1 \input texinfo @c -*-texinfo-*-
3 @c Free Software Foundation, Inc.
6 @c makeinfo ignores cmds prev to setfilename, so its arg cannot make use
7 @c of @set vars. However, you can override filename with makeinfo -o.
12 @settitle Debugging with @value{GDBN}
13 @setchapternewpage odd
24 @c readline appendices use @vindex
27 @c !!set GDB manual's edition---not the same as GDB version!
30 @c !!set GDB manual's revision date
31 @set DATE February 1999
33 @c THIS MANUAL REQUIRES TEXINFO-2 macros and info-makers to format properly.
36 @c This is a dir.info fragment to support semi-automated addition of
37 @c manuals to an info tree. zoo@cygnus.com is developing this facility.
40 * Gdb: (gdb). The @sc{gnu} debugger.
47 This file documents the @sc{gnu} debugger @value{GDBN}.
50 This is the @value{EDITION} Edition, @value{DATE},
51 of @cite{Debugging with @value{GDBN}: the @sc{gnu} Source-Level Debugger}
52 for @value{GDBN} Version @value{GDBVN}.
54 Copyright (C) 1988-1999 Free Software Foundation, Inc.
56 Permission is granted to make and distribute verbatim copies of
57 this manual provided the copyright notice and this permission notice
58 are preserved on all copies.
61 Permission is granted to process this file through TeX and print the
62 results, provided the printed document carries copying permission
63 notice identical to this one except for the removal of this paragraph
64 (this paragraph not being relevant to the printed manual).
67 Permission is granted to copy and distribute modified versions of this
68 manual under the conditions for verbatim copying, provided also that the
69 entire resulting derived work is distributed under the terms of a
70 permission notice identical to this one.
72 Permission is granted to copy and distribute translations of this manual
73 into another language, under the above conditions for modified versions.
77 @title Debugging with @value{GDBN}
78 @subtitle The @sc{gnu} Source-Level Debugger
80 @subtitle @value{EDITION} Edition, for @value{GDBN} version @value{GDBVN}
81 @subtitle @value{DATE}
82 @author Richard M. Stallman and Roland H. Pesch
86 \hfill (Send bugs and comments on @value{GDBN} to bug-gdb\@gnu.org.)\par
87 \hfill {\it Debugging with @value{GDBN}}\par
88 \hfill \TeX{}info \texinfoversion\par
92 @c ISBN seems to be wrong...
94 @vskip 0pt plus 1filll
95 Copyright @copyright{} 1988-1999 Free Software Foundation, Inc.
97 Published by the Free Software Foundation @*
98 59 Temple Place - Suite 330, @*
99 Boston, MA 02111-1307 USA @*
100 Printed copies are available for $20 each. @*
101 ISBN 1-882114-11-6 @*
103 Permission is granted to make and distribute verbatim copies of
104 this manual provided the copyright notice and this permission notice
105 are preserved on all copies.
107 Permission is granted to copy and distribute modified versions of this
108 manual under the conditions for verbatim copying, provided also that the
109 entire resulting derived work is distributed under the terms of a
110 permission notice identical to this one.
112 Permission is granted to copy and distribute translations of this manual
113 into another language, under the above conditions for modified versions.
119 @top Debugging with @value{GDBN}
121 This file describes @value{GDBN}, the @sc{gnu} symbolic debugger.
123 This is the @value{EDITION} Edition, @value{DATE}, for @value{GDBN} Version
126 Copyright (C) 1988-1999 Free Software Foundation, Inc.
128 * Summary:: Summary of @value{GDBN}
129 * Sample Session:: A sample @value{GDBN} session
131 * Invocation:: Getting in and out of @value{GDBN}
132 * Commands:: @value{GDBN} commands
133 * Running:: Running programs under @value{GDBN}
134 * Stopping:: Stopping and continuing
135 * Stack:: Examining the stack
136 * Source:: Examining source files
137 * Data:: Examining data
139 * Languages:: Using @value{GDBN} with different languages
141 * Symbols:: Examining the symbol table
142 * Altering:: Altering execution
143 * GDB Files:: @value{GDBN} files
144 * Targets:: Specifying a debugging target
145 * Configurations:: Configuration-specific information
146 * Controlling GDB:: Controlling @value{GDBN}
147 * Sequences:: Canned sequences of commands
148 * Emacs:: Using @value{GDBN} under @sc{gnu} Emacs
150 * GDB Bugs:: Reporting bugs in @value{GDBN}
151 * Formatting Documentation:: How to format and print @value{GDBN} documentation
153 * Command Line Editing:: Command Line Editing
154 * Using History Interactively:: Using History Interactively
155 * Installing GDB:: Installing GDB
162 @unnumbered Summary of @value{GDBN}
164 The purpose of a debugger such as @value{GDBN} is to allow you to see what is
165 going on ``inside'' another program while it executes---or what another
166 program was doing at the moment it crashed.
168 @value{GDBN} can do four main kinds of things (plus other things in support of
169 these) to help you catch bugs in the act:
173 Start your program, specifying anything that might affect its behavior.
176 Make your program stop on specified conditions.
179 Examine what has happened, when your program has stopped.
182 Change things in your program, so you can experiment with correcting the
183 effects of one bug and go on to learn about another.
186 You can use @value{GDBN} to debug programs written in C and C++.
187 For more information, see @ref{Support,,Supported languages}.
188 For more information, see @ref{C,,C and C++}.
192 Support for Modula-2 and Chill is partial. For information on Modula-2,
193 see @ref{Modula-2,,Modula-2}. For information on Chill, see @ref{Chill}.
196 Debugging Pascal programs which use sets, subranges, file variables, or
197 nested functions does not currently work. @value{GDBN} does not support
198 entering expressions, printing values, or similar features using Pascal
202 @value{GDBN} can be used to debug programs written in Fortran, although
203 it may be necessary to refer to some variables with a trailing
207 * Free Software:: Freely redistributable software
208 * Contributors:: Contributors to GDB
212 @unnumberedsec Free software
214 @value{GDBN} is @dfn{free software}, protected by the @sc{gnu}
215 General Public License
216 (GPL). The GPL gives you the freedom to copy or adapt a licensed
217 program---but every person getting a copy also gets with it the
218 freedom to modify that copy (which means that they must get access to
219 the source code), and the freedom to distribute further copies.
220 Typical software companies use copyrights to limit your freedoms; the
221 Free Software Foundation uses the GPL to preserve these freedoms.
223 Fundamentally, the General Public License is a license which says that
224 you have these freedoms and that you cannot take these freedoms away
228 @unnumberedsec Contributors to GDB
230 Richard Stallman was the original author of GDB, and of many other
231 @sc{gnu} programs. Many others have contributed to its development.
232 This section attempts to credit major contributors. One of the virtues
233 of free software is that everyone is free to contribute to it; with
234 regret, we cannot actually acknowledge everyone here. The file
235 @file{ChangeLog} in the @value{GDBN} distribution approximates a
236 blow-by-blow account.
238 Changes much prior to version 2.0 are lost in the mists of time.
241 @emph{Plea:} Additions to this section are particularly welcome. If you
242 or your friends (or enemies, to be evenhanded) have been unfairly
243 omitted from this list, we would like to add your names!
246 So that they may not regard their many labors as thankless, we
247 particularly thank those who shepherded @value{GDBN} through major
249 Jim Blandy (release 4.18);
250 Jason Molenda (release 4.17);
251 Stan Shebs (release 4.14);
252 Fred Fish (releases 4.16, 4.15, 4.13, 4.12, 4.11, 4.10, and 4.9);
253 Stu Grossman and John Gilmore (releases 4.8, 4.7, 4.6, 4.5, and 4.4);
254 John Gilmore (releases 4.3, 4.2, 4.1, 4.0, and 3.9);
255 Jim Kingdon (releases 3.5, 3.4, and 3.3);
256 and Randy Smith (releases 3.2, 3.1, and 3.0).
258 Richard Stallman, assisted at various times by Peter TerMaat, Chris
259 Hanson, and Richard Mlynarik, handled releases through 2.8.
261 Michael Tiemann is the author of most of the @sc{gnu} C++ support in GDB,
262 with significant additional contributions from Per Bothner. James
263 Clark wrote the @sc{gnu} C++ demangler. Early work on C++ was by Peter
264 TerMaat (who also did much general update work leading to release 3.0).
266 @value{GDBN} 4 uses the BFD subroutine library to examine multiple
267 object-file formats; BFD was a joint project of David V.
268 Henkel-Wallace, Rich Pixley, Steve Chamberlain, and John Gilmore.
270 David Johnson wrote the original COFF support; Pace Willison did
271 the original support for encapsulated COFF.
273 Brent Benson of Harris Computer Systems contributed DWARF 2 support.
275 Adam de Boor and Bradley Davis contributed the ISI Optimum V support.
276 Per Bothner, Noboyuki Hikichi, and Alessandro Forin contributed MIPS
278 Jean-Daniel Fekete contributed Sun 386i support.
279 Chris Hanson improved the HP9000 support.
280 Noboyuki Hikichi and Tomoyuki Hasei contributed Sony/News OS 3 support.
281 David Johnson contributed Encore Umax support.
282 Jyrki Kuoppala contributed Altos 3068 support.
283 Jeff Law contributed HP PA and SOM support.
284 Keith Packard contributed NS32K support.
285 Doug Rabson contributed Acorn Risc Machine support.
286 Bob Rusk contributed Harris Nighthawk CX-UX support.
287 Chris Smith contributed Convex support (and Fortran debugging).
288 Jonathan Stone contributed Pyramid support.
289 Michael Tiemann contributed SPARC support.
290 Tim Tucker contributed support for the Gould NP1 and Gould Powernode.
291 Pace Willison contributed Intel 386 support.
292 Jay Vosburgh contributed Symmetry support.
294 Andreas Schwab contributed M68K Linux support.
296 Rich Schaefer and Peter Schauer helped with support of SunOS shared
299 Jay Fenlason and Roland McGrath ensured that @value{GDBN} and GAS agree
300 about several machine instruction sets.
302 Patrick Duval, Ted Goldstein, Vikram Koka and Glenn Engel helped develop
303 remote debugging. Intel Corporation, Wind River Systems, AMD, and ARM
304 contributed remote debugging modules for the i960, VxWorks, A29K UDI,
305 and RDI targets, respectively.
307 Brian Fox is the author of the readline libraries providing
308 command-line editing and command history.
310 Andrew Beers of SUNY Buffalo wrote the language-switching code, the
311 Modula-2 support, and contributed the Languages chapter of this manual.
313 Fred Fish wrote most of the support for Unix System Vr4.
314 He also enhanced the command-completion support to cover C++ overloaded
317 Hitachi America, Ltd. sponsored the support for H8/300, H8/500, and
320 NEC sponsored the support for the v850, Vr4xxx, and Vr5xxx processors.
322 Mitsubishi sponsored the support for D10V, D30V, and M32R/D processors.
324 Toshiba sponsored the support for the TX39 Mips processor.
326 Matsushita sponsored the support for the MN10200 and MN10300 processors.
328 Fujitsu sponsored the support for SPARClite and FR30 processors
330 Kung Hsu, Jeff Law, and Rick Sladkey added support for hardware
333 Michael Snyder added support for tracepoints.
335 Stu Grossman wrote gdbserver.
337 Jim Kingdon, Peter Schauer, Ian Taylor, and Stu Grossman made
338 nearly innumerable bug fixes and cleanups throughout GDB.
340 The following people at the Hewlett-Packard Company contributed
341 support for the PA-RISC 2.0 architecture, HP-UX 10.20, 10.30, and 11.0
342 (narrow mode), HP's implementation of kernel threads, HP's aC++
343 compiler, and the terminal user interface: Ben Krepp, Richard Title,
344 John Bishop, Susan Macchia, Kathy Mann, Satish Pai, India Paul, Steve
345 Rehrauer, and Elena Zannoni. Kim Haase provided HP-specific
346 information in this manual.
348 Cygnus Solutions has sponsored GDB maintenance and much of its
349 development since 1991. Cygnus engineers who have worked on GDB
350 fulltime include Mark Alexander, Jim Blandy, Per Bothner, Edith Epstein,
351 Chris Faylor, Fred Fish, Martin Hunt, Jim Ingham, John Gilmore, Stu
352 Grossman, Kung Hsu, Jim Kingdon, John Metzler, Fernando Nasser, Geoffrey
353 Noer, Dawn Perchik, Rich Pixley, Zdenek Radouch, Keith Seitz, Stan
354 Shebs, David Taylor, and Elena Zannoni. In addition, Dave Brolley, Ian
355 Carmichael, Steve Chamberlain, Nick Clifton, JT Conklin, Stan Cox, DJ
356 Delorie, Ulrich Drepper, Frank Eigler, Doug Evans, Sean Fagan, David
357 Henkel-Wallace, Richard Henderson, Jeff Holcomb, Jeff Law, Jim Lemke,
358 Tom Lord, Bob Manson, Michael Meissner, Jason Merrill, Catherine Moore,
359 Drew Moseley, Ken Raeburn, Gavin Romig-Koch, Rob Savoye, Jamie Smith,
360 Mike Stump, Ian Taylor, Angela Thomas, Michael Tiemann, Tom Tromey, Ron
361 Unrau, Jim Wilson, and David Zuhn have made contributions both large
366 @chapter A Sample @value{GDBN} Session
368 You can use this manual at your leisure to read all about @value{GDBN}.
369 However, a handful of commands are enough to get started using the
370 debugger. This chapter illustrates those commands.
373 In this sample session, we emphasize user input like this: @b{input},
374 to make it easier to pick out from the surrounding output.
377 @c FIXME: this example may not be appropriate for some configs, where
378 @c FIXME...primary interest is in remote use.
380 One of the preliminary versions of @sc{gnu} @code{m4} (a generic macro
381 processor) exhibits the following bug: sometimes, when we change its
382 quote strings from the default, the commands used to capture one macro
383 definition within another stop working. In the following short @code{m4}
384 session, we define a macro @code{foo} which expands to @code{0000}; we
385 then use the @code{m4} built-in @code{defn} to define @code{bar} as the
386 same thing. However, when we change the open quote string to
387 @code{<QUOTE>} and the close quote string to @code{<UNQUOTE>}, the same
388 procedure fails to define a new synonym @code{baz}:
397 @b{define(bar,defn(`foo'))}
401 @b{changequote(<QUOTE>,<UNQUOTE>)}
403 @b{define(baz,defn(<QUOTE>foo<UNQUOTE>))}
406 m4: End of input: 0: fatal error: EOF in string
410 Let us use @value{GDBN} to try to see what is going on.
413 $ @b{@value{GDBP} m4}
414 @c FIXME: this falsifies the exact text played out, to permit smallbook
415 @c FIXME... format to come out better.
416 @value{GDBN} is free software and you are welcome to distribute copies
417 of it under certain conditions; type "show copying" to see
419 There is absolutely no warranty for @value{GDBN}; type "show warranty"
422 @value{GDBN} @value{GDBVN}, Copyright 1999 Free Software Foundation, Inc...
427 @value{GDBN} reads only enough symbol data to know where to find the
428 rest when needed; as a result, the first prompt comes up very quickly.
429 We now tell @value{GDBN} to use a narrower display width than usual, so
430 that examples fit in this manual.
433 (@value{GDBP}) @b{set width 70}
437 We need to see how the @code{m4} built-in @code{changequote} works.
438 Having looked at the source, we know the relevant subroutine is
439 @code{m4_changequote}, so we set a breakpoint there with the @value{GDBN}
440 @code{break} command.
443 (@value{GDBP}) @b{break m4_changequote}
444 Breakpoint 1 at 0x62f4: file builtin.c, line 879.
448 Using the @code{run} command, we start @code{m4} running under @value{GDBN}
449 control; as long as control does not reach the @code{m4_changequote}
450 subroutine, the program runs as usual:
453 (@value{GDBP}) @b{run}
454 Starting program: /work/Editorial/gdb/gnu/m4/m4
462 To trigger the breakpoint, we call @code{changequote}. @value{GDBN}
463 suspends execution of @code{m4}, displaying information about the
464 context where it stops.
467 @b{changequote(<QUOTE>,<UNQUOTE>)}
469 Breakpoint 1, m4_changequote (argc=3, argv=0x33c70)
471 879 if (bad_argc(TOKEN_DATA_TEXT(argv[0]),argc,1,3))
475 Now we use the command @code{n} (@code{next}) to advance execution to
476 the next line of the current function.
480 882 set_quotes((argc >= 2) ? TOKEN_DATA_TEXT(argv[1])\
485 @code{set_quotes} looks like a promising subroutine. We can go into it
486 by using the command @code{s} (@code{step}) instead of @code{next}.
487 @code{step} goes to the next line to be executed in @emph{any}
488 subroutine, so it steps into @code{set_quotes}.
492 set_quotes (lq=0x34c78 "<QUOTE>", rq=0x34c88 "<UNQUOTE>")
494 530 if (lquote != def_lquote)
498 The display that shows the subroutine where @code{m4} is now
499 suspended (and its arguments) is called a stack frame display. It
500 shows a summary of the stack. We can use the @code{backtrace}
501 command (which can also be spelled @code{bt}), to see where we are
502 in the stack as a whole: the @code{backtrace} command displays a
503 stack frame for each active subroutine.
506 (@value{GDBP}) @b{bt}
507 #0 set_quotes (lq=0x34c78 "<QUOTE>", rq=0x34c88 "<UNQUOTE>")
509 #1 0x6344 in m4_changequote (argc=3, argv=0x33c70)
511 #2 0x8174 in expand_macro (sym=0x33320) at macro.c:242
512 #3 0x7a88 in expand_token (obs=0x0, t=209696, td=0xf7fffa30)
514 #4 0x79dc in expand_input () at macro.c:40
515 #5 0x2930 in main (argc=0, argv=0xf7fffb20) at m4.c:195
519 We step through a few more lines to see what happens. The first two
520 times, we can use @samp{s}; the next two times we use @code{n} to avoid
521 falling into the @code{xstrdup} subroutine.
525 0x3b5c 532 if (rquote != def_rquote)
527 0x3b80 535 lquote = (lq == nil || *lq == '\0') ? \
528 def_lquote : xstrdup(lq);
530 536 rquote = (rq == nil || *rq == '\0') ? def_rquote\
533 538 len_lquote = strlen(rquote);
537 The last line displayed looks a little odd; we can examine the variables
538 @code{lquote} and @code{rquote} to see if they are in fact the new left
539 and right quotes we specified. We use the command @code{p}
540 (@code{print}) to see their values.
543 (@value{GDBP}) @b{p lquote}
544 $1 = 0x35d40 "<QUOTE>"
545 (@value{GDBP}) @b{p rquote}
546 $2 = 0x35d50 "<UNQUOTE>"
550 @code{lquote} and @code{rquote} are indeed the new left and right quotes.
551 To look at some context, we can display ten lines of source
552 surrounding the current line with the @code{l} (@code{list}) command.
558 535 lquote = (lq == nil || *lq == '\0') ? def_lquote\
560 536 rquote = (rq == nil || *rq == '\0') ? def_rquote\
563 538 len_lquote = strlen(rquote);
564 539 len_rquote = strlen(lquote);
571 Let us step past the two lines that set @code{len_lquote} and
572 @code{len_rquote}, and then examine the values of those variables.
576 539 len_rquote = strlen(lquote);
579 (@value{GDBP}) @b{p len_lquote}
581 (@value{GDBP}) @b{p len_rquote}
586 That certainly looks wrong, assuming @code{len_lquote} and
587 @code{len_rquote} are meant to be the lengths of @code{lquote} and
588 @code{rquote} respectively. We can set them to better values using
589 the @code{p} command, since it can print the value of
590 any expression---and that expression can include subroutine calls and
594 (@value{GDBP}) @b{p len_lquote=strlen(lquote)}
596 (@value{GDBP}) @b{p len_rquote=strlen(rquote)}
601 Is that enough to fix the problem of using the new quotes with the
602 @code{m4} built-in @code{defn}? We can allow @code{m4} to continue
603 executing with the @code{c} (@code{continue}) command, and then try the
604 example that caused trouble initially:
610 @b{define(baz,defn(<QUOTE>foo<UNQUOTE>))}
617 Success! The new quotes now work just as well as the default ones. The
618 problem seems to have been just the two typos defining the wrong
619 lengths. We allow @code{m4} exit by giving it an EOF as input:
623 Program exited normally.
627 The message @samp{Program exited normally.} is from @value{GDBN}; it
628 indicates @code{m4} has finished executing. We can end our @value{GDBN}
629 session with the @value{GDBN} @code{quit} command.
632 (@value{GDBP}) @b{quit}
636 @chapter Getting In and Out of @value{GDBN}
638 This chapter discusses how to start @value{GDBN}, and how to get out of it.
642 type @samp{@value{GDBP}} to start @value{GDBN}.
644 type @kbd{quit} or @kbd{C-d} to exit.
648 * Invoking GDB:: How to start @value{GDBN}
649 * Quitting GDB:: How to quit @value{GDBN}
650 * Shell Commands:: How to use shell commands inside @value{GDBN}
654 @section Invoking @value{GDBN}
656 Invoke @value{GDBN} by running the program @code{@value{GDBP}}. Once started,
657 @value{GDBN} reads commands from the terminal until you tell it to exit.
659 You can also run @code{@value{GDBP}} with a variety of arguments and options,
660 to specify more of your debugging environment at the outset.
662 The command-line options described here are designed
663 to cover a variety of situations; in some environments, some of these
664 options may effectively be unavailable.
666 The most usual way to start @value{GDBN} is with one argument,
667 specifying an executable program:
670 @value{GDBP} @var{program}
674 You can also start with both an executable program and a core file
678 @value{GDBP} @var{program} @var{core}
681 You can, instead, specify a process ID as a second argument, if you want
682 to debug a running process:
685 @value{GDBP} @var{program} 1234
689 would attach @value{GDBN} to process @code{1234} (unless you also have a file
690 named @file{1234}; @value{GDBN} does check for a core file first).
692 Taking advantage of the second command-line argument requires a fairly
693 complete operating system; when you use @value{GDBN} as a remote debugger
694 attached to a bare board, there may not be any notion of ``process'',
695 and there is often no way to get a core dump.
697 You can run @code{gdb} without printing the front material, which describes
698 @value{GDBN}'s non-warranty, by specifying @code{-silent}:
705 You can further control how @value{GDBN} starts up by using command-line
706 options. @value{GDBN} itself can remind you of the options available.
716 to display all available options and briefly describe their use
717 (@samp{@value{GDBP} -h} is a shorter equivalent).
719 All options and command line arguments you give are processed
720 in sequential order. The order makes a difference when the
721 @samp{-x} option is used.
725 * File Options:: Choosing files
726 * Mode Options:: Choosing modes
730 @subsection Choosing files
732 When @value{GDBN} starts
733 specifying an executable file and core file (or process ID). This is
734 the same as if the arguments were specified by the @samp{-se} and
735 @samp{-c} options respectively. (@value{GDBN} reads the first argument
736 that does not have an associated option flag as equivalent to the
737 @samp{-se} option followed by that argument; and the second argument
738 that does not have an associated option flag, if any, as equivalent to
739 the @samp{-c} option followed by that argument.)
741 If @value{GDBN} has not been configured to included core file support,
742 such as for most embedded targets, then it will complain about a second
743 argument and ignore it.
745 Many options have both long and short forms; both are shown in the
746 following list. @value{GDBN} also recognizes the long forms if you truncate
747 them, so long as enough of the option is present to be unambiguous.
748 (If you prefer, you can flag option arguments with @samp{--} rather
749 than @samp{-}, though we illustrate the more usual convention.)
752 @item -symbols @var{file}
754 Read symbol table from file @var{file}.
756 @item -exec @var{file}
758 Use file @var{file} as the executable file to execute when appropriate,
759 and for examining pure data in conjunction with a core dump.
762 Read symbol table from file @var{file} and use it as the executable
765 @item -core @var{file}
767 Use file @var{file} as a core dump to examine.
769 @item -c @var{number}
770 Connect to process ID @var{number}, as with the @code{attach} command
771 (unless there is a file in core-dump format named @var{number}, in which
772 case @samp{-c} specifies that file as a core dump to read).
774 @item -command @var{file}
776 Execute @value{GDBN} commands from file @var{file}. @xref{Command
777 Files,, Command files}.
779 @item -directory @var{directory}
780 @itemx -d @var{directory}
781 Add @var{directory} to the path to search for source files.
785 @emph{Warning: this option depends on operating system facilities that are not
786 supported on all systems.}@*
787 If memory-mapped files are available on your system through the @code{mmap}
788 system call, you can use this option
789 to have @value{GDBN} write the symbols from your
790 program into a reusable file in the current directory. If the program you are debugging is
791 called @file{/tmp/fred}, the mapped symbol file is @file{./fred.syms}.
792 Future @value{GDBN} debugging sessions notice the presence of this file,
793 and can quickly map in symbol information from it, rather than reading
794 the symbol table from the executable program.
796 The @file{.syms} file is specific to the host machine where @value{GDBN}
797 is run. It holds an exact image of the internal @value{GDBN} symbol
798 table. It cannot be shared across multiple host platforms.
802 Read each symbol file's entire symbol table immediately, rather than
803 the default, which is to read it incrementally as it is needed.
804 This makes startup slower, but makes future operations faster.
808 The @code{-mapped} and @code{-readnow} options are typically combined in
809 order to build a @file{.syms} file that contains complete symbol
810 information. (@xref{Files,,Commands to specify files}, for
811 information on @file{.syms} files.) A simple @value{GDBN} invocation to do
812 nothing but build a @file{.syms} file for future use is:
815 gdb -batch -nx -mapped -readnow programname
819 @subsection Choosing modes
821 You can run @value{GDBN} in various alternative modes---for example, in
822 batch mode or quiet mode.
827 Do not execute commands from any initialization files (normally called
828 @file{.gdbinit}, or @file{gdb.ini} on PCs). Normally, the commands in
829 these files are executed after all the command options and arguments
830 have been processed. @xref{Command Files,,Command files}.
834 ``Quiet''. Do not print the introductory and copyright messages. These
835 messages are also suppressed in batch mode.
838 Run in batch mode. Exit with status @code{0} after processing all the
839 command files specified with @samp{-x} (and all commands from
840 initialization files, if not inhibited with @samp{-n}). Exit with
841 nonzero status if an error occurs in executing the @value{GDBN} commands
842 in the command files.
844 Batch mode may be useful for running @value{GDBN} as a filter, for example to
845 download and run a program on another computer; in order to make this
846 more useful, the message
849 Program exited normally.
853 (which is ordinarily issued whenever a program running under @value{GDBN} control
854 terminates) is not issued when running in batch mode.
856 @item -cd @var{directory}
857 Run @value{GDBN} using @var{directory} as its working directory,
858 instead of the current directory.
862 @sc{gnu} Emacs sets this option when it runs @value{GDBN} as a
863 subprocess. It tells @value{GDBN} to output the full file name and line
864 number in a standard, recognizable fashion each time a stack frame is
865 displayed (which includes each time your program stops). This
866 recognizable format looks like two @samp{\032} characters, followed by
867 the file name, line number and character position separated by colons,
868 and a newline. The Emacs-to-@value{GDBN} interface program uses the two
869 @samp{\032} characters as a signal to display the source code for the
873 Set the line speed (baud rate or bits per second) of any serial
874 interface used by @value{GDBN} for remote debugging.
876 @item -tty @var{device}
877 Run using @var{device} for your program's standard input and output.
878 @c FIXME: kingdon thinks there is more to -tty. Investigate.
880 @c resolve the situation of these eventually
882 @c Use a Terminal User Interface. For information, use your Web browser to
883 @c read the file @file{TUI.html}, which is usually installed in the
884 @c directory @code{/opt/langtools/wdb/doc} on HP-UX systems. Do not use
885 @c this option if you run @value{GDBN} from Emacs (see @pxref{Emacs, ,Using
886 @c @value{GDBN} under @sc{gnu} Emacs}).
889 @c Run in XDB compatibility mode, allowing the use of certain XDB commands.
890 @c For information, see the file @file{xdb_trans.html}, which is usually
891 @c installed in the directory @code{/opt/langtools/wdb/doc} on HP-UX
897 @section Quitting @value{GDBN}
898 @cindex exiting @value{GDBN}
899 @cindex leaving @value{GDBN}
902 @kindex quit @r{[}@var{expression}@r{]}
905 To exit @value{GDBN}, use the @code{quit} command (abbreviated @code{q}), or
906 type an end-of-file character (usually @kbd{C-d}). If you do not supply
907 @var{expression}, @value{GDBN} will terminate normally; otherwise it will
908 terminate using the result of @var{expression} as the error code.
912 An interrupt (often @kbd{C-c}) does not exit from @value{GDBN}, but rather
913 terminates the action of any @value{GDBN} command that is in progress and
914 returns to @value{GDBN} command level. It is safe to type the interrupt
915 character at any time because @value{GDBN} does not allow it to take effect
916 until a time when it is safe.
918 If you have been using @value{GDBN} to control an attached process or
919 device, you can release it with the @code{detach} command
920 (@pxref{Attach, ,Debugging an already-running process}).
923 @section Shell commands
925 If you need to execute occasional shell commands during your
926 debugging session, there is no need to leave or suspend @value{GDBN}; you can
927 just use the @code{shell} command.
932 @item shell @var{command string}
933 Invoke a standard shell to execute @var{command string}.
934 If it exists, the environment variable @code{SHELL} determines which
935 shell to run. Otherwise @value{GDBN} uses @code{/bin/sh}.
938 The utility @code{make} is often needed in development environments.
939 You do not have to use the @code{shell} command for this purpose in
945 @item make @var{make-args}
946 Execute the @code{make} program with the specified
947 arguments. This is equivalent to @samp{shell make @var{make-args}}.
951 @chapter @value{GDBN} Commands
953 You can abbreviate a @value{GDBN} command to the first few letters of the command
954 name, if that abbreviation is unambiguous; and you can repeat certain
955 @value{GDBN} commands by typing just @key{RET}. You can also use the @key{TAB}
956 key to get @value{GDBN} to fill out the rest of a word in a command (or to
957 show you the alternatives available, if there is more than one possibility).
960 * Command Syntax:: How to give commands to @value{GDBN}
961 * Completion:: Command completion
962 * Help:: How to ask @value{GDBN} for help
966 @section Command syntax
968 A @value{GDBN} command is a single line of input. There is no limit on
969 how long it can be. It starts with a command name, which is followed by
970 arguments whose meaning depends on the command name. For example, the
971 command @code{step} accepts an argument which is the number of times to
972 step, as in @samp{step 5}. You can also use the @code{step} command
973 with no arguments. Some command names do not allow any arguments.
976 @value{GDBN} command names may always be truncated if that abbreviation is
977 unambiguous. Other possible command abbreviations are listed in the
978 documentation for individual commands. In some cases, even ambiguous
979 abbreviations are allowed; for example, @code{s} is specially defined as
980 equivalent to @code{step} even though there are other commands whose
981 names start with @code{s}. You can test abbreviations by using them as
982 arguments to the @code{help} command.
984 @cindex repeating commands
986 A blank line as input to @value{GDBN} (typing just @key{RET}) means to
987 repeat the previous command. Certain commands (for example, @code{run})
988 will not repeat this way; these are commands whose unintentional
989 repetition might cause trouble and which you are unlikely to want to
992 The @code{list} and @code{x} commands, when you repeat them with
993 @key{RET}, construct new arguments rather than repeating
994 exactly as typed. This permits easy scanning of source or memory.
996 @value{GDBN} can also use @key{RET} in another way: to partition lengthy
997 output, in a way similar to the common utility @code{more}
998 (@pxref{Screen Size,,Screen size}). Since it is easy to press one
999 @key{RET} too many in this situation, @value{GDBN} disables command
1000 repetition after any command that generates this sort of display.
1004 Any text from a @kbd{#} to the end of the line is a comment; it does
1005 nothing. This is useful mainly in command files (@pxref{Command
1006 Files,,Command files}).
1009 @section Command completion
1012 @cindex word completion
1013 @value{GDBN} can fill in the rest of a word in a command for you, if there is
1014 only one possibility; it can also show you what the valid possibilities
1015 are for the next word in a command, at any time. This works for @value{GDBN}
1016 commands, @value{GDBN} subcommands, and the names of symbols in your program.
1018 Press the @key{TAB} key whenever you want @value{GDBN} to fill out the rest
1019 of a word. If there is only one possibility, @value{GDBN} fills in the
1020 word, and waits for you to finish the command (or press @key{RET} to
1021 enter it). For example, if you type
1023 @c FIXME "@key" does not distinguish its argument sufficiently to permit
1024 @c complete accuracy in these examples; space introduced for clarity.
1025 @c If texinfo enhancements make it unnecessary, it would be nice to
1026 @c replace " @key" by "@key" in the following...
1028 (@value{GDBP}) info bre @key{TAB}
1032 @value{GDBN} fills in the rest of the word @samp{breakpoints}, since that is
1033 the only @code{info} subcommand beginning with @samp{bre}:
1036 (@value{GDBP}) info breakpoints
1040 You can either press @key{RET} at this point, to run the @code{info
1041 breakpoints} command, or backspace and enter something else, if
1042 @samp{breakpoints} does not look like the command you expected. (If you
1043 were sure you wanted @code{info breakpoints} in the first place, you
1044 might as well just type @key{RET} immediately after @samp{info bre},
1045 to exploit command abbreviations rather than command completion).
1047 If there is more than one possibility for the next word when you press
1048 @key{TAB}, @value{GDBN} sounds a bell. You can either supply more
1049 characters and try again, or just press @key{TAB} a second time;
1050 @value{GDBN} displays all the possible completions for that word. For
1051 example, you might want to set a breakpoint on a subroutine whose name
1052 begins with @samp{make_}, but when you type @kbd{b make_@key{TAB}} @value{GDBN}
1053 just sounds the bell. Typing @key{TAB} again displays all the
1054 function names in your program that begin with those characters, for
1058 (@value{GDBP}) b make_ @key{TAB}
1059 @exdent @value{GDBN} sounds bell; press @key{TAB} again, to see:
1060 make_a_section_from_file make_environ
1061 make_abs_section make_function_type
1062 make_blockvector make_pointer_type
1063 make_cleanup make_reference_type
1064 make_command make_symbol_completion_list
1065 (@value{GDBP}) b make_
1069 After displaying the available possibilities, @value{GDBN} copies your
1070 partial input (@samp{b make_} in the example) so you can finish the
1073 If you just want to see the list of alternatives in the first place, you
1074 can press @kbd{M-?} rather than pressing @key{TAB} twice. @kbd{M-?}
1075 means @kbd{@key{META} ?}. You can type this either by holding down a
1076 key designated as the @key{META} shift on your keyboard (if there is
1077 one) while typing @kbd{?}, or as @key{ESC} followed by @kbd{?}.
1079 @cindex quotes in commands
1080 @cindex completion of quoted strings
1081 Sometimes the string you need, while logically a ``word'', may contain
1082 parentheses or other characters that @value{GDBN} normally excludes from
1083 its notion of a word. To permit word completion to work in this
1084 situation, you may enclose words in @code{'} (single quote marks) in
1085 @value{GDBN} commands.
1087 The most likely situation where you might need this is in typing the
1088 name of a C++ function. This is because C++ allows function overloading
1089 (multiple definitions of the same function, distinguished by argument
1090 type). For example, when you want to set a breakpoint you may need to
1091 distinguish whether you mean the version of @code{name} that takes an
1092 @code{int} parameter, @code{name(int)}, or the version that takes a
1093 @code{float} parameter, @code{name(float)}. To use the word-completion
1094 facilities in this situation, type a single quote @code{'} at the
1095 beginning of the function name. This alerts @value{GDBN} that it may need to
1096 consider more information than usual when you press @key{TAB} or
1097 @kbd{M-?} to request word completion:
1100 (@value{GDBP}) b 'bubble( @key{M-?}
1101 bubble(double,double) bubble(int,int)
1102 (@value{GDBP}) b 'bubble(
1105 In some cases, @value{GDBN} can tell that completing a name requires using
1106 quotes. When this happens, @value{GDBN} inserts the quote for you (while
1107 completing as much as it can) if you do not type the quote in the first
1111 (@value{GDBP}) b bub @key{TAB}
1112 @exdent @value{GDBN} alters your input line to the following, and rings a bell:
1113 (@value{GDBP}) b 'bubble(
1117 In general, @value{GDBN} can tell that a quote is needed (and inserts it) if
1118 you have not yet started typing the argument list when you ask for
1119 completion on an overloaded symbol.
1121 For more information about overloaded functions, @pxref{C plus plus
1122 expressions, ,C++ expressions}. You can use the command @code{set
1123 overload-resolution off} to disable overload resolution;
1124 @pxref{Debugging C plus plus, ,@value{GDBN} features for C++}.
1128 @section Getting help
1129 @cindex online documentation
1132 You can always ask @value{GDBN} itself for information on its commands,
1133 using the command @code{help}.
1139 You can use @code{help} (abbreviated @code{h}) with no arguments to
1140 display a short list of named classes of commands:
1144 List of classes of commands:
1146 running -- Running the program
1147 stack -- Examining the stack
1148 data -- Examining data
1149 breakpoints -- Making program stop at certain points
1150 files -- Specifying and examining files
1151 status -- Status inquiries
1152 support -- Support facilities
1153 user-defined -- User-defined commands
1154 aliases -- Aliases of other commands
1155 obscure -- Obscure features
1157 Type "help" followed by a class name for a list of
1158 commands in that class.
1159 Type "help" followed by command name for full
1161 Command name abbreviations are allowed if unambiguous.
1165 @item help @var{class}
1166 Using one of the general help classes as an argument, you can get a
1167 list of the individual commands in that class. For example, here is the
1168 help display for the class @code{status}:
1171 (@value{GDBP}) help status
1176 @c Line break in "show" line falsifies real output, but needed
1177 @c to fit in smallbook page size.
1178 show -- Generic command for showing things set
1180 info -- Generic command for printing status
1182 Type "help" followed by command name for full
1184 Command name abbreviations are allowed if unambiguous.
1188 @item help @var{command}
1189 With a command name as @code{help} argument, @value{GDBN} displays a
1190 short paragraph on how to use that command.
1193 @item complete @var{args}
1194 The @code{complete @var{args}} command lists all the possible completions
1195 for the beginning of a command. Use @var{args} to specify the beginning of the
1196 command you want completed. For example:
1202 @noindent results in:
1212 @noindent This is intended for use by @sc{gnu} Emacs.
1215 In addition to @code{help}, you can use the @value{GDBN} commands @code{info}
1216 and @code{show} to inquire about the state of your program, or the state
1217 of @value{GDBN} itself. Each command supports many topics of inquiry; this
1218 manual introduces each of them in the appropriate context. The listings
1219 under @code{info} and under @code{show} in the Index point to
1220 all the sub-commands. @xref{Index}.
1227 This command (abbreviated @code{i}) is for describing the state of your
1228 program. For example, you can list the arguments given to your program
1229 with @code{info args}, list the registers currently in use with @code{info
1230 registers}, or list the breakpoints you have set with @code{info breakpoints}.
1231 You can get a complete list of the @code{info} sub-commands with
1232 @w{@code{help info}}.
1236 You can assign the result of an expression to an environment variable with
1237 @code{set}. For example, you can set the @value{GDBN} prompt to a $-sign with
1238 @code{set prompt $}.
1242 In contrast to @code{info}, @code{show} is for describing the state of
1243 @value{GDBN} itself.
1244 You can change most of the things you can @code{show}, by using the
1245 related command @code{set}; for example, you can control what number
1246 system is used for displays with @code{set radix}, or simply inquire
1247 which is currently in use with @code{show radix}.
1250 To display all the settable parameters and their current
1251 values, you can use @code{show} with no arguments; you may also use
1252 @code{info set}. Both commands produce the same display.
1253 @c FIXME: "info set" violates the rule that "info" is for state of
1254 @c FIXME...program. Ck w/ GNU: "info set" to be called something else,
1255 @c FIXME...or change desc of rule---eg "state of prog and debugging session"?
1259 Here are three miscellaneous @code{show} subcommands, all of which are
1260 exceptional in lacking corresponding @code{set} commands:
1263 @kindex show version
1264 @cindex version number
1266 Show what version of @value{GDBN} is running. You should include this
1267 information in @value{GDBN} bug-reports. If multiple versions of @value{GDBN} are in
1268 use at your site, you may occasionally want to determine which version
1269 of @value{GDBN} you are running; as @value{GDBN} evolves, new commands are introduced,
1270 and old ones may wither away. The version number is also announced
1271 when you start @value{GDBN}.
1273 @kindex show copying
1275 Display information about permission for copying @value{GDBN}.
1277 @kindex show warranty
1279 Display the @sc{gnu} ``NO WARRANTY'' statement.
1283 @chapter Running Programs Under @value{GDBN}
1285 When you run a program under @value{GDBN}, you must first generate
1286 debugging information when you compile it.
1288 You may start @value{GDBN} with its arguments, if any, in an environment
1289 of your choice. If you are doing native debugging, you may redirect
1290 your program's input and output, debug an already running process, or
1291 kill a child process.
1294 * Compilation:: Compiling for debugging
1295 * Starting:: Starting your program
1296 * Arguments:: Your program's arguments
1297 * Environment:: Your program's environment
1299 * Working Directory:: Your program's working directory
1300 * Input/Output:: Your program's input and output
1301 * Attach:: Debugging an already-running process
1302 * Kill Process:: Killing the child process
1304 * Threads:: Debugging programs with multiple threads
1305 * Processes:: Debugging programs with multiple processes
1309 @section Compiling for debugging
1311 In order to debug a program effectively, you need to generate
1312 debugging information when you compile it. This debugging information
1313 is stored in the object file; it describes the data type of each
1314 variable or function and the correspondence between source line numbers
1315 and addresses in the executable code.
1317 To request debugging information, specify the @samp{-g} option when you run
1320 Many C compilers are unable to handle the @samp{-g} and @samp{-O}
1321 options together. Using those compilers, you cannot generate optimized
1322 executables containing debugging information.
1324 @value{NGCC}, the @sc{gnu} C compiler, supports @samp{-g} with or
1325 without @samp{-O}, making it possible to debug optimized code. We
1326 recommend that you @emph{always} use @samp{-g} whenever you compile a
1327 program. You may think your program is correct, but there is no sense
1328 in pushing your luck.
1330 @cindex optimized code, debugging
1331 @cindex debugging optimized code
1332 When you debug a program compiled with @samp{-g -O}, remember that the
1333 optimizer is rearranging your code; the debugger shows you what is
1334 really there. Do not be too surprised when the execution path does not
1335 exactly match your source file! An extreme example: if you define a
1336 variable, but never use it, @value{GDBN} never sees that
1337 variable---because the compiler optimizes it out of existence.
1339 Some things do not work as well with @samp{-g -O} as with just
1340 @samp{-g}, particularly on machines with instruction scheduling. If in
1341 doubt, recompile with @samp{-g} alone, and if this fixes the problem,
1342 please report it to us as a bug (including a test case!).
1344 Older versions of the @sc{gnu} C compiler permitted a variant option
1345 @w{@samp{-gg}} for debugging information. @value{GDBN} no longer supports this
1346 format; if your @sc{gnu} C compiler has this option, do not use it.
1350 @section Starting your program
1358 Use the @code{run} command to start your program under @value{GDBN}.
1359 You must first specify the program name (except on VxWorks) with an
1360 argument to @value{GDBN} (@pxref{Invocation, ,Getting In and Out of
1361 @value{GDBN}}), or by using the @code{file} or @code{exec-file} command
1362 (@pxref{Files, ,Commands to specify files}).
1366 If you are running your program in an execution environment that
1367 supports processes, @code{run} creates an inferior process and makes
1368 that process run your program. (In environments without processes,
1369 @code{run} jumps to the start of your program.)
1371 The execution of a program is affected by certain information it
1372 receives from its superior. @value{GDBN} provides ways to specify this
1373 information, which you must do @emph{before} starting your program. (You
1374 can change it after starting your program, but such changes only affect
1375 your program the next time you start it.) This information may be
1376 divided into four categories:
1379 @item The @emph{arguments.}
1380 Specify the arguments to give your program as the arguments of the
1381 @code{run} command. If a shell is available on your target, the shell
1382 is used to pass the arguments, so that you may use normal conventions
1383 (such as wildcard expansion or variable substitution) in describing
1385 In Unix systems, you can control which shell is used with the
1386 @code{SHELL} environment variable.
1387 @xref{Arguments, ,Your program's arguments}.
1389 @item The @emph{environment.}
1390 Your program normally inherits its environment from @value{GDBN}, but you can
1391 use the @value{GDBN} commands @code{set environment} and @code{unset
1392 environment} to change parts of the environment that affect
1393 your program. @xref{Environment, ,Your program's environment}.
1395 @item The @emph{working directory.}
1396 Your program inherits its working directory from @value{GDBN}. You can set
1397 the @value{GDBN} working directory with the @code{cd} command in @value{GDBN}.
1398 @xref{Working Directory, ,Your program's working directory}.
1400 @item The @emph{standard input and output.}
1401 Your program normally uses the same device for standard input and
1402 standard output as @value{GDBN} is using. You can redirect input and output
1403 in the @code{run} command line, or you can use the @code{tty} command to
1404 set a different device for your program.
1405 @xref{Input/Output, ,Your program's input and output}.
1408 @emph{Warning:} While input and output redirection work, you cannot use
1409 pipes to pass the output of the program you are debugging to another
1410 program; if you attempt this, @value{GDBN} is likely to wind up debugging the
1414 When you issue the @code{run} command, your program begins to execute
1415 immediately. @xref{Stopping, ,Stopping and continuing}, for discussion
1416 of how to arrange for your program to stop. Once your program has
1417 stopped, you may call functions in your program, using the @code{print}
1418 or @code{call} commands. @xref{Data, ,Examining Data}.
1420 If the modification time of your symbol file has changed since the last
1421 time @value{GDBN} read its symbols, @value{GDBN} discards its symbol
1422 table, and reads it again. When it does this, @value{GDBN} tries to retain
1423 your current breakpoints.
1426 @section Your program's arguments
1428 @cindex arguments (to your program)
1429 The arguments to your program can be specified by the arguments of the
1431 They are passed to a shell, which expands wildcard characters and
1432 performs redirection of I/O, and thence to your program. Your
1433 @code{SHELL} environment variable (if it exists) specifies what shell
1434 @value{GDBN} uses. If you do not define @code{SHELL}, @value{GDBN} uses
1437 @code{run} with no arguments uses the same arguments used by the previous
1438 @code{run}, or those set by the @code{set args} command.
1443 Specify the arguments to be used the next time your program is run. If
1444 @code{set args} has no arguments, @code{run} executes your program
1445 with no arguments. Once you have run your program with arguments,
1446 using @code{set args} before the next @code{run} is the only way to run
1447 it again without arguments.
1451 Show the arguments to give your program when it is started.
1455 @section Your program's environment
1457 @cindex environment (of your program)
1458 The @dfn{environment} consists of a set of environment variables and
1459 their values. Environment variables conventionally record such things as
1460 your user name, your home directory, your terminal type, and your search
1461 path for programs to run. Usually you set up environment variables with
1462 the shell and they are inherited by all the other programs you run. When
1463 debugging, it can be useful to try running your program with a modified
1464 environment without having to start @value{GDBN} over again.
1468 @item path @var{directory}
1469 Add @var{directory} to the front of the @code{PATH} environment variable
1470 (the search path for executables), for both @value{GDBN} and your program.
1471 You may specify several directory names, separated by @samp{:} or
1472 whitespace. If @var{directory} is already in the path, it is moved to
1473 the front, so it is searched sooner.
1475 You can use the string @samp{$cwd} to refer to whatever is the current
1476 working directory at the time @value{GDBN} searches the path. If you
1477 use @samp{.} instead, it refers to the directory where you executed the
1478 @code{path} command. @value{GDBN} replaces @samp{.} in the
1479 @var{directory} argument (with the current path) before adding
1480 @var{directory} to the search path.
1481 @c 'path' is explicitly nonrepeatable, but RMS points out it is silly to
1482 @c document that, since repeating it would be a no-op.
1486 Display the list of search paths for executables (the @code{PATH}
1487 environment variable).
1489 @kindex show environment
1490 @item show environment @r{[}@var{varname}@r{]}
1491 Print the value of environment variable @var{varname} to be given to
1492 your program when it starts. If you do not supply @var{varname},
1493 print the names and values of all environment variables to be given to
1494 your program. You can abbreviate @code{environment} as @code{env}.
1496 @kindex set environment
1497 @item set environment @var{varname} @r{[}=@var{value}@r{]}
1498 Set environment variable @var{varname} to @var{value}. The value
1499 changes for your program only, not for @value{GDBN} itself. @var{value} may
1500 be any string; the values of environment variables are just strings, and
1501 any interpretation is supplied by your program itself. The @var{value}
1502 parameter is optional; if it is eliminated, the variable is set to a
1504 @c "any string" here does not include leading, trailing
1505 @c blanks. Gnu asks: does anyone care?
1507 For example, this command:
1514 tells a Unix program, when subsequently run, that its user is named
1515 @samp{foo}. (The spaces around @samp{=} are used for clarity here; they
1516 are not actually required.)
1518 @kindex unset environment
1519 @item unset environment @var{varname}
1520 Remove variable @var{varname} from the environment to be passed to your
1521 program. This is different from @samp{set env @var{varname} =};
1522 @code{unset environment} removes the variable from the environment,
1523 rather than assigning it an empty value.
1526 @emph{Warning:} @value{GDBN} runs your program using the shell indicated
1527 by your @code{SHELL} environment variable if it exists (or
1528 @code{/bin/sh} if not). If your @code{SHELL} variable names a shell
1529 that runs an initialization file---such as @file{.cshrc} for C-shell, or
1530 @file{.bashrc} for BASH---any variables you set in that file affect
1531 your program. You may wish to move setting of environment variables to
1532 files that are only run when you sign on, such as @file{.login} or
1535 @node Working Directory
1536 @section Your program's working directory
1538 @cindex working directory (of your program)
1539 Each time you start your program with @code{run}, it inherits its
1540 working directory from the current working directory of @value{GDBN}.
1541 The @value{GDBN} working directory is initially whatever it inherited
1542 from its parent process (typically the shell), but you can specify a new
1543 working directory in @value{GDBN} with the @code{cd} command.
1545 The @value{GDBN} working directory also serves as a default for the commands
1546 that specify files for @value{GDBN} to operate on. @xref{Files, ,Commands to
1551 @item cd @var{directory}
1552 Set the @value{GDBN} working directory to @var{directory}.
1556 Print the @value{GDBN} working directory.
1560 @section Your program's input and output
1565 By default, the program you run under @value{GDBN} does input and output to
1566 the same terminal that @value{GDBN} uses. @value{GDBN} switches the terminal
1567 to its own terminal modes to interact with you, but it records the terminal
1568 modes your program was using and switches back to them when you continue
1569 running your program.
1572 @kindex info terminal
1574 Displays information recorded by @value{GDBN} about the terminal modes your
1578 You can redirect your program's input and/or output using shell
1579 redirection with the @code{run} command. For example,
1586 starts your program, diverting its output to the file @file{outfile}.
1589 @cindex controlling terminal
1590 Another way to specify where your program should do input and output is
1591 with the @code{tty} command. This command accepts a file name as
1592 argument, and causes this file to be the default for future @code{run}
1593 commands. It also resets the controlling terminal for the child
1594 process, for future @code{run} commands. For example,
1601 directs that processes started with subsequent @code{run} commands
1602 default to do input and output on the terminal @file{/dev/ttyb} and have
1603 that as their controlling terminal.
1605 An explicit redirection in @code{run} overrides the @code{tty} command's
1606 effect on the input/output device, but not its effect on the controlling
1609 When you use the @code{tty} command or redirect input in the @code{run}
1610 command, only the input @emph{for your program} is affected. The input
1611 for @value{GDBN} still comes from your terminal.
1614 @section Debugging an already-running process
1619 @item attach @var{process-id}
1620 This command attaches to a running process---one that was started
1621 outside @value{GDBN}. (@code{info files} shows your active
1622 targets.) The command takes as argument a process ID. The usual way to
1623 find out the process-id of a Unix process is with the @code{ps} utility,
1624 or with the @samp{jobs -l} shell command.
1626 @code{attach} does not repeat if you press @key{RET} a second time after
1627 executing the command.
1630 To use @code{attach}, your program must be running in an environment
1631 which supports processes; for example, @code{attach} does not work for
1632 programs on bare-board targets that lack an operating system. You must
1633 also have permission to send the process a signal.
1635 When you use @code{attach}, the debugger finds the program running in
1636 the process first by looking in the current working directory, then (if
1637 the program is not found) by using the source file search path
1638 (@pxref{Source Path, ,Specifying source directories}). You can also use
1639 the @code{file} command to load the program. @xref{Files, ,Commands to
1642 The first thing @value{GDBN} does after arranging to debug the specified
1643 process is to stop it. You can examine and modify an attached process
1644 with all the @value{GDBN} commands that are ordinarily available when
1645 you start processes with @code{run}. You can insert breakpoints; you
1646 can step and continue; you can modify storage. If you would rather the
1647 process continue running, you may use the @code{continue} command after
1648 attaching @value{GDBN} to the process.
1653 When you have finished debugging the attached process, you can use the
1654 @code{detach} command to release it from @value{GDBN} control. Detaching
1655 the process continues its execution. After the @code{detach} command,
1656 that process and @value{GDBN} become completely independent once more, and you
1657 are ready to @code{attach} another process or start one with @code{run}.
1658 @code{detach} does not repeat if you press @key{RET} again after
1659 executing the command.
1662 If you exit @value{GDBN} or use the @code{run} command while you have an
1663 attached process, you kill that process. By default, @value{GDBN} asks
1664 for confirmation if you try to do either of these things; you can
1665 control whether or not you need to confirm by using the @code{set
1666 confirm} command (@pxref{Messages/Warnings, ,Optional warnings and
1670 @section Killing the child process
1675 Kill the child process in which your program is running under @value{GDBN}.
1678 This command is useful if you wish to debug a core dump instead of a
1679 running process. @value{GDBN} ignores any core dump file while your program
1682 On some operating systems, a program cannot be executed outside @value{GDBN}
1683 while you have breakpoints set on it inside @value{GDBN}. You can use the
1684 @code{kill} command in this situation to permit running your program
1685 outside the debugger.
1687 The @code{kill} command is also useful if you wish to recompile and
1688 relink your program, since on many systems it is impossible to modify an
1689 executable file while it is running in a process. In this case, when you
1690 next type @code{run}, @value{GDBN} notices that the file has changed, and
1691 reads the symbol table again (while trying to preserve your current
1692 breakpoint settings).
1695 @section Debugging programs with multiple threads
1697 @cindex threads of execution
1698 @cindex multiple threads
1699 @cindex switching threads
1700 In some operating systems, such as HP-UX and Solaris, a single program
1701 may have more than one @dfn{thread} of execution. The precise semantics
1702 of threads differ from one operating system to another, but in general
1703 the threads of a single program are akin to multiple processes---except
1704 that they share one address space (that is, they can all examine and
1705 modify the same variables). On the other hand, each thread has its own
1706 registers and execution stack, and perhaps private memory.
1708 @value{GDBN} provides these facilities for debugging multi-thread
1712 @item automatic notification of new threads
1713 @item @samp{thread @var{threadno}}, a command to switch among threads
1714 @item @samp{info threads}, a command to inquire about existing threads
1715 @item @samp{thread apply [@var{threadno}] [@var{all}] @var{args}},
1716 a command to apply a command to a list of threads
1717 @item thread-specific breakpoints
1721 @emph{Warning:} These facilities are not yet available on every
1722 @value{GDBN} configuration where the operating system supports threads.
1723 If your @value{GDBN} does not support threads, these commands have no
1724 effect. For example, a system without thread support shows no output
1725 from @samp{info threads}, and always rejects the @code{thread} command,
1729 (@value{GDBP}) info threads
1730 (@value{GDBP}) thread 1
1731 Thread ID 1 not known. Use the "info threads" command to
1732 see the IDs of currently known threads.
1734 @c FIXME to implementors: how hard would it be to say "sorry, this GDB
1735 @c doesn't support threads"?
1738 @cindex focus of debugging
1739 @cindex current thread
1740 The @value{GDBN} thread debugging facility allows you to observe all
1741 threads while your program runs---but whenever @value{GDBN} takes
1742 control, one thread in particular is always the focus of debugging.
1743 This thread is called the @dfn{current thread}. Debugging commands show
1744 program information from the perspective of the current thread.
1746 @kindex New @var{systag}
1747 @cindex thread identifier (system)
1748 @c FIXME-implementors!! It would be more helpful if the [New...] message
1749 @c included GDB's numeric thread handle, so you could just go to that
1750 @c thread without first checking `info threads'.
1751 Whenever @value{GDBN} detects a new thread in your program, it displays
1752 the target system's identification for the thread with a message in the
1753 form @samp{[New @var{systag}]}. @var{systag} is a thread identifier
1754 whose form varies depending on the particular system. For example, on
1755 LynxOS, you might see
1758 [New process 35 thread 27]
1762 when @value{GDBN} notices a new thread. In contrast, on an SGI system,
1763 the @var{systag} is simply something like @samp{process 368}, with no
1766 @c FIXME!! (1) Does the [New...] message appear even for the very first
1767 @c thread of a program, or does it only appear for the
1768 @c second---i.e., when it becomes obvious we have a multithread
1770 @c (2) *Is* there necessarily a first thread always? Or do some
1771 @c multithread systems permit starting a program with multiple
1772 @c threads ab initio?
1774 @cindex thread number
1775 @cindex thread identifier (GDB)
1776 For debugging purposes, @value{GDBN} associates its own thread
1777 number---always a single integer---with each thread in your program.
1780 @kindex info threads
1782 Display a summary of all threads currently in your
1783 program. @value{GDBN} displays for each thread (in this order):
1786 @item the thread number assigned by @value{GDBN}
1788 @item the target system's thread identifier (@var{systag})
1790 @item the current stack frame summary for that thread
1794 An asterisk @samp{*} to the left of the @value{GDBN} thread number
1795 indicates the current thread.
1799 @c end table here to get a little more width for example
1802 (@value{GDBP}) info threads
1803 3 process 35 thread 27 0x34e5 in sigpause ()
1804 2 process 35 thread 23 0x34e5 in sigpause ()
1805 * 1 process 35 thread 13 main (argc=1, argv=0x7ffffff8)
1811 @cindex thread number
1812 @cindex thread identifier (GDB)
1813 For debugging purposes, @value{GDBN} associates its own thread
1814 number---a small integer assigned in thread-creation order---with each
1815 thread in your program.
1817 @kindex New @var{systag}
1818 @cindex thread identifier (system)
1819 @c FIXME-implementors!! It would be more helpful if the [New...] message
1820 @c included GDB's numeric thread handle, so you could just go to that
1821 @c thread without first checking `info threads'.
1822 Whenever @value{GDBN} detects a new thread in your program, it displays
1823 both @value{GDBN}'s thread number and the target system's identification for the thread with a message in the
1824 form @samp{[New @var{systag}]}. @var{systag} is a thread identifier
1825 whose form varies depending on the particular system. For example, on
1829 [New thread 2 (system thread 26594)]
1833 when @value{GDBN} notices a new thread.
1836 @kindex info threads
1838 Display a summary of all threads currently in your
1839 program. @value{GDBN} displays for each thread (in this order):
1842 @item the thread number assigned by @value{GDBN}
1844 @item the target system's thread identifier (@var{systag})
1846 @item the current stack frame summary for that thread
1850 An asterisk @samp{*} to the left of the @value{GDBN} thread number
1851 indicates the current thread.
1855 @c end table here to get a little more width for example
1858 (@value{GDBP}) info threads
1859 * 3 system thread 26607 worker (wptr=0x7b09c318 "@@") at quicksort.c:137
1860 2 system thread 26606 0x7b0030d8 in __ksleep () from /usr/lib/libc.2
1861 1 system thread 27905 0x7b003498 in _brk () from /usr/lib/libc.2
1865 @kindex thread @var{threadno}
1866 @item thread @var{threadno}
1867 Make thread number @var{threadno} the current thread. The command
1868 argument @var{threadno} is the internal @value{GDBN} thread number, as
1869 shown in the first field of the @samp{info threads} display.
1870 @value{GDBN} responds by displaying the system identifier of the thread
1871 you selected, and its current stack frame summary:
1874 @c FIXME!! This example made up; find a @value{GDBN} w/threads and get real one
1875 (@value{GDBP}) thread 2
1876 [Switching to process 35 thread 23]
1877 0x34e5 in sigpause ()
1881 As with the @samp{[New @dots{}]} message, the form of the text after
1882 @samp{Switching to} depends on your system's conventions for identifying
1885 @kindex thread apply
1886 @item thread apply [@var{threadno}] [@var{all}] @var{args}
1887 The @code{thread apply} command allows you to apply a command to one or
1888 more threads. Specify the numbers of the threads that you want affected
1889 with the command argument @var{threadno}. @var{threadno} is the internal
1890 @value{GDBN} thread number, as shown in the first field of the @samp{info
1891 threads} display. To apply a command to all threads, use
1892 @code{thread apply all} @var{args}.
1895 @cindex automatic thread selection
1896 @cindex switching threads automatically
1897 @cindex threads, automatic switching
1898 Whenever @value{GDBN} stops your program, due to a breakpoint or a
1899 signal, it automatically selects the thread where that breakpoint or
1900 signal happened. @value{GDBN} alerts you to the context switch with a
1901 message of the form @samp{[Switching to @var{systag}]} to identify the
1904 @xref{Thread Stops,,Stopping and starting multi-thread programs}, for
1905 more information about how @value{GDBN} behaves when you stop and start
1906 programs with multiple threads.
1908 @xref{Set Watchpoints,,Setting watchpoints}, for information about
1909 watchpoints in programs with multiple threads.
1912 @section Debugging programs with multiple processes
1914 @cindex fork, debugging programs which call
1915 @cindex multiple processes
1916 @cindex processes, multiple
1917 On most systems, @value{GDBN} has no special support for debugging
1918 programs which create additional processes using the @code{fork}
1919 function. When a program forks, @value{GDBN} will continue to debug the
1920 parent process and the child process will run unimpeded. If you have
1921 set a breakpoint in any code which the child then executes, the child
1922 will get a @code{SIGTRAP} signal which (unless it catches the signal)
1923 will cause it to terminate.
1925 However, if you want to debug the child process there is a workaround
1926 which isn't too painful. Put a call to @code{sleep} in the code which
1927 the child process executes after the fork. It may be useful to sleep
1928 only if a certain environment variable is set, or a certain file exists,
1929 so that the delay need not occur when you don't want to run @value{GDBN}
1930 on the child. While the child is sleeping, use the @code{ps} program to
1931 get its process ID. Then tell @value{GDBN} (a new invocation of
1932 @value{GDBN} if you are also debugging the parent process) to attach to
1933 the child process (see @ref{Attach}). From that point on you can debug
1934 the child process just like any other process which you attached to.
1936 On HP-UX (11.x and later only?), @value{GDBN} provides support for
1937 debugging programs that create additional processes using the
1938 @code{fork} or @code{vfork} function.
1940 By default, when a program forks, @value{GDBN} will continue to debug
1941 the parent process and the child process will run unimpeded.
1943 If you want to follow the child process instead of the parent process,
1944 use the command @w{@code{set follow-fork-mode}}.
1947 @kindex set follow-fork-mode
1948 @item set follow-fork-mode @var{mode}
1949 Set the debugger response to a program call of @code{fork} or
1950 @code{vfork}. A call to @code{fork} or @code{vfork} creates a new
1951 process. The @var{mode} can be:
1955 The original process is debugged after a fork. The child process runs
1959 The new process is debugged after a fork. The parent process runs
1963 The debugger will ask for one of the above choices.
1966 @item show follow-fork-mode
1967 Display the current debugger response to a fork or vfork call.
1970 If you ask to debug a child process and a @code{vfork} is followed by an
1971 @code{exec}, @value{GDBN} executes the new target up to the first
1972 breakpoint in the new target. If you have a breakpoint set on
1973 @code{main} in your original program, the breakpoint will also be set on
1974 the child process's @code{main}.
1976 When a child process is spawned by @code{vfork}, you cannot debug the
1977 child or parent until an @code{exec} call completes.
1979 If you issue a @code{run} command to @value{GDBN} after an @code{exec}
1980 call executes, the new target restarts. To restart the parent process,
1981 use the @code{file} command with the parent executable name as its
1984 You can use the @code{catch} command to make @value{GDBN} stop whenever
1985 a @code{fork}, @code{vfork}, or @code{exec} call is made. @xref{Set
1986 Catchpoints, ,Setting catchpoints}.
1989 @chapter Stopping and Continuing
1991 The principal purposes of using a debugger are so that you can stop your
1992 program before it terminates; or so that, if your program runs into
1993 trouble, you can investigate and find out why.
1995 Inside @value{GDBN}, your program may stop for any of several reasons,
1996 such as a signal, a breakpoint, or reaching a new line after a
1997 @value{GDBN} command such as @code{step}. You may then examine and
1998 change variables, set new breakpoints or remove old ones, and then
1999 continue execution. Usually, the messages shown by @value{GDBN} provide
2000 ample explanation of the status of your program---but you can also
2001 explicitly request this information at any time.
2004 @kindex info program
2006 Display information about the status of your program: whether it is
2007 running or not, what process it is, and why it stopped.
2011 * Breakpoints:: Breakpoints, watchpoints, and catchpoints
2012 * Continuing and Stepping:: Resuming execution
2014 * Thread Stops:: Stopping and starting multi-thread programs
2018 @section Breakpoints, watchpoints, and catchpoints
2021 A @dfn{breakpoint} makes your program stop whenever a certain point in
2022 the program is reached. For each breakpoint, you can add conditions to
2023 control in finer detail whether your program stops. You can set
2024 breakpoints with the @code{break} command and its variants (@pxref{Set
2025 Breaks, ,Setting breakpoints}), to specify the place where your program
2026 should stop by line number, function name or exact address in the
2029 In HP-UX, SunOS 4.x, SVR4, and Alpha OSF/1 configurations, you can set
2030 breakpoints in shared libraries before the executable is run. There is
2031 a minor limitation on HP-UX systems: you must wait until the executable
2032 is run in order to set breakpoints in shared library routines that are
2033 not called directly by the program (for example, routines that are
2034 arguments in a @code{pthread_create} call).
2037 @cindex memory tracing
2038 @cindex breakpoint on memory address
2039 @cindex breakpoint on variable modification
2040 A @dfn{watchpoint} is a special breakpoint that stops your program
2041 when the value of an expression changes. You must use a different
2042 command to set watchpoints (@pxref{Set Watchpoints, ,Setting
2043 watchpoints}), but aside from that, you can manage a watchpoint like
2044 any other breakpoint: you enable, disable, and delete both breakpoints
2045 and watchpoints using the same commands.
2047 You can arrange to have values from your program displayed automatically
2048 whenever @value{GDBN} stops at a breakpoint. @xref{Auto Display,,
2052 @cindex breakpoint on events
2053 A @dfn{catchpoint} is another special breakpoint that stops your program
2054 when a certain kind of event occurs, such as the throwing of a C++
2055 exception or the loading of a library. As with watchpoints, you use a
2056 different command to set a catchpoint (@pxref{Set Catchpoints, ,Setting
2057 catchpoints}), but aside from that, you can manage a catchpoint like any
2058 other breakpoint. (To stop when your program receives a signal, use the
2059 @code{handle} command; @pxref{Signals, ,Signals}.)
2061 @cindex breakpoint numbers
2062 @cindex numbers for breakpoints
2063 @value{GDBN} assigns a number to each breakpoint, watchpoint, or
2064 catchpoint when you create it; these numbers are successive integers
2065 starting with one. In many of the commands for controlling various
2066 features of breakpoints you use the breakpoint number to say which
2067 breakpoint you want to change. Each breakpoint may be @dfn{enabled} or
2068 @dfn{disabled}; if disabled, it has no effect on your program until you
2072 * Set Breaks:: Setting breakpoints
2073 * Set Watchpoints:: Setting watchpoints
2074 * Set Catchpoints:: Setting catchpoints
2075 * Delete Breaks:: Deleting breakpoints
2076 * Disabling:: Disabling breakpoints
2077 * Conditions:: Break conditions
2078 * Break Commands:: Breakpoint command lists
2079 * Breakpoint Menus:: Breakpoint menus
2081 @c * Error in Breakpoints:: ``Cannot insert breakpoints''
2085 @subsection Setting breakpoints
2087 @c FIXME LMB what does GDB do if no code on line of breakpt?
2088 @c consider in particular declaration with/without initialization.
2090 @c FIXME 2 is there stuff on this already? break at fun start, already init?
2095 @cindex latest breakpoint
2096 Breakpoints are set with the @code{break} command (abbreviated
2097 @code{b}). The debugger convenience variable @samp{$bpnum} records the
2098 number of the breakpoints you've set most recently; see @ref{Convenience
2099 Vars,, Convenience variables}, for a discussion of what you can do with
2100 convenience variables.
2102 You have several ways to say where the breakpoint should go.
2105 @item break @var{function}
2106 Set a breakpoint at entry to function @var{function}.
2107 When using source languages that permit overloading of symbols, such as
2108 C++, @var{function} may refer to more than one possible place to break.
2109 @xref{Breakpoint Menus,,Breakpoint menus}, for a discussion of that situation.
2111 @item break +@var{offset}
2112 @itemx break -@var{offset}
2113 Set a breakpoint some number of lines forward or back from the position
2114 at which execution stopped in the currently selected frame.
2116 @item break @var{linenum}
2117 Set a breakpoint at line @var{linenum} in the current source file.
2118 That file is the last file whose source text was printed. This
2119 breakpoint stops your program just before it executes any of the
2122 @item break @var{filename}:@var{linenum}
2123 Set a breakpoint at line @var{linenum} in source file @var{filename}.
2125 @item break @var{filename}:@var{function}
2126 Set a breakpoint at entry to function @var{function} found in file
2127 @var{filename}. Specifying a file name as well as a function name is
2128 superfluous except when multiple files contain similarly named
2131 @item break *@var{address}
2132 Set a breakpoint at address @var{address}. You can use this to set
2133 breakpoints in parts of your program which do not have debugging
2134 information or source files.
2137 When called without any arguments, @code{break} sets a breakpoint at
2138 the next instruction to be executed in the selected stack frame
2139 (@pxref{Stack, ,Examining the Stack}). In any selected frame but the
2140 innermost, this makes your program stop as soon as control
2141 returns to that frame. This is similar to the effect of a
2142 @code{finish} command in the frame inside the selected frame---except
2143 that @code{finish} does not leave an active breakpoint. If you use
2144 @code{break} without an argument in the innermost frame, @value{GDBN} stops
2145 the next time it reaches the current location; this may be useful
2148 @value{GDBN} normally ignores breakpoints when it resumes execution, until at
2149 least one instruction has been executed. If it did not do this, you
2150 would be unable to proceed past a breakpoint without first disabling the
2151 breakpoint. This rule applies whether or not the breakpoint already
2152 existed when your program stopped.
2154 @item break @dots{} if @var{cond}
2155 Set a breakpoint with condition @var{cond}; evaluate the expression
2156 @var{cond} each time the breakpoint is reached, and stop only if the
2157 value is nonzero---that is, if @var{cond} evaluates as true.
2158 @samp{@dots{}} stands for one of the possible arguments described
2159 above (or no argument) specifying where to break. @xref{Conditions,
2160 ,Break conditions}, for more information on breakpoint conditions.
2163 @item tbreak @var{args}
2164 Set a breakpoint enabled only for one stop. @var{args} are the
2165 same as for the @code{break} command, and the breakpoint is set in the same
2166 way, but the breakpoint is automatically deleted after the first time your
2167 program stops there. @xref{Disabling, ,Disabling breakpoints}.
2170 @item hbreak @var{args}
2171 Set a hardware-assisted breakpoint. @var{args} are the same as for the
2172 @code{break} command and the breakpoint is set in the same way, but the
2173 breakpoint requires hardware support and some target hardware may not
2174 have this support. The main purpose of this is EPROM/ROM code
2175 debugging, so you can set a breakpoint at an instruction without
2176 changing the instruction. This can be used with the new trap-generation
2177 provided by SPARClite DSU. DSU will generate traps when a program accesses
2178 some data or instruction address that is assigned to the debug registers.
2179 However the hardware breakpoint registers can only take two data breakpoints,
2180 and @value{GDBN} will reject this command if more than two are used.
2181 Delete or disable unused hardware breakpoints before setting
2182 new ones. @xref{Conditions, ,Break conditions}.
2185 @item thbreak @var{args}
2186 Set a hardware-assisted breakpoint enabled only for one stop. @var{args}
2187 are the same as for the @code{hbreak} command and the breakpoint is set in
2188 the same way. However, like the @code{tbreak} command,
2189 the breakpoint is automatically deleted after the
2190 first time your program stops there. Also, like the @code{hbreak}
2191 command, the breakpoint requires hardware support and some target hardware
2192 may not have this support. @xref{Disabling, ,Disabling breakpoints}.
2193 Also @xref{Conditions, ,Break conditions}.
2196 @cindex regular expression
2197 @item rbreak @var{regex}
2198 @c FIXME what kind of regexp?
2199 Set breakpoints on all functions matching the regular expression
2200 @var{regex}. This command
2201 sets an unconditional breakpoint on all matches, printing a list of all
2202 breakpoints it set. Once these breakpoints are set, they are treated
2203 just like the breakpoints set with the @code{break} command. You can
2204 delete them, disable them, or make them conditional the same way as any
2207 When debugging C++ programs, @code{rbreak} is useful for setting
2208 breakpoints on overloaded functions that are not members of any special
2211 @kindex info breakpoints
2212 @cindex @code{$_} and @code{info breakpoints}
2213 @item info breakpoints @r{[}@var{n}@r{]}
2214 @itemx info break @r{[}@var{n}@r{]}
2215 @itemx info watchpoints @r{[}@var{n}@r{]}
2216 Print a table of all breakpoints, watchpoints, and catchpoints set and
2217 not deleted, with the following columns for each breakpoint:
2220 @item Breakpoint Numbers
2222 Breakpoint, watchpoint, or catchpoint.
2224 Whether the breakpoint is marked to be disabled or deleted when hit.
2225 @item Enabled or Disabled
2226 Enabled breakpoints are marked with @samp{y}. @samp{n} marks breakpoints
2227 that are not enabled.
2229 Where the breakpoint is in your program, as a memory address
2231 Where the breakpoint is in the source for your program, as a file and
2236 If a breakpoint is conditional, @code{info break} shows the condition on
2237 the line following the affected breakpoint; breakpoint commands, if any,
2238 are listed after that.
2241 @code{info break} with a breakpoint
2242 number @var{n} as argument lists only that breakpoint. The
2243 convenience variable @code{$_} and the default examining-address for
2244 the @code{x} command are set to the address of the last breakpoint
2245 listed (@pxref{Memory, ,Examining memory}).
2248 @code{info break} displays a count of the number of times the breakpoint
2249 has been hit. This is especially useful in conjunction with the
2250 @code{ignore} command. You can ignore a large number of breakpoint
2251 hits, look at the breakpoint info to see how many times the breakpoint
2252 was hit, and then run again, ignoring one less than that number. This
2253 will get you quickly to the last hit of that breakpoint.
2256 @value{GDBN} allows you to set any number of breakpoints at the same place in
2257 your program. There is nothing silly or meaningless about this. When
2258 the breakpoints are conditional, this is even useful
2259 (@pxref{Conditions, ,Break conditions}).
2261 @cindex negative breakpoint numbers
2262 @cindex internal @value{GDBN} breakpoints
2263 @value{GDBN} itself sometimes sets breakpoints in your program for special
2264 purposes, such as proper handling of @code{longjmp} (in C programs).
2265 These internal breakpoints are assigned negative numbers, starting with
2266 @code{-1}; @samp{info breakpoints} does not display them.
2268 You can see these breakpoints with the @value{GDBN} maintenance command
2269 @samp{maint info breakpoints}.
2272 @kindex maint info breakpoints
2273 @item maint info breakpoints
2274 Using the same format as @samp{info breakpoints}, display both the
2275 breakpoints you've set explicitly, and those @value{GDBN} is using for
2276 internal purposes. Internal breakpoints are shown with negative
2277 breakpoint numbers. The type column identifies what kind of breakpoint
2282 Normal, explicitly set breakpoint.
2285 Normal, explicitly set watchpoint.
2288 Internal breakpoint, used to handle correctly stepping through
2289 @code{longjmp} calls.
2291 @item longjmp resume
2292 Internal breakpoint at the target of a @code{longjmp}.
2295 Temporary internal breakpoint used by the @value{GDBN} @code{until} command.
2298 Temporary internal breakpoint used by the @value{GDBN} @code{finish} command.
2301 Shared library events.
2308 @node Set Watchpoints
2309 @subsection Setting watchpoints
2311 @cindex setting watchpoints
2312 @cindex software watchpoints
2313 @cindex hardware watchpoints
2314 You can use a watchpoint to stop execution whenever the value of an
2315 expression changes, without having to predict a particular place where
2318 Depending on your system, watchpoints may be implemented in software or
2319 hardware. GDB does software watchpointing by single-stepping your
2320 program and testing the variable's value each time, which is hundreds of
2321 times slower than normal execution. (But this may still be worth it, to
2322 catch errors where you have no clue what part of your program is the
2325 On some systems, such as HP-UX and Linux, GDB includes support for
2326 hardware watchpoints, which do not slow down the running of your
2331 @item watch @var{expr}
2332 Set a watchpoint for an expression. @value{GDBN} will break when @var{expr}
2333 is written into by the program and its value changes.
2336 @item rwatch @var{expr}
2337 Set a watchpoint that will break when watch @var{expr} is read by the program.
2340 @item awatch @var{expr}
2341 Set a watchpoint that will break when @var{args} is read and written into
2344 @kindex info watchpoints
2345 @item info watchpoints
2346 This command prints a list of watchpoints, breakpoints, and catchpoints;
2347 it is the same as @code{info break}.
2350 @value{GDBN} sets a @dfn{hardware watchpoint} if possible. Hardware
2351 watchpoints execute very quickly, and the debugger reports a change in
2352 value at the exact instruction where the change occurs. If @value{GDBN}
2353 cannot set a hardware watchpoint, it sets a software watchpoint, which
2354 executes more slowly and reports the change in value at the next
2355 statement, not the instruction, after the change occurs.
2357 When you issue the @code{watch} command, @value{GDBN} reports
2360 Hardware watchpoint @var{num}: @var{expr}
2364 if it was able to set a hardware watchpoint.
2366 Currently, the @code{awatch} and @code{rwatch} commands can only set
2367 hardware watchpoints, because accesses to data that don't change the
2368 value of the watched expression cannot be detected without examining
2369 every instruction as it is being executed, and @value{GDBN} does not do
2370 that currently. If @value{GDBN} finds that it is unable to set a
2371 hardware breakpoint with the @code{awatch} or @code{rwatch} command, it
2372 will print a message like this:
2375 Expression cannot be implemented with read/access watchpoint.
2378 Sometimes, @value{GDBN} cannot set a hardware watchpoint because the
2379 data type of the watched expression is wider than what a hardware
2380 watchpoint on the target machine can handle. For example, some systems
2381 can only watch regions that are up to 4 bytes wide; on such systems you
2382 cannot set hardware watchpoints for an expression that yields a
2383 double-precision floating-point number (which is typically 8 bytes
2384 wide). As a work-around, it might be possible to break the large region
2385 into a series of smaller ones and watch them with separate watchpoints.
2387 If you set too many hardware watchpoints, @value{GDBN} might be unable
2388 to insert all of them when you resume the execution of your program.
2389 Since the precise number of active watchpoints is unknown until such
2390 time as the program is about to be resumed, @value{GDBN} might not be
2391 able to warn you about this when you set the watchpoints, and the
2392 warning will be printed only when the program is resumed:
2395 Hardware watchpoint @var{num}: Could not insert watchpoint
2399 If this happens, delete or disable some of the watchpoints.
2401 The SPARClite DSU will generate traps when a program accesses some data
2402 or instruction address that is assigned to the debug registers. For the
2403 data addresses, DSU facilitates the @code{watch} command. However the
2404 hardware breakpoint registers can only take two data watchpoints, and
2405 both watchpoints must be the same kind. For example, you can set two
2406 watchpoints with @code{watch} commands, two with @code{rwatch} commands,
2407 @strong{or} two with @code{awatch} commands, but you cannot set one
2408 watchpoint with one command and the other with a different command.
2409 @value{GDBN} will reject the command if you try to mix watchpoints.
2410 Delete or disable unused watchpoint commands before setting new ones.
2412 If you call a function interactively using @code{print} or @code{call},
2413 any watchpoints you have set will be inactive until GDB reaches another
2414 kind of breakpoint or the call completes.
2416 @value{GDBN} automatically deletes watchpoints that watch local
2417 (automatic) variables, or expressions that involve such variables, when
2418 they go out of scope, that is, when the execution leaves the block in
2419 which these variables were defined. In particular, when the program
2420 being debugged terminates, @emph{all} local variables go out of scope,
2421 and so only watchpoints that watch global variables remain set. If you
2422 rerun the program, you will need to set all such watchpoints again. One
2423 way of doing that would be to set a code breakpoint at the entry to the
2424 @code{main} function and when it breaks, set all the watchpoints.
2427 @cindex watchpoints and threads
2428 @cindex threads and watchpoints
2429 @emph{Warning:} In multi-thread programs, watchpoints have only limited
2430 usefulness. With the current watchpoint implementation, @value{GDBN}
2431 can only watch the value of an expression @emph{in a single thread}. If
2432 you are confident that the expression can only change due to the current
2433 thread's activity (and if you are also confident that no other thread
2434 can become current), then you can use watchpoints as usual. However,
2435 @value{GDBN} may not notice when a non-current thread's activity changes
2438 @emph{HP-UX Warning:} In multi-thread programs, software watchpoints
2439 have only limited usefulness. If @value{GDBN} creates a software
2440 watchpoint, it can only watch the value of an expression @emph{in a
2441 single thread}. If you are confident that the expression can only
2442 change due to the current thread's activity (and if you are also
2443 confident that no other thread can become current), then you can use
2444 software watchpoints as usual. However, @value{GDBN} may not notice
2445 when a non-current thread's activity changes the expression. (Hardware
2446 watchpoints, in contrast, watch an expression in all threads.)
2449 @node Set Catchpoints
2450 @subsection Setting catchpoints
2452 @cindex exception handlers
2453 @cindex event handling
2455 You can use @dfn{catchpoints} to cause the debugger to stop for certain
2456 kinds of program events, such as C++ exceptions or the loading of a
2457 shared library. Use the @code{catch} command to set a catchpoint.
2461 @item catch @var{event}
2462 Stop when @var{event} occurs. @var{event} can be any of the following:
2466 The throwing of a C++ exception.
2470 The catching of a C++ exception.
2474 A call to @code{exec}. This is currently only available for HP-UX.
2478 A call to @code{fork}. This is currently only available for HP-UX.
2482 A call to @code{vfork}. This is currently only available for HP-UX.
2485 @itemx load @var{libname}
2487 The dynamic loading of any shared library, or the loading of the library
2488 @var{libname}. This is currently only available for HP-UX.
2491 @itemx unload @var{libname}
2492 @kindex catch unload
2493 The unloading of any dynamically loaded shared library, or the unloading
2494 of the library @var{libname}. This is currently only available for HP-UX.
2497 @item tcatch @var{event}
2498 Set a catchpoint that is enabled only for one stop. The catchpoint is
2499 automatically deleted after the first time the event is caught.
2503 Use the @code{info break} command to list the current catchpoints.
2505 There are currently some limitations to C++ exception handling
2506 (@code{catch throw} and @code{catch catch}) in @value{GDBN}:
2510 If you call a function interactively, @value{GDBN} normally returns
2511 control to you when the function has finished executing. If the call
2512 raises an exception, however, the call may bypass the mechanism that
2513 returns control to you and cause your program either to abort or to
2514 simply continue running until it hits a breakpoint, catches a signal
2515 that @value{GDBN} is listening for, or exits. This is the case even if
2516 you set a catchpoint for the exception; catchpoints on exceptions are
2517 disabled within interactive calls.
2520 You cannot raise an exception interactively.
2523 You cannot install an exception handler interactively.
2526 @cindex raise exceptions
2527 Sometimes @code{catch} is not the best way to debug exception handling:
2528 if you need to know exactly where an exception is raised, it is better to
2529 stop @emph{before} the exception handler is called, since that way you
2530 can see the stack before any unwinding takes place. If you set a
2531 breakpoint in an exception handler instead, it may not be easy to find
2532 out where the exception was raised.
2534 To stop just before an exception handler is called, you need some
2535 knowledge of the implementation. In the case of @sc{gnu} C++, exceptions are
2536 raised by calling a library function named @code{__raise_exception}
2537 which has the following ANSI C interface:
2540 /* @var{addr} is where the exception identifier is stored.
2541 ID is the exception identifier. */
2542 void __raise_exception (void **@var{addr}, void *@var{id});
2546 To make the debugger catch all exceptions before any stack
2547 unwinding takes place, set a breakpoint on @code{__raise_exception}
2548 (@pxref{Breakpoints, ,Breakpoints; watchpoints; and exceptions}).
2550 With a conditional breakpoint (@pxref{Conditions, ,Break conditions})
2551 that depends on the value of @var{id}, you can stop your program when
2552 a specific exception is raised. You can use multiple conditional
2553 breakpoints to stop your program when any of a number of exceptions are
2558 @subsection Deleting breakpoints
2560 @cindex clearing breakpoints, watchpoints, catchpoints
2561 @cindex deleting breakpoints, watchpoints, catchpoints
2562 It is often necessary to eliminate a breakpoint, watchpoint, or
2563 catchpoint once it has done its job and you no longer want your program
2564 to stop there. This is called @dfn{deleting} the breakpoint. A
2565 breakpoint that has been deleted no longer exists; it is forgotten.
2567 With the @code{clear} command you can delete breakpoints according to
2568 where they are in your program. With the @code{delete} command you can
2569 delete individual breakpoints, watchpoints, or catchpoints by specifying
2570 their breakpoint numbers.
2572 It is not necessary to delete a breakpoint to proceed past it. @value{GDBN}
2573 automatically ignores breakpoints on the first instruction to be executed
2574 when you continue execution without changing the execution address.
2579 Delete any breakpoints at the next instruction to be executed in the
2580 selected stack frame (@pxref{Selection, ,Selecting a frame}). When
2581 the innermost frame is selected, this is a good way to delete a
2582 breakpoint where your program just stopped.
2584 @item clear @var{function}
2585 @itemx clear @var{filename}:@var{function}
2586 Delete any breakpoints set at entry to the function @var{function}.
2588 @item clear @var{linenum}
2589 @itemx clear @var{filename}:@var{linenum}
2590 Delete any breakpoints set at or within the code of the specified line.
2592 @cindex delete breakpoints
2595 @item delete @r{[}breakpoints@r{]} @r{[}@var{bnums}@dots{}@r{]}
2596 Delete the breakpoints, watchpoints, or catchpoints of the numbers
2597 specified as arguments. If no argument is specified, delete all
2598 breakpoints (@value{GDBN} asks confirmation, unless you have @code{set
2599 confirm off}). You can abbreviate this command as @code{d}.
2603 @subsection Disabling breakpoints
2605 @kindex disable breakpoints
2606 @kindex enable breakpoints
2607 Rather than deleting a breakpoint, watchpoint, or catchpoint, you might
2608 prefer to @dfn{disable} it. This makes the breakpoint inoperative as if
2609 it had been deleted, but remembers the information on the breakpoint so
2610 that you can @dfn{enable} it again later.
2612 You disable and enable breakpoints, watchpoints, and catchpoints with
2613 the @code{enable} and @code{disable} commands, optionally specifying one
2614 or more breakpoint numbers as arguments. Use @code{info break} or
2615 @code{info watch} to print a list of breakpoints, watchpoints, and
2616 catchpoints if you do not know which numbers to use.
2618 A breakpoint, watchpoint, or catchpoint can have any of four different
2619 states of enablement:
2623 Enabled. The breakpoint stops your program. A breakpoint set
2624 with the @code{break} command starts out in this state.
2626 Disabled. The breakpoint has no effect on your program.
2628 Enabled once. The breakpoint stops your program, but then becomes
2629 disabled. A breakpoint set with the @code{tbreak} command starts out in
2632 Enabled for deletion. The breakpoint stops your program, but
2633 immediately after it does so it is deleted permanently.
2636 You can use the following commands to enable or disable breakpoints,
2637 watchpoints, and catchpoints:
2640 @kindex disable breakpoints
2643 @item disable @r{[}breakpoints@r{]} @r{[}@var{bnums}@dots{}@r{]}
2644 Disable the specified breakpoints---or all breakpoints, if none are
2645 listed. A disabled breakpoint has no effect but is not forgotten. All
2646 options such as ignore-counts, conditions and commands are remembered in
2647 case the breakpoint is enabled again later. You may abbreviate
2648 @code{disable} as @code{dis}.
2650 @kindex enable breakpoints
2652 @item enable @r{[}breakpoints@r{]} @r{[}@var{bnums}@dots{}@r{]}
2653 Enable the specified breakpoints (or all defined breakpoints). They
2654 become effective once again in stopping your program.
2656 @item enable @r{[}breakpoints@r{]} once @var{bnums}@dots{}
2657 Enable the specified breakpoints temporarily. @value{GDBN} disables any
2658 of these breakpoints immediately after stopping your program.
2660 @item enable @r{[}breakpoints@r{]} delete @var{bnums}@dots{}
2661 Enable the specified breakpoints to work once, then die. @value{GDBN}
2662 deletes any of these breakpoints as soon as your program stops there.
2665 Except for a breakpoint set with @code{tbreak} (@pxref{Set Breaks,
2666 ,Setting breakpoints}), breakpoints that you set are initially enabled;
2667 subsequently, they become disabled or enabled only when you use one of
2668 the commands above. (The command @code{until} can set and delete a
2669 breakpoint of its own, but it does not change the state of your other
2670 breakpoints; see @ref{Continuing and Stepping, ,Continuing and
2674 @subsection Break conditions
2675 @cindex conditional breakpoints
2676 @cindex breakpoint conditions
2678 @c FIXME what is scope of break condition expr? Context where wanted?
2679 @c in particular for a watchpoint?
2680 The simplest sort of breakpoint breaks every time your program reaches a
2681 specified place. You can also specify a @dfn{condition} for a
2682 breakpoint. A condition is just a Boolean expression in your
2683 programming language (@pxref{Expressions, ,Expressions}). A breakpoint with
2684 a condition evaluates the expression each time your program reaches it,
2685 and your program stops only if the condition is @emph{true}.
2687 This is the converse of using assertions for program validation; in that
2688 situation, you want to stop when the assertion is violated---that is,
2689 when the condition is false. In C, if you want to test an assertion expressed
2690 by the condition @var{assert}, you should set the condition
2691 @samp{! @var{assert}} on the appropriate breakpoint.
2693 Conditions are also accepted for watchpoints; you may not need them,
2694 since a watchpoint is inspecting the value of an expression anyhow---but
2695 it might be simpler, say, to just set a watchpoint on a variable name,
2696 and specify a condition that tests whether the new value is an interesting
2699 Break conditions can have side effects, and may even call functions in
2700 your program. This can be useful, for example, to activate functions
2701 that log program progress, or to use your own print functions to
2702 format special data structures. The effects are completely predictable
2703 unless there is another enabled breakpoint at the same address. (In
2704 that case, @value{GDBN} might see the other breakpoint first and stop your
2705 program without checking the condition of this one.) Note that
2706 breakpoint commands are usually more convenient and flexible for the
2707 purpose of performing side effects when a breakpoint is reached
2708 (@pxref{Break Commands, ,Breakpoint command lists}).
2710 Break conditions can be specified when a breakpoint is set, by using
2711 @samp{if} in the arguments to the @code{break} command. @xref{Set
2712 Breaks, ,Setting breakpoints}. They can also be changed at any time
2713 with the @code{condition} command.
2715 You can also use the @code{if} keyword with the @code{watch} command.
2716 The @code{catch} command does not recognize the @code{if} keyword;
2717 @code{condition} is the only way to impose a further condition on a
2722 @item condition @var{bnum} @var{expression}
2723 Specify @var{expression} as the break condition for breakpoint,
2724 watchpoint, or catchpoint number @var{bnum}. After you set a condition,
2725 breakpoint @var{bnum} stops your program only if the value of
2726 @var{expression} is true (nonzero, in C). When you use
2727 @code{condition}, @value{GDBN} checks @var{expression} immediately for
2728 syntactic correctness, and to determine whether symbols in it have
2729 referents in the context of your breakpoint.
2730 @c FIXME so what does GDB do if there is no referent? Moreover, what
2731 @c about watchpoints?
2733 not actually evaluate @var{expression} at the time the @code{condition}
2734 command is given, however. @xref{Expressions, ,Expressions}.
2736 @item condition @var{bnum}
2737 Remove the condition from breakpoint number @var{bnum}. It becomes
2738 an ordinary unconditional breakpoint.
2741 @cindex ignore count (of breakpoint)
2742 A special case of a breakpoint condition is to stop only when the
2743 breakpoint has been reached a certain number of times. This is so
2744 useful that there is a special way to do it, using the @dfn{ignore
2745 count} of the breakpoint. Every breakpoint has an ignore count, which
2746 is an integer. Most of the time, the ignore count is zero, and
2747 therefore has no effect. But if your program reaches a breakpoint whose
2748 ignore count is positive, then instead of stopping, it just decrements
2749 the ignore count by one and continues. As a result, if the ignore count
2750 value is @var{n}, the breakpoint does not stop the next @var{n} times
2751 your program reaches it.
2755 @item ignore @var{bnum} @var{count}
2756 Set the ignore count of breakpoint number @var{bnum} to @var{count}.
2757 The next @var{count} times the breakpoint is reached, your program's
2758 execution does not stop; other than to decrement the ignore count, @value{GDBN}
2761 To make the breakpoint stop the next time it is reached, specify
2764 When you use @code{continue} to resume execution of your program from a
2765 breakpoint, you can specify an ignore count directly as an argument to
2766 @code{continue}, rather than using @code{ignore}. @xref{Continuing and
2767 Stepping,,Continuing and stepping}.
2769 If a breakpoint has a positive ignore count and a condition, the
2770 condition is not checked. Once the ignore count reaches zero,
2771 @value{GDBN} resumes checking the condition.
2773 You could achieve the effect of the ignore count with a condition such
2774 as @w{@samp{$foo-- <= 0}} using a debugger convenience variable that
2775 is decremented each time. @xref{Convenience Vars, ,Convenience
2779 Ignore counts apply to breakpoints, watchpoints, and catchpoints.
2782 @node Break Commands
2783 @subsection Breakpoint command lists
2785 @cindex breakpoint commands
2786 You can give any breakpoint (or watchpoint or catchpoint) a series of
2787 commands to execute when your program stops due to that breakpoint. For
2788 example, you might want to print the values of certain expressions, or
2789 enable other breakpoints.
2794 @item commands @r{[}@var{bnum}@r{]}
2795 @itemx @dots{} @var{command-list} @dots{}
2797 Specify a list of commands for breakpoint number @var{bnum}. The commands
2798 themselves appear on the following lines. Type a line containing just
2799 @code{end} to terminate the commands.
2801 To remove all commands from a breakpoint, type @code{commands} and
2802 follow it immediately with @code{end}; that is, give no commands.
2804 With no @var{bnum} argument, @code{commands} refers to the last
2805 breakpoint, watchpoint, or catchpoint set (not to the breakpoint most
2806 recently encountered).
2809 Pressing @key{RET} as a means of repeating the last @value{GDBN} command is
2810 disabled within a @var{command-list}.
2812 You can use breakpoint commands to start your program up again. Simply
2813 use the @code{continue} command, or @code{step}, or any other command
2814 that resumes execution.
2816 Any other commands in the command list, after a command that resumes
2817 execution, are ignored. This is because any time you resume execution
2818 (even with a simple @code{next} or @code{step}), you may encounter
2819 another breakpoint---which could have its own command list, leading to
2820 ambiguities about which list to execute.
2823 If the first command you specify in a command list is @code{silent}, the
2824 usual message about stopping at a breakpoint is not printed. This may
2825 be desirable for breakpoints that are to print a specific message and
2826 then continue. If none of the remaining commands print anything, you
2827 see no sign that the breakpoint was reached. @code{silent} is
2828 meaningful only at the beginning of a breakpoint command list.
2830 The commands @code{echo}, @code{output}, and @code{printf} allow you to
2831 print precisely controlled output, and are often useful in silent
2832 breakpoints. @xref{Output, ,Commands for controlled output}.
2834 For example, here is how you could use breakpoint commands to print the
2835 value of @code{x} at entry to @code{foo} whenever @code{x} is positive.
2841 printf "x is %d\n",x
2846 One application for breakpoint commands is to compensate for one bug so
2847 you can test for another. Put a breakpoint just after the erroneous line
2848 of code, give it a condition to detect the case in which something
2849 erroneous has been done, and give it commands to assign correct values
2850 to any variables that need them. End with the @code{continue} command
2851 so that your program does not stop, and start with the @code{silent}
2852 command so that no output is produced. Here is an example:
2863 @node Breakpoint Menus
2864 @subsection Breakpoint menus
2866 @cindex symbol overloading
2868 Some programming languages (notably C++) permit a single function name
2869 to be defined several times, for application in different contexts.
2870 This is called @dfn{overloading}. When a function name is overloaded,
2871 @samp{break @var{function}} is not enough to tell @value{GDBN} where you want
2872 a breakpoint. If you realize this is a problem, you can use
2873 something like @samp{break @var{function}(@var{types})} to specify which
2874 particular version of the function you want. Otherwise, @value{GDBN} offers
2875 you a menu of numbered choices for different possible breakpoints, and
2876 waits for your selection with the prompt @samp{>}. The first two
2877 options are always @samp{[0] cancel} and @samp{[1] all}. Typing @kbd{1}
2878 sets a breakpoint at each definition of @var{function}, and typing
2879 @kbd{0} aborts the @code{break} command without setting any new
2882 For example, the following session excerpt shows an attempt to set a
2883 breakpoint at the overloaded symbol @code{String::after}.
2884 We choose three particular definitions of that function name:
2886 @c FIXME! This is likely to change to show arg type lists, at least
2889 (@value{GDBP}) b String::after
2892 [2] file:String.cc; line number:867
2893 [3] file:String.cc; line number:860
2894 [4] file:String.cc; line number:875
2895 [5] file:String.cc; line number:853
2896 [6] file:String.cc; line number:846
2897 [7] file:String.cc; line number:735
2899 Breakpoint 1 at 0xb26c: file String.cc, line 867.
2900 Breakpoint 2 at 0xb344: file String.cc, line 875.
2901 Breakpoint 3 at 0xafcc: file String.cc, line 846.
2902 Multiple breakpoints were set.
2903 Use the "delete" command to delete unwanted
2909 @c @ifclear BARETARGET
2910 @c @node Error in Breakpoints
2911 @c @subsection ``Cannot insert breakpoints''
2913 @c FIXME!! 14/6/95 Is there a real example of this? Let's use it.
2915 @c Under some operating systems
2916 @c any other process is running that program. In this situation,
2917 @c attempting to run or continue a program with a breakpoint causes
2918 @c @value{GDBN} to stop the other process.
2920 @c When this happens, you have three ways to proceed:
2924 @c Remove or disable the breakpoints, then continue.
2927 @c Suspend @value{GDBN}, and copy the file containing your program to a new
2928 @c name. Resume @value{GDBN} and use the @code{exec-file} command to specify
2929 @c that @value{GDBN} should run your program under that name.
2930 @c Then start your program again.
2933 @c Relink your program so that the text segment is nonsharable, using the
2934 @c linker option @samp{-N}. The operating system limitation may not apply
2935 @c to nonsharable executables.
2939 @node Continuing and Stepping
2940 @section Continuing and stepping
2944 @cindex resuming execution
2945 @dfn{Continuing} means resuming program execution until your program
2946 completes normally. In contrast, @dfn{stepping} means executing just
2947 one more ``step'' of your program, where ``step'' may mean either one
2948 line of source code, or one machine instruction (depending on what
2949 particular command you use). Either when continuing or when stepping,
2950 your program may stop even sooner, due to a breakpoint or a signal. (If
2951 due to a signal, you may want to use @code{handle}, or use @samp{signal
2952 0} to resume execution. @xref{Signals, ,Signals}.)
2958 @item continue @r{[}@var{ignore-count}@r{]}
2959 @itemx c @r{[}@var{ignore-count}@r{]}
2960 @itemx fg @r{[}@var{ignore-count}@r{]}
2961 Resume program execution, at the address where your program last stopped;
2962 any breakpoints set at that address are bypassed. The optional argument
2963 @var{ignore-count} allows you to specify a further number of times to
2964 ignore a breakpoint at this location; its effect is like that of
2965 @code{ignore} (@pxref{Conditions, ,Break conditions}).
2967 The argument @var{ignore-count} is meaningful only when your program
2968 stopped due to a breakpoint. At other times, the argument to
2969 @code{continue} is ignored.
2971 The synonyms @code{c} and @code{fg} are provided purely for convenience,
2972 and have exactly the same behavior as @code{continue}.
2975 To resume execution at a different place, you can use @code{return}
2976 (@pxref{Returning, ,Returning from a function}) to go back to the
2977 calling function; or @code{jump} (@pxref{Jumping, ,Continuing at a
2978 different address}) to go to an arbitrary location in your program.
2980 A typical technique for using stepping is to set a breakpoint
2981 (@pxref{Breakpoints, ,Breakpoints; watchpoints; and catchpoints}) at the
2982 beginning of the function or the section of your program where a problem
2983 is believed to lie, run your program until it stops at that breakpoint,
2984 and then step through the suspect area, examining the variables that are
2985 interesting, until you see the problem happen.
2991 Continue running your program until control reaches a different source
2992 line, then stop it and return control to @value{GDBN}. This command is
2993 abbreviated @code{s}.
2996 @c "without debugging information" is imprecise; actually "without line
2997 @c numbers in the debugging information". (gcc -g1 has debugging info but
2998 @c not line numbers). But it seems complex to try to make that
2999 @c distinction here.
3000 @emph{Warning:} If you use the @code{step} command while control is
3001 within a function that was compiled without debugging information,
3002 execution proceeds until control reaches a function that does have
3003 debugging information. Likewise, it will not step into a function which
3004 is compiled without debugging information. To step through functions
3005 without debugging information, use the @code{stepi} command, described
3009 The @code{step} command now only stops at the first instruction of a
3010 source line. This prevents the multiple stops that used to occur in
3011 switch statements, for loops, etc. @code{step} continues to stop if a
3012 function that has debugging information is called within the line.
3014 Also, the @code{step} command now only enters a subroutine if there is line
3015 number information for the subroutine. Otherwise it acts like the
3016 @code{next} command. This avoids problems when using @code{cc -gl}
3017 on MIPS machines. Previously, @code{step} entered subroutines if there
3018 was any debugging information about the routine.
3020 @item step @var{count}
3021 Continue running as in @code{step}, but do so @var{count} times. If a
3022 breakpoint is reached, or a signal not related to stepping occurs before
3023 @var{count} steps, stepping stops right away.
3027 @item next @r{[}@var{count}@r{]}
3028 Continue to the next source line in the current (innermost) stack frame.
3029 This is similar to @code{step}, but function calls that appear within
3030 the line of code are executed without stopping. Execution stops when
3031 control reaches a different line of code at the original stack level
3032 that was executing when you gave the @code{next} command. This command
3033 is abbreviated @code{n}.
3035 An argument @var{count} is a repeat count, as for @code{step}.
3038 @c FIX ME!! Do we delete this, or is there a way it fits in with
3039 @c the following paragraph? --- Vctoria
3041 @c @code{next} within a function that lacks debugging information acts like
3042 @c @code{step}, but any function calls appearing within the code of the
3043 @c function are executed without stopping.
3045 The @code{next} command now only stops at the first instruction of a
3046 source line. This prevents the multiple stops that used to occur in
3047 switch statements, for loops, etc.
3051 Continue running until just after function in the selected stack frame
3052 returns. Print the returned value (if any).
3054 Contrast this with the @code{return} command (@pxref{Returning,
3055 ,Returning from a function}).
3061 Continue running until a source line past the current line, in the
3062 current stack frame, is reached. This command is used to avoid single
3063 stepping through a loop more than once. It is like the @code{next}
3064 command, except that when @code{until} encounters a jump, it
3065 automatically continues execution until the program counter is greater
3066 than the address of the jump.
3068 This means that when you reach the end of a loop after single stepping
3069 though it, @code{until} makes your program continue execution until it
3070 exits the loop. In contrast, a @code{next} command at the end of a loop
3071 simply steps back to the beginning of the loop, which forces you to step
3072 through the next iteration.
3074 @code{until} always stops your program if it attempts to exit the current
3077 @code{until} may produce somewhat counterintuitive results if the order
3078 of machine code does not match the order of the source lines. For
3079 example, in the following excerpt from a debugging session, the @code{f}
3080 (@code{frame}) command shows that execution is stopped at line
3081 @code{206}; yet when we use @code{until}, we get to line @code{195}:
3085 #0 main (argc=4, argv=0xf7fffae8) at m4.c:206
3087 (@value{GDBP}) until
3088 195 for ( ; argc > 0; NEXTARG) @{
3091 This happened because, for execution efficiency, the compiler had
3092 generated code for the loop closure test at the end, rather than the
3093 start, of the loop---even though the test in a C @code{for}-loop is
3094 written before the body of the loop. The @code{until} command appeared
3095 to step back to the beginning of the loop when it advanced to this
3096 expression; however, it has not really gone to an earlier
3097 statement---not in terms of the actual machine code.
3099 @code{until} with no argument works by means of single
3100 instruction stepping, and hence is slower than @code{until} with an
3103 @item until @var{location}
3104 @itemx u @var{location}
3105 Continue running your program until either the specified location is
3106 reached, or the current stack frame returns. @var{location} is any of
3107 the forms of argument acceptable to @code{break} (@pxref{Set Breaks,
3108 ,Setting breakpoints}). This form of the command uses breakpoints,
3109 and hence is quicker than @code{until} without an argument.
3115 Execute one machine instruction, then stop and return to the debugger.
3117 It is often useful to do @samp{display/i $pc} when stepping by machine
3118 instructions. This makes @value{GDBN} automatically display the next
3119 instruction to be executed, each time your program stops. @xref{Auto
3120 Display,, Automatic display}.
3122 An argument is a repeat count, as in @code{step}.
3129 Execute one machine instruction, but if it is a function call,
3130 proceed until the function returns.
3132 An argument is a repeat count, as in @code{next}.
3139 A signal is an asynchronous event that can happen in a program. The
3140 operating system defines the possible kinds of signals, and gives each
3141 kind a name and a number. For example, in Unix @code{SIGINT} is the
3142 signal a program gets when you type an interrupt (often @kbd{C-c});
3143 @code{SIGSEGV} is the signal a program gets from referencing a place in
3144 memory far away from all the areas in use; @code{SIGALRM} occurs when
3145 the alarm clock timer goes off (which happens only if your program has
3146 requested an alarm).
3148 @cindex fatal signals
3149 Some signals, including @code{SIGALRM}, are a normal part of the
3150 functioning of your program. Others, such as @code{SIGSEGV}, indicate
3151 errors; these signals are @dfn{fatal} (kill your program immediately) if the
3152 program has not specified in advance some other way to handle the signal.
3153 @code{SIGINT} does not indicate an error in your program, but it is normally
3154 fatal so it can carry out the purpose of the interrupt: to kill the program.
3156 @value{GDBN} has the ability to detect any occurrence of a signal in your
3157 program. You can tell @value{GDBN} in advance what to do for each kind of
3160 @cindex handling signals
3161 Normally, @value{GDBN} is set up to ignore non-erroneous signals like @code{SIGALRM}
3162 (so as not to interfere with their role in the functioning of your program)
3163 but to stop your program immediately whenever an error signal happens.
3164 You can change these settings with the @code{handle} command.
3167 @kindex info signals
3169 Print a table of all the kinds of signals and how @value{GDBN} has been told to
3170 handle each one. You can use this to see the signal numbers of all
3171 the defined types of signals.
3173 @code{info handle} is the new alias for @code{info signals}.
3176 @item handle @var{signal} @var{keywords}@dots{}
3177 Change the way @value{GDBN} handles signal @var{signal}. @var{signal} can
3178 be the number of a signal or its name (with or without the @samp{SIG} at the
3179 beginning). The @var{keywords} say what change to make.
3183 The keywords allowed by the @code{handle} command can be abbreviated.
3184 Their full names are:
3188 @value{GDBN} should not stop your program when this signal happens. It may
3189 still print a message telling you that the signal has come in.
3192 @value{GDBN} should stop your program when this signal happens. This implies
3193 the @code{print} keyword as well.
3196 @value{GDBN} should print a message when this signal happens.
3199 @value{GDBN} should not mention the occurrence of the signal at all. This
3200 implies the @code{nostop} keyword as well.
3203 @value{GDBN} should allow your program to see this signal; your program
3204 can handle the signal, or else it may terminate if the signal is fatal
3208 @value{GDBN} should not allow your program to see this signal.
3212 When a signal stops your program, the signal is not visible until you
3213 continue. Your program sees the signal then, if @code{pass} is in
3214 effect for the signal in question @emph{at that time}. In other words,
3215 after @value{GDBN} reports a signal, you can use the @code{handle}
3216 command with @code{pass} or @code{nopass} to control whether your
3217 program sees that signal when you continue.
3219 You can also use the @code{signal} command to prevent your program from
3220 seeing a signal, or cause it to see a signal it normally would not see,
3221 or to give it any signal at any time. For example, if your program stopped
3222 due to some sort of memory reference error, you might store correct
3223 values into the erroneous variables and continue, hoping to see more
3224 execution; but your program would probably terminate immediately as
3225 a result of the fatal signal once it saw the signal. To prevent this,
3226 you can continue with @samp{signal 0}. @xref{Signaling, ,Giving your
3230 @section Stopping and starting multi-thread programs
3232 When your program has multiple threads (@pxref{Threads,, Debugging
3233 programs with multiple threads}), you can choose whether to set
3234 breakpoints on all threads, or on a particular thread.
3237 @cindex breakpoints and threads
3238 @cindex thread breakpoints
3239 @kindex break @dots{} thread @var{threadno}
3240 @item break @var{linespec} thread @var{threadno}
3241 @itemx break @var{linespec} thread @var{threadno} if @dots{}
3242 @var{linespec} specifies source lines; there are several ways of
3243 writing them, but the effect is always to specify some source line.
3245 Use the qualifier @samp{thread @var{threadno}} with a breakpoint command
3246 to specify that you only want @value{GDBN} to stop the program when a
3247 particular thread reaches this breakpoint. @var{threadno} is one of the
3248 numeric thread identifiers assigned by @value{GDBN}, shown in the first
3249 column of the @samp{info threads} display.
3251 If you do not specify @samp{thread @var{threadno}} when you set a
3252 breakpoint, the breakpoint applies to @emph{all} threads of your
3255 You can use the @code{thread} qualifier on conditional breakpoints as
3256 well; in this case, place @samp{thread @var{threadno}} before the
3257 breakpoint condition, like this:
3260 (gdb) break frik.c:13 thread 28 if bartab > lim
3265 @cindex stopped threads
3266 @cindex threads, stopped
3267 Whenever your program stops under @value{GDBN} for any reason,
3268 @emph{all} threads of execution stop, not just the current thread. This
3269 allows you to examine the overall state of the program, including
3270 switching between threads, without worrying that things may change
3273 @cindex continuing threads
3274 @cindex threads, continuing
3275 Conversely, whenever you restart the program, @emph{all} threads start
3276 executing. @emph{This is true even when single-stepping} with commands
3277 like @code{step} or @code{next}.
3279 In particular, @value{GDBN} cannot single-step all threads in lockstep.
3280 Since thread scheduling is up to your debugging target's operating
3281 system (not controlled by @value{GDBN}), other threads may
3282 execute more than one statement while the current thread completes a
3283 single step. Moreover, in general other threads stop in the middle of a
3284 statement, rather than at a clean statement boundary, when the program
3287 You might even find your program stopped in another thread after
3288 continuing or even single-stepping. This happens whenever some other
3289 thread runs into a breakpoint, a signal, or an exception before the
3290 first thread completes whatever you requested.
3292 On some OSes, you can lock the OS scheduler and thus allow only a single
3296 @item set scheduler-locking @var{mode}
3297 Set the scheduler locking mode. If it is @code{off}, then there is no
3298 locking and any thread may run at any time. If @code{on}, then only the
3299 current thread may run when the inferior is resumed. The @code{step}
3300 mode optimizes for single-stepping. It stops other threads from
3301 ``seizing the prompt'' by preempting the current thread while you are
3302 stepping. Other threads will only rarely (or never) get a chance to run
3303 when you step. They are more likely to run when you ``next'' over a
3304 function call, and they are completely free to run when you use commands
3305 like ``continue'', ``until'', or ``finish''. However, unless another
3306 thread hits a breakpoint during its timeslice, they will never steal the
3307 GDB prompt away from the thread that you are debugging.
3309 @item show scheduler-locking
3310 Display the current scheduler locking mode.
3315 @chapter Examining the Stack
3317 When your program has stopped, the first thing you need to know is where it
3318 stopped and how it got there.
3321 Each time your program performs a function call, information about the call
3323 That information includes the location of the call in your program,
3324 the arguments of the call,
3325 and the local variables of the function being called.
3326 The information is saved in a block of data called a @dfn{stack frame}.
3327 The stack frames are allocated in a region of memory called the @dfn{call
3330 When your program stops, the @value{GDBN} commands for examining the
3331 stack allow you to see all of this information.
3333 @cindex selected frame
3334 One of the stack frames is @dfn{selected} by @value{GDBN} and many
3335 @value{GDBN} commands refer implicitly to the selected frame. In
3336 particular, whenever you ask @value{GDBN} for the value of a variable in
3337 your program, the value is found in the selected frame. There are
3338 special @value{GDBN} commands to select whichever frame you are
3339 interested in. @xref{Selection, ,Selecting a frame}.
3341 When your program stops, @value{GDBN} automatically selects the
3342 currently executing frame and describes it briefly, similar to the
3343 @code{frame} command (@pxref{Frame Info, ,Information about a frame}).
3346 * Frames:: Stack frames
3347 * Backtrace:: Backtraces
3348 * Selection:: Selecting a frame
3349 * Frame Info:: Information on a frame
3354 @section Stack frames
3358 The call stack is divided up into contiguous pieces called @dfn{stack
3359 frames}, or @dfn{frames} for short; each frame is the data associated
3360 with one call to one function. The frame contains the arguments given
3361 to the function, the function's local variables, and the address at
3362 which the function is executing.
3364 @cindex initial frame
3365 @cindex outermost frame
3366 @cindex innermost frame
3367 When your program is started, the stack has only one frame, that of the
3368 function @code{main}. This is called the @dfn{initial} frame or the
3369 @dfn{outermost} frame. Each time a function is called, a new frame is
3370 made. Each time a function returns, the frame for that function invocation
3371 is eliminated. If a function is recursive, there can be many frames for
3372 the same function. The frame for the function in which execution is
3373 actually occurring is called the @dfn{innermost} frame. This is the most
3374 recently created of all the stack frames that still exist.
3376 @cindex frame pointer
3377 Inside your program, stack frames are identified by their addresses. A
3378 stack frame consists of many bytes, each of which has its own address; each
3379 kind of computer has a convention for choosing one byte whose
3380 address serves as the address of the frame. Usually this address is kept
3381 in a register called the @dfn{frame pointer register} while execution is
3382 going on in that frame.
3384 @cindex frame number
3385 @value{GDBN} assigns numbers to all existing stack frames, starting with
3386 zero for the innermost frame, one for the frame that called it,
3387 and so on upward. These numbers do not really exist in your program;
3388 they are assigned by @value{GDBN} to give you a way of designating stack
3389 frames in @value{GDBN} commands.
3391 @c below produces an acceptable overful hbox. --mew 13aug1993
3392 @cindex frameless execution
3393 Some compilers provide a way to compile functions so that they operate
3394 without stack frames. (For example, the @code{@value{GCC}} option
3395 @samp{-fomit-frame-pointer} generates functions without a frame.)
3396 This is occasionally done with heavily used library functions to save
3397 the frame setup time. @value{GDBN} has limited facilities for dealing
3398 with these function invocations. If the innermost function invocation
3399 has no stack frame, @value{GDBN} nevertheless regards it as though
3400 it had a separate frame, which is numbered zero as usual, allowing
3401 correct tracing of the function call chain. However, @value{GDBN} has
3402 no provision for frameless functions elsewhere in the stack.
3406 @item frame @var{args}
3407 The @code{frame} command allows you to move from one stack frame to another,
3408 and to print the stack frame you select. @var{args} may be either the
3409 address of the frame or the stack frame number. Without an argument,
3410 @code{frame} prints the current stack frame.
3412 @kindex select-frame
3414 The @code{select-frame} command allows you to move from one stack frame
3415 to another without printing the frame. This is the silent version of
3424 @cindex stack traces
3425 A backtrace is a summary of how your program got where it is. It shows one
3426 line per frame, for many frames, starting with the currently executing
3427 frame (frame zero), followed by its caller (frame one), and on up the
3435 Print a backtrace of the entire stack: one line per frame for all
3436 frames in the stack.
3438 You can stop the backtrace at any time by typing the system interrupt
3439 character, normally @kbd{C-c}.
3441 @item backtrace @var{n}
3443 Similar, but print only the innermost @var{n} frames.
3445 @item backtrace -@var{n}
3447 Similar, but print only the outermost @var{n} frames.
3453 The names @code{where} and @code{info stack} (abbreviated @code{info s})
3454 are additional aliases for @code{backtrace}.
3456 Each line in the backtrace shows the frame number and the function name.
3457 The program counter value is also shown---unless you use @code{set
3458 print address off}. The backtrace also shows the source file name and
3459 line number, as well as the arguments to the function. The program
3460 counter value is omitted if it is at the beginning of the code for that
3463 Here is an example of a backtrace. It was made with the command
3464 @samp{bt 3}, so it shows the innermost three frames.
3468 #0 m4_traceon (obs=0x24eb0, argc=1, argv=0x2b8c8)
3470 #1 0x6e38 in expand_macro (sym=0x2b600) at macro.c:242
3471 #2 0x6840 in expand_token (obs=0x0, t=177664, td=0xf7fffb08)
3473 (More stack frames follow...)
3478 The display for frame zero does not begin with a program counter
3479 value, indicating that your program has stopped at the beginning of the
3480 code for line @code{993} of @code{builtin.c}.
3483 @section Selecting a frame
3485 Most commands for examining the stack and other data in your program work on
3486 whichever stack frame is selected at the moment. Here are the commands for
3487 selecting a stack frame; all of them finish by printing a brief description
3488 of the stack frame just selected.
3495 Select frame number @var{n}. Recall that frame zero is the innermost
3496 (currently executing) frame, frame one is the frame that called the
3497 innermost one, and so on. The highest-numbered frame is the one for
3500 @item frame @var{addr}
3502 Select the frame at address @var{addr}. This is useful mainly if the
3503 chaining of stack frames has been damaged by a bug, making it
3504 impossible for @value{GDBN} to assign numbers properly to all frames. In
3505 addition, this can be useful when your program has multiple stacks and
3506 switches between them.
3508 On the SPARC architecture, @code{frame} needs two addresses to
3509 select an arbitrary frame: a frame pointer and a stack pointer.
3511 On the MIPS and Alpha architecture, it needs two addresses: a stack
3512 pointer and a program counter.
3514 On the 29k architecture, it needs three addresses: a register stack
3515 pointer, a program counter, and a memory stack pointer.
3516 @c note to future updaters: this is conditioned on a flag
3517 @c SETUP_ARBITRARY_FRAME in the tm-*.h files. The above is up to date
3518 @c as of 27 Jan 1994.
3522 Move @var{n} frames up the stack. For positive numbers @var{n}, this
3523 advances toward the outermost frame, to higher frame numbers, to frames
3524 that have existed longer. @var{n} defaults to one.
3529 Move @var{n} frames down the stack. For positive numbers @var{n}, this
3530 advances toward the innermost frame, to lower frame numbers, to frames
3531 that were created more recently. @var{n} defaults to one. You may
3532 abbreviate @code{down} as @code{do}.
3535 All of these commands end by printing two lines of output describing the
3536 frame. The first line shows the frame number, the function name, the
3537 arguments, and the source file and line number of execution in that
3538 frame. The second line shows the text of that source line.
3546 #1 0x22f0 in main (argc=1, argv=0xf7fffbf4, env=0xf7fffbfc)
3548 10 read_input_file (argv[i]);
3552 After such a printout, the @code{list} command with no arguments
3553 prints ten lines centered on the point of execution in the frame.
3554 @xref{List, ,Printing source lines}.
3557 @kindex down-silently
3559 @item up-silently @var{n}
3560 @itemx down-silently @var{n}
3561 These two commands are variants of @code{up} and @code{down},
3562 respectively; they differ in that they do their work silently, without
3563 causing display of the new frame. They are intended primarily for use
3564 in @value{GDBN} command scripts, where the output might be unnecessary and
3569 @section Information about a frame
3571 There are several other commands to print information about the selected
3577 When used without any argument, this command does not change which
3578 frame is selected, but prints a brief description of the currently
3579 selected stack frame. It can be abbreviated @code{f}. With an
3580 argument, this command is used to select a stack frame.
3581 @xref{Selection, ,Selecting a frame}.
3587 This command prints a verbose description of the selected stack frame,
3592 the address of the frame
3594 the address of the next frame down (called by this frame)
3596 the address of the next frame up (caller of this frame)
3598 the language in which the source code corresponding to this frame is written
3600 the address of the frame's arguments
3602 the program counter saved in it (the address of execution in the caller frame)
3604 which registers were saved in the frame
3607 @noindent The verbose description is useful when
3608 something has gone wrong that has made the stack format fail to fit
3609 the usual conventions.
3611 @item info frame @var{addr}
3612 @itemx info f @var{addr}
3613 Print a verbose description of the frame at address @var{addr}, without
3614 selecting that frame. The selected frame remains unchanged by this
3615 command. This requires the same kind of address (more than one for some
3616 architectures) that you specify in the @code{frame} command.
3617 @xref{Selection, ,Selecting a frame}.
3621 Print the arguments of the selected frame, each on a separate line.
3625 Print the local variables of the selected frame, each on a separate
3626 line. These are all variables (declared either static or automatic)
3627 accessible at the point of execution of the selected frame.
3630 @cindex catch exceptions
3631 @cindex exception handlers
3633 Print a list of all the exception handlers that are active in the
3634 current stack frame at the current point of execution. To see other
3635 exception handlers, visit the associated frame (using the @code{up},
3636 @code{down}, or @code{frame} commands); then type @code{info catch}.
3637 @xref{Set Catchpoints, , Setting catchpoints}.
3643 @chapter Examining Source Files
3645 @value{GDBN} can print parts of your program's source, since the debugging
3646 information recorded in the program tells @value{GDBN} what source files were
3647 used to build it. When your program stops, @value{GDBN} spontaneously prints
3648 the line where it stopped. Likewise, when you select a stack frame
3649 (@pxref{Selection, ,Selecting a frame}), @value{GDBN} prints the line where
3650 execution in that frame has stopped. You can print other portions of
3651 source files by explicit command.
3653 If you use @value{GDBN} through its @sc{gnu} Emacs interface, you may
3654 prefer to use Emacs facilities to view source; @pxref{Emacs, ,Using
3655 @value{GDBN} under @sc{gnu} Emacs}.
3658 * List:: Printing source lines
3659 * Search:: Searching source files
3660 * Source Path:: Specifying source directories
3661 * Machine Code:: Source and machine code
3665 @section Printing source lines
3669 To print lines from a source file, use the @code{list} command
3670 (abbreviated @code{l}). By default, ten lines are printed.
3671 There are several ways to specify what part of the file you want to print.
3673 Here are the forms of the @code{list} command most commonly used:
3676 @item list @var{linenum}
3677 Print lines centered around line number @var{linenum} in the
3678 current source file.
3680 @item list @var{function}
3681 Print lines centered around the beginning of function
3685 Print more lines. If the last lines printed were printed with a
3686 @code{list} command, this prints lines following the last lines
3687 printed; however, if the last line printed was a solitary line printed
3688 as part of displaying a stack frame (@pxref{Stack, ,Examining the
3689 Stack}), this prints lines centered around that line.
3692 Print lines just before the lines last printed.
3695 By default, @value{GDBN} prints ten source lines with any of these forms of
3696 the @code{list} command. You can change this using @code{set listsize}:
3699 @kindex set listsize
3700 @item set listsize @var{count}
3701 Make the @code{list} command display @var{count} source lines (unless
3702 the @code{list} argument explicitly specifies some other number).
3704 @kindex show listsize
3706 Display the number of lines that @code{list} prints.
3709 Repeating a @code{list} command with @key{RET} discards the argument,
3710 so it is equivalent to typing just @code{list}. This is more useful
3711 than listing the same lines again. An exception is made for an
3712 argument of @samp{-}; that argument is preserved in repetition so that
3713 each repetition moves up in the source file.
3716 In general, the @code{list} command expects you to supply zero, one or two
3717 @dfn{linespecs}. Linespecs specify source lines; there are several ways
3718 of writing them but the effect is always to specify some source line.
3719 Here is a complete description of the possible arguments for @code{list}:
3722 @item list @var{linespec}
3723 Print lines centered around the line specified by @var{linespec}.
3725 @item list @var{first},@var{last}
3726 Print lines from @var{first} to @var{last}. Both arguments are
3729 @item list ,@var{last}
3730 Print lines ending with @var{last}.
3732 @item list @var{first},
3733 Print lines starting with @var{first}.
3736 Print lines just after the lines last printed.
3739 Print lines just before the lines last printed.
3742 As described in the preceding table.
3745 Here are the ways of specifying a single source line---all the
3750 Specifies line @var{number} of the current source file.
3751 When a @code{list} command has two linespecs, this refers to
3752 the same source file as the first linespec.
3755 Specifies the line @var{offset} lines after the last line printed.
3756 When used as the second linespec in a @code{list} command that has
3757 two, this specifies the line @var{offset} lines down from the
3761 Specifies the line @var{offset} lines before the last line printed.
3763 @item @var{filename}:@var{number}
3764 Specifies line @var{number} in the source file @var{filename}.
3766 @item @var{function}
3767 Specifies the line that begins the body of the function @var{function}.
3768 For example: in C, this is the line with the open brace.
3770 @item @var{filename}:@var{function}
3771 Specifies the line of the open-brace that begins the body of the
3772 function @var{function} in the file @var{filename}. You only need the
3773 file name with a function name to avoid ambiguity when there are
3774 identically named functions in different source files.
3776 @item *@var{address}
3777 Specifies the line containing the program address @var{address}.
3778 @var{address} may be any expression.
3782 @section Searching source files
3784 @kindex reverse-search
3786 There are two commands for searching through the current source file for a
3791 @kindex forward-search
3792 @item forward-search @var{regexp}
3793 @itemx search @var{regexp}
3794 The command @samp{forward-search @var{regexp}} checks each line,
3795 starting with the one following the last line listed, for a match for
3796 @var{regexp}. It lists the line that is found. You can use the
3797 synonym @samp{search @var{regexp}} or abbreviate the command name as
3800 @item reverse-search @var{regexp}
3801 The command @samp{reverse-search @var{regexp}} checks each line, starting
3802 with the one before the last line listed and going backward, for a match
3803 for @var{regexp}. It lists the line that is found. You can abbreviate
3804 this command as @code{rev}.
3808 @section Specifying source directories
3811 @cindex directories for source files
3812 Executable programs sometimes do not record the directories of the source
3813 files from which they were compiled, just the names. Even when they do,
3814 the directories could be moved between the compilation and your debugging
3815 session. @value{GDBN} has a list of directories to search for source files;
3816 this is called the @dfn{source path}. Each time @value{GDBN} wants a source file,
3817 it tries all the directories in the list, in the order they are present
3818 in the list, until it finds a file with the desired name. Note that
3819 the executable search path is @emph{not} used for this purpose. Neither is
3820 the current working directory, unless it happens to be in the source
3823 If @value{GDBN} cannot find a source file in the source path, and the
3824 object program records a directory, @value{GDBN} tries that directory
3825 too. If the source path is empty, and there is no record of the
3826 compilation directory, @value{GDBN} looks in the current directory as a
3829 Whenever you reset or rearrange the source path, @value{GDBN} clears out
3830 any information it has cached about where source files are found and where
3831 each line is in the file.
3835 When you start @value{GDBN}, its source path is empty.
3836 To add other directories, use the @code{directory} command.
3839 @item directory @var{dirname} @dots{}
3840 @item dir @var{dirname} @dots{}
3841 Add directory @var{dirname} to the front of the source path. Several
3842 directory names may be given to this command, separated by @samp{:} or
3843 whitespace. You may specify a directory that is already in the source
3844 path; this moves it forward, so @value{GDBN} searches it sooner.
3850 @cindex compilation directory
3851 @cindex current directory
3852 @cindex working directory
3853 @cindex directory, current
3854 @cindex directory, compilation
3855 You can use the string @samp{$cdir} to refer to the compilation
3856 directory (if one is recorded), and @samp{$cwd} to refer to the current
3857 working directory. @samp{$cwd} is not the same as @samp{.}---the former
3858 tracks the current working directory as it changes during your @value{GDBN}
3859 session, while the latter is immediately expanded to the current
3860 directory at the time you add an entry to the source path.
3863 Reset the source path to empty again. This requires confirmation.
3865 @c RET-repeat for @code{directory} is explicitly disabled, but since
3866 @c repeating it would be a no-op we do not say that. (thanks to RMS)
3868 @item show directories
3869 @kindex show directories
3870 Print the source path: show which directories it contains.
3873 If your source path is cluttered with directories that are no longer of
3874 interest, @value{GDBN} may sometimes cause confusion by finding the wrong
3875 versions of source. You can correct the situation as follows:
3879 Use @code{directory} with no argument to reset the source path to empty.
3882 Use @code{directory} with suitable arguments to reinstall the
3883 directories you want in the source path. You can add all the
3884 directories in one command.
3888 @section Source and machine code
3890 You can use the command @code{info line} to map source lines to program
3891 addresses (and vice versa), and the command @code{disassemble} to display
3892 a range of addresses as machine instructions. When run under @sc{gnu} Emacs
3893 mode, the @code{info line} command now causes the arrow to point to the
3894 line specified. Also, @code{info line} prints addresses in symbolic form as
3899 @item info line @var{linespec}
3900 Print the starting and ending addresses of the compiled code for
3901 source line @var{linespec}. You can specify source lines in any of
3902 the ways understood by the @code{list} command (@pxref{List, ,Printing
3906 For example, we can use @code{info line} to discover the location of
3907 the object code for the first line of function
3908 @code{m4_changequote}:
3911 (@value{GDBP}) info line m4_changecom
3912 Line 895 of "builtin.c" starts at pc 0x634c and ends at 0x6350.
3916 We can also inquire (using @code{*@var{addr}} as the form for
3917 @var{linespec}) what source line covers a particular address:
3919 (@value{GDBP}) info line *0x63ff
3920 Line 926 of "builtin.c" starts at pc 0x63e4 and ends at 0x6404.
3923 @cindex @code{$_} and @code{info line}
3924 After @code{info line}, the default address for the @code{x} command
3925 is changed to the starting address of the line, so that @samp{x/i} is
3926 sufficient to begin examining the machine code (@pxref{Memory,
3927 ,Examining memory}). Also, this address is saved as the value of the
3928 convenience variable @code{$_} (@pxref{Convenience Vars, ,Convenience
3933 @cindex assembly instructions
3934 @cindex instructions, assembly
3935 @cindex machine instructions
3936 @cindex listing machine instructions
3938 This specialized command dumps a range of memory as machine
3939 instructions. The default memory range is the function surrounding the
3940 program counter of the selected frame. A single argument to this
3941 command is a program counter value; @value{GDBN} dumps the function
3942 surrounding this value. Two arguments specify a range of addresses
3943 (first inclusive, second exclusive) to dump.
3946 The following example shows the disassembly of a range of addresses of
3947 HP PA-RISC 2.0 code:
3950 (@value{GDBP}) disas 0x32c4 0x32e4
3951 Dump of assembler code from 0x32c4 to 0x32e4:
3952 0x32c4 <main+204>: addil 0,dp
3953 0x32c8 <main+208>: ldw 0x22c(sr0,r1),r26
3954 0x32cc <main+212>: ldil 0x3000,r31
3955 0x32d0 <main+216>: ble 0x3f8(sr4,r31)
3956 0x32d4 <main+220>: ldo 0(r31),rp
3957 0x32d8 <main+224>: addil -0x800,dp
3958 0x32dc <main+228>: ldo 0x588(r1),r26
3959 0x32e0 <main+232>: ldil 0x3000,r31
3960 End of assembler dump.
3963 Some architectures have more than one commonly-used set of instruction
3964 mnemonics or other syntax.
3967 @kindex set assembly-language
3968 @cindex assembly instructions
3969 @cindex instructions, assembly
3970 @cindex machine instructions
3971 @cindex listing machine instructions
3972 @item set assembly-language @var{instruction-set}
3973 Select the instruction set to use when disassembling the
3974 program via the @code{disassemble} or @code{x/i} commands.
3976 Currently this command is only defined for the Intel x86 family. You
3977 can set @var{instruction-set} to either @code{i386} or @code{i8086}.
3978 The default is @code{i386}.
3983 @chapter Examining Data
3985 @cindex printing data
3986 @cindex examining data
3989 @c "inspect" is not quite a synonym if you are using Epoch, which we do not
3990 @c document because it is nonstandard... Under Epoch it displays in a
3991 @c different window or something like that.
3992 The usual way to examine data in your program is with the @code{print}
3993 command (abbreviated @code{p}), or its synonym @code{inspect}. It
3994 evaluates and prints the value of an expression of the language your
3995 program is written in (@pxref{Languages, ,Using @value{GDBN} with
3996 Different Languages}).
3999 @item print @var{exp}
4000 @itemx print /@var{f} @var{exp}
4001 @var{exp} is an expression (in the source language). By default the
4002 value of @var{exp} is printed in a format appropriate to its data type;
4003 you can choose a different format by specifying @samp{/@var{f}}, where
4004 @var{f} is a letter specifying the format; @pxref{Output Formats,,Output
4008 @itemx print /@var{f}
4009 If you omit @var{exp}, @value{GDBN} displays the last value again (from the
4010 @dfn{value history}; @pxref{Value History, ,Value history}). This allows you to
4011 conveniently inspect the same value in an alternative format.
4014 A more low-level way of examining data is with the @code{x} command.
4015 It examines data in memory at a specified address and prints it in a
4016 specified format. @xref{Memory, ,Examining memory}.
4018 If you are interested in information about types, or about how the
4019 fields of a struct or class are declared, use the @code{ptype @var{exp}}
4020 command rather than @code{print}. @xref{Symbols, ,Examining the Symbol
4024 * Expressions:: Expressions
4025 * Variables:: Program variables
4026 * Arrays:: Artificial arrays
4027 * Output Formats:: Output formats
4028 * Memory:: Examining memory
4029 * Auto Display:: Automatic display
4030 * Print Settings:: Print settings
4031 * Value History:: Value history
4032 * Convenience Vars:: Convenience variables
4033 * Registers:: Registers
4034 * Floating Point Hardware:: Floating point hardware
4038 @section Expressions
4041 @code{print} and many other @value{GDBN} commands accept an expression and
4042 compute its value. Any kind of constant, variable or operator defined
4043 by the programming language you are using is valid in an expression in
4044 @value{GDBN}. This includes conditional expressions, function calls, casts
4045 and string constants. It unfortunately does not include symbols defined
4046 by preprocessor @code{#define} commands.
4048 @value{GDBN} now supports array constants in expressions input by
4049 the user. The syntax is @var{@{element, element@dots{}@}}. For example,
4050 you can now use the command @code{print @{1, 2, 3@}} to build up an array in
4051 memory that is malloc'd in the target program.
4053 Because C is so widespread, most of the expressions shown in examples in
4054 this manual are in C. @xref{Languages, , Using @value{GDBN} with Different
4055 Languages}, for information on how to use expressions in other
4058 In this section, we discuss operators that you can use in @value{GDBN}
4059 expressions regardless of your programming language.
4061 Casts are supported in all languages, not just in C, because it is so
4062 useful to cast a number into a pointer in order to examine a structure
4063 at that address in memory.
4064 @c FIXME: casts supported---Mod2 true?
4066 @value{GDBN} supports these operators, in addition to those common
4067 to programming languages:
4071 @samp{@@} is a binary operator for treating parts of memory as arrays.
4072 @xref{Arrays, ,Artificial arrays}, for more information.
4075 @samp{::} allows you to specify a variable in terms of the file or
4076 function where it is defined. @xref{Variables, ,Program variables}.
4078 @cindex @{@var{type}@}
4079 @cindex type casting memory
4080 @cindex memory, viewing as typed object
4081 @cindex casts, to view memory
4082 @item @{@var{type}@} @var{addr}
4083 Refers to an object of type @var{type} stored at address @var{addr} in
4084 memory. @var{addr} may be any expression whose value is an integer or
4085 pointer (but parentheses are required around binary operators, just as in
4086 a cast). This construct is allowed regardless of what kind of data is
4087 normally supposed to reside at @var{addr}.
4091 @section Program variables
4093 The most common kind of expression to use is the name of a variable
4096 Variables in expressions are understood in the selected stack frame
4097 (@pxref{Selection, ,Selecting a frame}); they must be either:
4101 global (or file-static)
4108 visible according to the scope rules of the
4109 programming language from the point of execution in that frame
4112 @noindent This means that in the function
4127 you can examine and use the variable @code{a} whenever your program is
4128 executing within the function @code{foo}, but you can only use or
4129 examine the variable @code{b} while your program is executing inside
4130 the block where @code{b} is declared.
4132 @cindex variable name conflict
4133 There is an exception: you can refer to a variable or function whose
4134 scope is a single source file even if the current execution point is not
4135 in this file. But it is possible to have more than one such variable or
4136 function with the same name (in different source files). If that
4137 happens, referring to that name has unpredictable effects. If you wish,
4138 you can specify a static variable in a particular function or file,
4139 using the colon-colon notation:
4143 @c info cannot cope with a :: index entry, but why deprive hard copy readers?
4147 @var{file}::@var{variable}
4148 @var{function}::@var{variable}
4152 Here @var{file} or @var{function} is the name of the context for the
4153 static @var{variable}. In the case of file names, you can use quotes to
4154 make sure @value{GDBN} parses the file name as a single word---for example,
4155 to print a global value of @code{x} defined in @file{f2.c}:
4158 (@value{GDBP}) p 'f2.c'::x
4161 @cindex C++ scope resolution
4162 This use of @samp{::} is very rarely in conflict with the very similar
4163 use of the same notation in C++. @value{GDBN} also supports use of the C++
4164 scope resolution operator in @value{GDBN} expressions.
4165 @c FIXME: Um, so what happens in one of those rare cases where it's in
4168 @cindex wrong values
4169 @cindex variable values, wrong
4171 @emph{Warning:} Occasionally, a local variable may appear to have the
4172 wrong value at certain points in a function---just after entry to a new
4173 scope, and just before exit.
4175 You may see this problem when you are stepping by machine instructions.
4176 This is because, on most machines, it takes more than one instruction to
4177 set up a stack frame (including local variable definitions); if you are
4178 stepping by machine instructions, variables may appear to have the wrong
4179 values until the stack frame is completely built. On exit, it usually
4180 also takes more than one machine instruction to destroy a stack frame;
4181 after you begin stepping through that group of instructions, local
4182 variable definitions may be gone.
4184 This may also happen when the compiler does significant optimizations.
4185 To be sure of always seeing accurate values, turn off all optimization
4189 @section Artificial arrays
4191 @cindex artificial array
4193 It is often useful to print out several successive objects of the
4194 same type in memory; a section of an array, or an array of
4195 dynamically determined size for which only a pointer exists in the
4198 You can do this by referring to a contiguous span of memory as an
4199 @dfn{artificial array}, using the binary operator @samp{@@}. The left
4200 operand of @samp{@@} should be the first element of the desired array
4201 and be an individual object. The right operand should be the desired length
4202 of the array. The result is an array value whose elements are all of
4203 the type of the left argument. The first element is actually the left
4204 argument; the second element comes from bytes of memory immediately
4205 following those that hold the first element, and so on. Here is an
4206 example. If a program says
4209 int *array = (int *) malloc (len * sizeof (int));
4213 you can print the contents of @code{array} with
4219 The left operand of @samp{@@} must reside in memory. Array values made
4220 with @samp{@@} in this way behave just like other arrays in terms of
4221 subscripting, and are coerced to pointers when used in expressions.
4222 Artificial arrays most often appear in expressions via the value history
4223 (@pxref{Value History, ,Value history}), after printing one out.
4225 Another way to create an artificial array is to use a cast.
4226 This re-interprets a value as if it were an array.
4227 The value need not be in memory:
4229 (@value{GDBP}) p/x (short[2])0x12345678
4230 $1 = @{0x1234, 0x5678@}
4233 As a convenience, if you leave the array length out (as in
4234 @samp{(@var{type})[])@var{value}}) gdb calculates the size to fill
4235 the value (as @samp{sizeof(@var{value})/sizeof(@var{type})}:
4237 (@value{GDBP}) p/x (short[])0x12345678
4238 $2 = @{0x1234, 0x5678@}
4241 Sometimes the artificial array mechanism is not quite enough; in
4242 moderately complex data structures, the elements of interest may not
4243 actually be adjacent---for example, if you are interested in the values
4244 of pointers in an array. One useful work-around in this situation is
4245 to use a convenience variable (@pxref{Convenience Vars, ,Convenience
4246 variables}) as a counter in an expression that prints the first
4247 interesting value, and then repeat that expression via @key{RET}. For
4248 instance, suppose you have an array @code{dtab} of pointers to
4249 structures, and you are interested in the values of a field @code{fv}
4250 in each structure. Here is an example of what you might type:
4260 @node Output Formats
4261 @section Output formats
4263 @cindex formatted output
4264 @cindex output formats
4265 By default, @value{GDBN} prints a value according to its data type. Sometimes
4266 this is not what you want. For example, you might want to print a number
4267 in hex, or a pointer in decimal. Or you might want to view data in memory
4268 at a certain address as a character string or as an instruction. To do
4269 these things, specify an @dfn{output format} when you print a value.
4271 The simplest use of output formats is to say how to print a value
4272 already computed. This is done by starting the arguments of the
4273 @code{print} command with a slash and a format letter. The format
4274 letters supported are:
4278 Regard the bits of the value as an integer, and print the integer in
4282 Print as integer in signed decimal.
4285 Print as integer in unsigned decimal.
4288 Print as integer in octal.
4291 Print as integer in binary. The letter @samp{t} stands for ``two''.
4292 @footnote{@samp{b} cannot be used because these format letters are also
4293 used with the @code{x} command, where @samp{b} stands for ``byte'';
4294 @pxref{Memory,,Examining memory}.}
4297 @cindex unknown address, locating
4298 Print as an address, both absolute in hexadecimal and as an offset from
4299 the nearest preceding symbol. You can use this format used to discover
4300 where (in what function) an unknown address is located:
4303 (@value{GDBP}) p/a 0x54320
4304 $3 = 0x54320 <_initialize_vx+396>
4308 Regard as an integer and print it as a character constant.
4311 Regard the bits of the value as a floating point number and print
4312 using typical floating point syntax.
4315 For example, to print the program counter in hex (@pxref{Registers}), type
4322 Note that no space is required before the slash; this is because command
4323 names in @value{GDBN} cannot contain a slash.
4325 To reprint the last value in the value history with a different format,
4326 you can use the @code{print} command with just a format and no
4327 expression. For example, @samp{p/x} reprints the last value in hex.
4330 @section Examining memory
4332 You can use the command @code{x} (for ``examine'') to examine memory in
4333 any of several formats, independently of your program's data types.
4335 @cindex examining memory
4338 @item x/@var{nfu} @var{addr}
4341 Use the @code{x} command to examine memory.
4344 @var{n}, @var{f}, and @var{u} are all optional parameters that specify how
4345 much memory to display and how to format it; @var{addr} is an
4346 expression giving the address where you want to start displaying memory.
4347 If you use defaults for @var{nfu}, you need not type the slash @samp{/}.
4348 Several commands set convenient defaults for @var{addr}.
4351 @item @var{n}, the repeat count
4352 The repeat count is a decimal integer; the default is 1. It specifies
4353 how much memory (counting by units @var{u}) to display.
4354 @c This really is **decimal**; unaffected by 'set radix' as of GDB
4357 @item @var{f}, the display format
4358 The display format is one of the formats used by @code{print},
4359 @samp{s} (null-terminated string), or @samp{i} (machine instruction).
4360 The default is @samp{x} (hexadecimal) initially.
4361 The default changes each time you use either @code{x} or @code{print}.
4363 @item @var{u}, the unit size
4364 The unit size is any of
4370 Halfwords (two bytes).
4372 Words (four bytes). This is the initial default.
4374 Giant words (eight bytes).
4377 Each time you specify a unit size with @code{x}, that size becomes the
4378 default unit the next time you use @code{x}. (For the @samp{s} and
4379 @samp{i} formats, the unit size is ignored and is normally not written.)
4381 @item @var{addr}, starting display address
4382 @var{addr} is the address where you want @value{GDBN} to begin displaying
4383 memory. The expression need not have a pointer value (though it may);
4384 it is always interpreted as an integer address of a byte of memory.
4385 @xref{Expressions, ,Expressions}, for more information on expressions. The default for
4386 @var{addr} is usually just after the last address examined---but several
4387 other commands also set the default address: @code{info breakpoints} (to
4388 the address of the last breakpoint listed), @code{info line} (to the
4389 starting address of a line), and @code{print} (if you use it to display
4390 a value from memory).
4393 For example, @samp{x/3uh 0x54320} is a request to display three halfwords
4394 (@code{h}) of memory, formatted as unsigned decimal integers (@samp{u}),
4395 starting at address @code{0x54320}. @samp{x/4xw $sp} prints the four
4396 words (@samp{w}) of memory above the stack pointer (here, @samp{$sp};
4397 @pxref{Registers}) in hexadecimal (@samp{x}).
4399 Since the letters indicating unit sizes are all distinct from the
4400 letters specifying output formats, you do not have to remember whether
4401 unit size or format comes first; either order works. The output
4402 specifications @samp{4xw} and @samp{4wx} mean exactly the same thing.
4403 (However, the count @var{n} must come first; @samp{wx4} does not work.)
4405 Even though the unit size @var{u} is ignored for the formats @samp{s}
4406 and @samp{i}, you might still want to use a count @var{n}; for example,
4407 @samp{3i} specifies that you want to see three machine instructions,
4408 including any operands. The command @code{disassemble} gives an
4409 alternative way of inspecting machine instructions; @pxref{Machine
4410 Code,,Source and machine code}.
4412 All the defaults for the arguments to @code{x} are designed to make it
4413 easy to continue scanning memory with minimal specifications each time
4414 you use @code{x}. For example, after you have inspected three machine
4415 instructions with @samp{x/3i @var{addr}}, you can inspect the next seven
4416 with just @samp{x/7}. If you use @key{RET} to repeat the @code{x} command,
4417 the repeat count @var{n} is used again; the other arguments default as
4418 for successive uses of @code{x}.
4420 @cindex @code{$_}, @code{$__}, and value history
4421 The addresses and contents printed by the @code{x} command are not saved
4422 in the value history because there is often too much of them and they
4423 would get in the way. Instead, @value{GDBN} makes these values available for
4424 subsequent use in expressions as values of the convenience variables
4425 @code{$_} and @code{$__}. After an @code{x} command, the last address
4426 examined is available for use in expressions in the convenience variable
4427 @code{$_}. The contents of that address, as examined, are available in
4428 the convenience variable @code{$__}.
4430 If the @code{x} command has a repeat count, the address and contents saved
4431 are from the last memory unit printed; this is not the same as the last
4432 address printed if several units were printed on the last line of output.
4435 @section Automatic display
4436 @cindex automatic display
4437 @cindex display of expressions
4439 If you find that you want to print the value of an expression frequently
4440 (to see how it changes), you might want to add it to the @dfn{automatic
4441 display list} so that @value{GDBN} prints its value each time your program stops.
4442 Each expression added to the list is given a number to identify it;
4443 to remove an expression from the list, you specify that number.
4444 The automatic display looks like this:
4448 3: bar[5] = (struct hack *) 0x3804
4452 This display shows item numbers, expressions and their current values. As with
4453 displays you request manually using @code{x} or @code{print}, you can
4454 specify the output format you prefer; in fact, @code{display} decides
4455 whether to use @code{print} or @code{x} depending on how elaborate your
4456 format specification is---it uses @code{x} if you specify a unit size,
4457 or one of the two formats (@samp{i} and @samp{s}) that are only
4458 supported by @code{x}; otherwise it uses @code{print}.
4462 @item display @var{exp}
4463 Add the expression @var{exp} to the list of expressions to display
4464 each time your program stops. @xref{Expressions, ,Expressions}.
4466 @code{display} does not repeat if you press @key{RET} again after using it.
4468 @item display/@var{fmt} @var{exp}
4469 For @var{fmt} specifying only a display format and not a size or
4470 count, add the expression @var{exp} to the auto-display list but
4471 arrange to display it each time in the specified format @var{fmt}.
4472 @xref{Output Formats,,Output formats}.
4474 @item display/@var{fmt} @var{addr}
4475 For @var{fmt} @samp{i} or @samp{s}, or including a unit-size or a
4476 number of units, add the expression @var{addr} as a memory address to
4477 be examined each time your program stops. Examining means in effect
4478 doing @samp{x/@var{fmt} @var{addr}}. @xref{Memory, ,Examining memory}.
4481 For example, @samp{display/i $pc} can be helpful, to see the machine
4482 instruction about to be executed each time execution stops (@samp{$pc}
4483 is a common name for the program counter; @pxref{Registers}).
4486 @kindex delete display
4488 @item undisplay @var{dnums}@dots{}
4489 @itemx delete display @var{dnums}@dots{}
4490 Remove item numbers @var{dnums} from the list of expressions to display.
4492 @code{undisplay} does not repeat if you press @key{RET} after using it.
4493 (Otherwise you would just get the error @samp{No display number @dots{}}.)
4495 @kindex disable display
4496 @item disable display @var{dnums}@dots{}
4497 Disable the display of item numbers @var{dnums}. A disabled display
4498 item is not printed automatically, but is not forgotten. It may be
4499 enabled again later.
4501 @kindex enable display
4502 @item enable display @var{dnums}@dots{}
4503 Enable display of item numbers @var{dnums}. It becomes effective once
4504 again in auto display of its expression, until you specify otherwise.
4507 Display the current values of the expressions on the list, just as is
4508 done when your program stops.
4510 @kindex info display
4512 Print the list of expressions previously set up to display
4513 automatically, each one with its item number, but without showing the
4514 values. This includes disabled expressions, which are marked as such.
4515 It also includes expressions which would not be displayed right now
4516 because they refer to automatic variables not currently available.
4519 If a display expression refers to local variables, then it does not make
4520 sense outside the lexical context for which it was set up. Such an
4521 expression is disabled when execution enters a context where one of its
4522 variables is not defined. For example, if you give the command
4523 @code{display last_char} while inside a function with an argument
4524 @code{last_char}, @value{GDBN} displays this argument while your program
4525 continues to stop inside that function. When it stops elsewhere---where
4526 there is no variable @code{last_char}---the display is disabled
4527 automatically. The next time your program stops where @code{last_char}
4528 is meaningful, you can enable the display expression once again.
4530 @node Print Settings
4531 @section Print settings
4533 @cindex format options
4534 @cindex print settings
4535 @value{GDBN} provides the following ways to control how arrays, structures,
4536 and symbols are printed.
4539 These settings are useful for debugging programs in any language:
4542 @kindex set print address
4543 @item set print address
4544 @itemx set print address on
4545 @value{GDBN} prints memory addresses showing the location of stack
4546 traces, structure values, pointer values, breakpoints, and so forth,
4547 even when it also displays the contents of those addresses. The default
4548 is @code{on}. For example, this is what a stack frame display looks like with
4549 @code{set print address on}:
4554 #0 set_quotes (lq=0x34c78 "<<", rq=0x34c88 ">>")
4556 530 if (lquote != def_lquote)
4560 @item set print address off
4561 Do not print addresses when displaying their contents. For example,
4562 this is the same stack frame displayed with @code{set print address off}:
4566 (@value{GDBP}) set print addr off
4568 #0 set_quotes (lq="<<", rq=">>") at input.c:530
4569 530 if (lquote != def_lquote)
4573 You can use @samp{set print address off} to eliminate all machine
4574 dependent displays from the @value{GDBN} interface. For example, with
4575 @code{print address off}, you should get the same text for backtraces on
4576 all machines---whether or not they involve pointer arguments.
4578 @kindex show print address
4579 @item show print address
4580 Show whether or not addresses are to be printed.
4583 When @value{GDBN} prints a symbolic address, it normally prints the
4584 closest earlier symbol plus an offset. If that symbol does not uniquely
4585 identify the address (for example, it is a name whose scope is a single
4586 source file), you may need to clarify. One way to do this is with
4587 @code{info line}, for example @samp{info line *0x4537}. Alternately,
4588 you can set @value{GDBN} to print the source file and line number when
4589 it prints a symbolic address:
4592 @kindex set print symbol-filename
4593 @item set print symbol-filename on
4594 Tell @value{GDBN} to print the source file name and line number of a
4595 symbol in the symbolic form of an address.
4597 @item set print symbol-filename off
4598 Do not print source file name and line number of a symbol. This is the
4601 @kindex show print symbol-filename
4602 @item show print symbol-filename
4603 Show whether or not @value{GDBN} will print the source file name and
4604 line number of a symbol in the symbolic form of an address.
4607 Another situation where it is helpful to show symbol filenames and line
4608 numbers is when disassembling code; @value{GDBN} shows you the line
4609 number and source file that corresponds to each instruction.
4611 Also, you may wish to see the symbolic form only if the address being
4612 printed is reasonably close to the closest earlier symbol:
4615 @kindex set print max-symbolic-offset
4616 @item set print max-symbolic-offset @var{max-offset}
4617 Tell @value{GDBN} to only display the symbolic form of an address if the
4618 offset between the closest earlier symbol and the address is less than
4619 @var{max-offset}. The default is 0, which tells @value{GDBN}
4620 to always print the symbolic form of an address if any symbol precedes it.
4622 @kindex show print max-symbolic-offset
4623 @item show print max-symbolic-offset
4624 Ask how large the maximum offset is that @value{GDBN} prints in a
4628 @cindex wild pointer, interpreting
4629 @cindex pointer, finding referent
4630 If you have a pointer and you are not sure where it points, try
4631 @samp{set print symbol-filename on}. Then you can determine the name
4632 and source file location of the variable where it points, using
4633 @samp{p/a @var{pointer}}. This interprets the address in symbolic form.
4634 For example, here @value{GDBN} shows that a variable @code{ptt} points
4635 at another variable @code{t}, defined in @file{hi2.c}:
4638 (@value{GDBP}) set print symbol-filename on
4639 (@value{GDBP}) p/a ptt
4640 $4 = 0xe008 <t in hi2.c>
4644 @emph{Warning:} For pointers that point to a local variable, @samp{p/a}
4645 does not show the symbol name and filename of the referent, even with
4646 the appropriate @code{set print} options turned on.
4649 Other settings control how different kinds of objects are printed:
4652 @kindex set print array
4653 @item set print array
4654 @itemx set print array on
4655 Pretty print arrays. This format is more convenient to read,
4656 but uses more space. The default is off.
4658 @item set print array off
4659 Return to compressed format for arrays.
4661 @kindex show print array
4662 @item show print array
4663 Show whether compressed or pretty format is selected for displaying
4666 @kindex set print elements
4667 @item set print elements @var{number-of-elements}
4668 Set a limit on how many elements of an array @value{GDBN} will print.
4669 If @value{GDBN} is printing a large array, it stops printing after it has
4670 printed the number of elements set by the @code{set print elements} command.
4671 This limit also applies to the display of strings.
4672 Setting @var{number-of-elements} to zero means that the printing is unlimited.
4674 @kindex show print elements
4675 @item show print elements
4676 Display the number of elements of a large array that @value{GDBN} will print.
4677 If the number is 0, then the printing is unlimited.
4679 @kindex set print null-stop
4680 @item set print null-stop
4681 Cause @value{GDBN} to stop printing the characters of an array when the first
4682 @sc{NULL} is encountered. This is useful when large arrays actually
4683 contain only short strings.
4685 @kindex set print pretty
4686 @item set print pretty on
4687 Cause @value{GDBN} to print structures in an indented format with one member
4688 per line, like this:
4703 @item set print pretty off
4704 Cause @value{GDBN} to print structures in a compact format, like this:
4708 $1 = @{next = 0x0, flags = @{sweet = 1, sour = 1@}, \
4709 meat = 0x54 "Pork"@}
4714 This is the default format.
4716 @kindex show print pretty
4717 @item show print pretty
4718 Show which format @value{GDBN} is using to print structures.
4720 @kindex set print sevenbit-strings
4721 @item set print sevenbit-strings on
4722 Print using only seven-bit characters; if this option is set,
4723 @value{GDBN} displays any eight-bit characters (in strings or
4724 character values) using the notation @code{\}@var{nnn}. This setting is
4725 best if you are working in English (@sc{ascii}) and you use the
4726 high-order bit of characters as a marker or ``meta'' bit.
4728 @item set print sevenbit-strings off
4729 Print full eight-bit characters. This allows the use of more
4730 international character sets, and is the default.
4732 @kindex show print sevenbit-strings
4733 @item show print sevenbit-strings
4734 Show whether or not @value{GDBN} is printing only seven-bit characters.
4736 @kindex set print union
4737 @item set print union on
4738 Tell @value{GDBN} to print unions which are contained in structures. This
4739 is the default setting.
4741 @item set print union off
4742 Tell @value{GDBN} not to print unions which are contained in structures.
4744 @kindex show print union
4745 @item show print union
4746 Ask @value{GDBN} whether or not it will print unions which are contained in
4749 For example, given the declarations
4752 typedef enum @{Tree, Bug@} Species;
4753 typedef enum @{Big_tree, Acorn, Seedling@} Tree_forms;
4754 typedef enum @{Caterpillar, Cocoon, Butterfly@}
4765 struct thing foo = @{Tree, @{Acorn@}@};
4769 with @code{set print union on} in effect @samp{p foo} would print
4772 $1 = @{it = Tree, form = @{tree = Acorn, bug = Cocoon@}@}
4776 and with @code{set print union off} in effect it would print
4779 $1 = @{it = Tree, form = @{...@}@}
4785 These settings are of interest when debugging C++ programs:
4789 @kindex set print demangle
4790 @item set print demangle
4791 @itemx set print demangle on
4792 Print C++ names in their source form rather than in the encoded
4793 (``mangled'') form passed to the assembler and linker for type-safe
4794 linkage. The default is @samp{on}.
4796 @kindex show print demangle
4797 @item show print demangle
4798 Show whether C++ names are printed in mangled or demangled form.
4800 @kindex set print asm-demangle
4801 @item set print asm-demangle
4802 @itemx set print asm-demangle on
4803 Print C++ names in their source form rather than their mangled form, even
4804 in assembler code printouts such as instruction disassemblies.
4807 @kindex show print asm-demangle
4808 @item show print asm-demangle
4809 Show whether C++ names in assembly listings are printed in mangled
4812 @kindex set demangle-style
4813 @cindex C++ symbol decoding style
4814 @cindex symbol decoding style, C++
4815 @item set demangle-style @var{style}
4816 Choose among several encoding schemes used by different compilers to
4817 represent C++ names. The choices for @var{style} are currently:
4821 Allow @value{GDBN} to choose a decoding style by inspecting your program.
4824 Decode based on the @sc{gnu} C++ compiler (@code{g++}) encoding algorithm.
4825 This is the default.
4828 Decode based on the HP ANSI C++ (@code{aCC}) encoding algorithm.
4831 Decode based on the Lucid C++ compiler (@code{lcc}) encoding algorithm.
4834 Decode using the algorithm in the @cite{C++ Annotated Reference Manual}.
4835 @strong{Warning:} this setting alone is not sufficient to allow
4836 debugging @code{cfront}-generated executables. @value{GDBN} would
4837 require further enhancement to permit that.
4840 If you omit @var{style}, you will see a list of possible formats.
4842 @kindex show demangle-style
4843 @item show demangle-style
4844 Display the encoding style currently in use for decoding C++ symbols.
4846 @kindex set print object
4847 @item set print object
4848 @itemx set print object on
4849 When displaying a pointer to an object, identify the @emph{actual}
4850 (derived) type of the object rather than the @emph{declared} type, using
4851 the virtual function table.
4853 @item set print object off
4854 Display only the declared type of objects, without reference to the
4855 virtual function table. This is the default setting.
4857 @kindex show print object
4858 @item show print object
4859 Show whether actual, or declared, object types are displayed.
4861 @kindex set print static-members
4862 @item set print static-members
4863 @itemx set print static-members on
4864 Print static members when displaying a C++ object. The default is on.
4866 @item set print static-members off
4867 Do not print static members when displaying a C++ object.
4869 @kindex show print static-members
4870 @item show print static-members
4871 Show whether C++ static members are printed, or not.
4873 @c These don't work with HP ANSI C++ yet.
4874 @kindex set print vtbl
4875 @item set print vtbl
4876 @itemx set print vtbl on
4877 Pretty print C++ virtual function tables. The default is off.
4878 (The @code{vtbl} commands do not work on programs compiled with the HP
4879 ANSI C++ compiler (@code{aCC}).)
4881 @item set print vtbl off
4882 Do not pretty print C++ virtual function tables.
4884 @kindex show print vtbl
4885 @item show print vtbl
4886 Show whether C++ virtual function tables are pretty printed, or not.
4890 @section Value history
4892 @cindex value history
4893 Values printed by the @code{print} command are saved in the @value{GDBN}
4894 @dfn{value history}. This allows you to refer to them in other expressions.
4895 Values are kept until the symbol table is re-read or discarded
4896 (for example with the @code{file} or @code{symbol-file} commands).
4897 When the symbol table changes, the value history is discarded,
4898 since the values may contain pointers back to the types defined in the
4903 @cindex history number
4904 The values printed are given @dfn{history numbers} by which you can
4905 refer to them. These are successive integers starting with one.
4906 @code{print} shows you the history number assigned to a value by
4907 printing @samp{$@var{num} = } before the value; here @var{num} is the
4910 To refer to any previous value, use @samp{$} followed by the value's
4911 history number. The way @code{print} labels its output is designed to
4912 remind you of this. Just @code{$} refers to the most recent value in
4913 the history, and @code{$$} refers to the value before that.
4914 @code{$$@var{n}} refers to the @var{n}th value from the end; @code{$$2}
4915 is the value just prior to @code{$$}, @code{$$1} is equivalent to
4916 @code{$$}, and @code{$$0} is equivalent to @code{$}.
4918 For example, suppose you have just printed a pointer to a structure and
4919 want to see the contents of the structure. It suffices to type
4925 If you have a chain of structures where the component @code{next} points
4926 to the next one, you can print the contents of the next one with this:
4933 You can print successive links in the chain by repeating this
4934 command---which you can do by just typing @key{RET}.
4936 Note that the history records values, not expressions. If the value of
4937 @code{x} is 4 and you type these commands:
4945 then the value recorded in the value history by the @code{print} command
4946 remains 4 even though the value of @code{x} has changed.
4951 Print the last ten values in the value history, with their item numbers.
4952 This is like @samp{p@ $$9} repeated ten times, except that @code{show
4953 values} does not change the history.
4955 @item show values @var{n}
4956 Print ten history values centered on history item number @var{n}.
4959 Print ten history values just after the values last printed. If no more
4960 values are available, @code{show values +} produces no display.
4963 Pressing @key{RET} to repeat @code{show values @var{n}} has exactly the
4964 same effect as @samp{show values +}.
4966 @node Convenience Vars
4967 @section Convenience variables
4969 @cindex convenience variables
4970 @value{GDBN} provides @dfn{convenience variables} that you can use within
4971 @value{GDBN} to hold on to a value and refer to it later. These variables
4972 exist entirely within @value{GDBN}; they are not part of your program, and
4973 setting a convenience variable has no direct effect on further execution
4974 of your program. That is why you can use them freely.
4976 Convenience variables are prefixed with @samp{$}. Any name preceded by
4977 @samp{$} can be used for a convenience variable, unless it is one of
4978 the predefined machine-specific register names (@pxref{Registers}).
4979 (Value history references, in contrast, are @emph{numbers} preceded
4980 by @samp{$}. @xref{Value History, ,Value history}.)
4982 You can save a value in a convenience variable with an assignment
4983 expression, just as you would set a variable in your program.
4987 set $foo = *object_ptr
4991 would save in @code{$foo} the value contained in the object pointed to by
4994 Using a convenience variable for the first time creates it, but its
4995 value is @code{void} until you assign a new value. You can alter the
4996 value with another assignment at any time.
4998 Convenience variables have no fixed types. You can assign a convenience
4999 variable any type of value, including structures and arrays, even if
5000 that variable already has a value of a different type. The convenience
5001 variable, when used as an expression, has the type of its current value.
5004 @kindex show convenience
5005 @item show convenience
5006 Print a list of convenience variables used so far, and their values.
5007 Abbreviated @code{show con}.
5010 One of the ways to use a convenience variable is as a counter to be
5011 incremented or a pointer to be advanced. For example, to print
5012 a field from successive elements of an array of structures:
5016 print bar[$i++]->contents
5019 @noindent Repeat that command by typing @key{RET}.
5021 Some convenience variables are created automatically by @value{GDBN} and given
5022 values likely to be useful.
5027 The variable @code{$_} is automatically set by the @code{x} command to
5028 the last address examined (@pxref{Memory, ,Examining memory}). Other
5029 commands which provide a default address for @code{x} to examine also
5030 set @code{$_} to that address; these commands include @code{info line}
5031 and @code{info breakpoint}. The type of @code{$_} is @code{void *}
5032 except when set by the @code{x} command, in which case it is a pointer
5033 to the type of @code{$__}.
5037 The variable @code{$__} is automatically set by the @code{x} command
5038 to the value found in the last address examined. Its type is chosen
5039 to match the format in which the data was printed.
5043 The variable @code{$_exitcode} is automatically set to the exit code when
5044 the program being debugged terminates.
5047 On HP-UX systems, if you refer to a function or variable name that
5048 begins with a dollar sign, @value{GDBN} searches for a user or system
5049 name first, before it searches for a convenience variable.
5055 You can refer to machine register contents, in expressions, as variables
5056 with names starting with @samp{$}. The names of registers are different
5057 for each machine; use @code{info registers} to see the names used on
5061 @kindex info registers
5062 @item info registers
5063 Print the names and values of all registers except floating-point
5064 registers (in the selected stack frame).
5066 @kindex info all-registers
5067 @cindex floating point registers
5068 @item info all-registers
5069 Print the names and values of all registers, including floating-point
5072 @item info registers @var{regname} @dots{}
5073 Print the @dfn{relativized} value of each specified register @var{regname}.
5074 As discussed in detail below, register values are normally relative to
5075 the selected stack frame. @var{regname} may be any register name valid on
5076 the machine you are using, with or without the initial @samp{$}.
5079 @value{GDBN} has four ``standard'' register names that are available (in
5080 expressions) on most machines---whenever they do not conflict with an
5081 architecture's canonical mnemonics for registers. The register names
5082 @code{$pc} and @code{$sp} are used for the program counter register and
5083 the stack pointer. @code{$fp} is used for a register that contains a
5084 pointer to the current stack frame, and @code{$ps} is used for a
5085 register that contains the processor status. For example,
5086 you could print the program counter in hex with
5093 or print the instruction to be executed next with
5100 or add four to the stack pointer@footnote{This is a way of removing
5101 one word from the stack, on machines where stacks grow downward in
5102 memory (most machines, nowadays). This assumes that the innermost
5103 stack frame is selected; setting @code{$sp} is not allowed when other
5104 stack frames are selected. To pop entire frames off the stack,
5105 regardless of machine architecture, use @code{return};
5106 @pxref{Returning, ,Returning from a function}.} with
5112 Whenever possible, these four standard register names are available on
5113 your machine even though the machine has different canonical mnemonics,
5114 so long as there is no conflict. The @code{info registers} command
5115 shows the canonical names. For example, on the SPARC, @code{info
5116 registers} displays the processor status register as @code{$psr} but you
5117 can also refer to it as @code{$ps}.
5119 @value{GDBN} always considers the contents of an ordinary register as an
5120 integer when the register is examined in this way. Some machines have
5121 special registers which can hold nothing but floating point; these
5122 registers are considered to have floating point values. There is no way
5123 to refer to the contents of an ordinary register as floating point value
5124 (although you can @emph{print} it as a floating point value with
5125 @samp{print/f $@var{regname}}).
5127 Some registers have distinct ``raw'' and ``virtual'' data formats. This
5128 means that the data format in which the register contents are saved by
5129 the operating system is not the same one that your program normally
5130 sees. For example, the registers of the 68881 floating point
5131 coprocessor are always saved in ``extended'' (raw) format, but all C
5132 programs expect to work with ``double'' (virtual) format. In such
5133 cases, @value{GDBN} normally works with the virtual format only (the format
5134 that makes sense for your program), but the @code{info registers} command
5135 prints the data in both formats.
5137 Normally, register values are relative to the selected stack frame
5138 (@pxref{Selection, ,Selecting a frame}). This means that you get the
5139 value that the register would contain if all stack frames farther in
5140 were exited and their saved registers restored. In order to see the
5141 true contents of hardware registers, you must select the innermost
5142 frame (with @samp{frame 0}).
5144 However, @value{GDBN} must deduce where registers are saved, from the machine
5145 code generated by your compiler. If some registers are not saved, or if
5146 @value{GDBN} is unable to locate the saved registers, the selected stack
5147 frame makes no difference.
5149 @node Floating Point Hardware
5150 @section Floating point hardware
5151 @cindex floating point
5153 Depending on the configuration, @value{GDBN} may be able to give
5154 you more information about the status of the floating point hardware.
5159 Display hardware-dependent information about the floating
5160 point unit. The exact contents and layout vary depending on the
5161 floating point chip. Currently, @samp{info float} is supported on
5162 the ARM and x86 machines.
5166 @chapter Using @value{GDBN} with Different Languages
5169 Although programming languages generally have common aspects, they are
5170 rarely expressed in the same manner. For instance, in ANSI C,
5171 dereferencing a pointer @code{p} is accomplished by @code{*p}, but in
5172 Modula-2, it is accomplished by @code{p^}. Values can also be
5173 represented (and displayed) differently. Hex numbers in C appear as
5174 @samp{0x1ae}, while in Modula-2 they appear as @samp{1AEH}.
5176 @cindex working language
5177 Language-specific information is built into @value{GDBN} for some languages,
5178 allowing you to express operations like the above in your program's
5179 native language, and allowing @value{GDBN} to output values in a manner
5180 consistent with the syntax of your program's native language. The
5181 language you use to build expressions is called the @dfn{working
5185 * Setting:: Switching between source languages
5186 * Show:: Displaying the language
5187 * Checks:: Type and range checks
5188 * Support:: Supported languages
5192 @section Switching between source languages
5194 There are two ways to control the working language---either have @value{GDBN}
5195 set it automatically, or select it manually yourself. You can use the
5196 @code{set language} command for either purpose. On startup, @value{GDBN}
5197 defaults to setting the language automatically. The working language is
5198 used to determine how expressions you type are interpreted, how values
5201 In addition to the working language, every source file that
5202 @value{GDBN} knows about has its own working language. For some object
5203 file formats, the compiler might indicate which language a particular
5204 source file is in. However, most of the time @value{GDBN} infers the
5205 language from the name of the file. The language of a source file
5206 controls whether C++ names are demangled---this way @code{backtrace} can
5207 show each frame appropriately for its own language. There is no way to
5208 set the language of a source file from within @value{GDBN}.
5210 This is most commonly a problem when you use a program, such
5211 as @code{cfront} or @code{f2c}, that generates C but is written in
5212 another language. In that case, make the
5213 program use @code{#line} directives in its C output; that way
5214 @value{GDBN} will know the correct language of the source code of the original
5215 program, and will display that source code, not the generated C code.
5218 * Filenames:: Filename extensions and languages.
5219 * Manually:: Setting the working language manually
5220 * Automatically:: Having @value{GDBN} infer the source language
5224 @subsection List of filename extensions and languages
5226 If a source file name ends in one of the following extensions, then
5227 @value{GDBN} infers that its language is the one indicated.
5252 Modula-2 source file
5256 Assembler source file. This actually behaves almost like C, but
5257 @value{GDBN} does not skip over function prologues when stepping.
5260 In addition, you may set the language associated with a filename
5261 extension. @xref{Show, , Displaying the language}.
5264 @subsection Setting the working language
5266 If you allow @value{GDBN} to set the language automatically,
5267 expressions are interpreted the same way in your debugging session and
5270 @kindex set language
5271 If you wish, you may set the language manually. To do this, issue the
5272 command @samp{set language @var{lang}}, where @var{lang} is the name of
5274 @code{c} or @code{modula-2}.
5275 For a list of the supported languages, type @samp{set language}.
5277 Setting the language manually prevents @value{GDBN} from updating the working
5278 language automatically. This can lead to confusion if you try
5279 to debug a program when the working language is not the same as the
5280 source language, when an expression is acceptable to both
5281 languages---but means different things. For instance, if the current
5282 source file were written in C, and @value{GDBN} was parsing Modula-2, a
5290 might not have the effect you intended. In C, this means to add
5291 @code{b} and @code{c} and place the result in @code{a}. The result
5292 printed would be the value of @code{a}. In Modula-2, this means to compare
5293 @code{a} to the result of @code{b+c}, yielding a @code{BOOLEAN} value.
5296 @subsection Having @value{GDBN} infer the source language
5298 To have @value{GDBN} set the working language automatically, use
5299 @samp{set language local} or @samp{set language auto}. @value{GDBN}
5300 then infers the working language. That is, when your program stops in a
5301 frame (usually by encountering a breakpoint), @value{GDBN} sets the
5302 working language to the language recorded for the function in that
5303 frame. If the language for a frame is unknown (that is, if the function
5304 or block corresponding to the frame was defined in a source file that
5305 does not have a recognized extension), the current working language is
5306 not changed, and @value{GDBN} issues a warning.
5308 This may not seem necessary for most programs, which are written
5309 entirely in one source language. However, program modules and libraries
5310 written in one source language can be used by a main program written in
5311 a different source language. Using @samp{set language auto} in this
5312 case frees you from having to set the working language manually.
5315 @section Displaying the language
5317 The following commands help you find out which language is the
5318 working language, and also what language source files were written in.
5320 @kindex show language
5325 Display the current working language. This is the
5326 language you can use with commands such as @code{print} to
5327 build and compute expressions that may involve variables in your program.
5330 Display the source language for this frame. This language becomes the
5331 working language if you use an identifier from this frame.
5332 @xref{Frame Info, ,Information about a frame}, to identify the other
5333 information listed here.
5336 Display the source language of this source file.
5337 @xref{Symbols, ,Examining the Symbol Table}, to identify the other
5338 information listed here.
5341 In unusual circumstances, you may have source files with extensions
5342 not in the standard list. You can then set the extension associated
5343 with a language explicitly:
5345 @kindex set extension-language
5346 @kindex info extensions
5348 @item set extension-language @var{.ext} @var{language}
5349 Set source files with extension @var{.ext} to be assumed to be in
5350 the source language @var{language}.
5352 @item info extensions
5353 List all the filename extensions and the associated languages.
5357 @section Type and range checking
5360 @emph{Warning:} In this release, the @value{GDBN} commands for type and range
5361 checking are included, but they do not yet have any effect. This
5362 section documents the intended facilities.
5364 @c FIXME remove warning when type/range code added
5366 Some languages are designed to guard you against making seemingly common
5367 errors through a series of compile- and run-time checks. These include
5368 checking the type of arguments to functions and operators, and making
5369 sure mathematical overflows are caught at run time. Checks such as
5370 these help to ensure a program's correctness once it has been compiled
5371 by eliminating type mismatches, and providing active checks for range
5372 errors when your program is running.
5374 @value{GDBN} can check for conditions like the above if you wish.
5375 Although @value{GDBN} does not check the statements in your program, it
5376 can check expressions entered directly into @value{GDBN} for evaluation via
5377 the @code{print} command, for example. As with the working language,
5378 @value{GDBN} can also decide whether or not to check automatically based on
5379 your program's source language. @xref{Support, ,Supported languages},
5380 for the default settings of supported languages.
5383 * Type Checking:: An overview of type checking
5384 * Range Checking:: An overview of range checking
5387 @cindex type checking
5388 @cindex checks, type
5390 @subsection An overview of type checking
5392 Some languages, such as Modula-2, are strongly typed, meaning that the
5393 arguments to operators and functions have to be of the correct type,
5394 otherwise an error occurs. These checks prevent type mismatch
5395 errors from ever causing any run-time problems. For example,
5403 The second example fails because the @code{CARDINAL} 1 is not
5404 type-compatible with the @code{REAL} 2.3.
5406 For the expressions you use in @value{GDBN} commands, you can tell the
5407 @value{GDBN} type checker to skip checking;
5408 to treat any mismatches as errors and abandon the expression;
5409 or to only issue warnings when type mismatches occur,
5410 but evaluate the expression anyway. When you choose the last of
5411 these, @value{GDBN} evaluates expressions like the second example above, but
5412 also issues a warning.
5414 Even if you turn type checking off, there may be other reasons
5415 related to type that prevent @value{GDBN} from evaluating an expression.
5416 For instance, @value{GDBN} does not know how to add an @code{int} and
5417 a @code{struct foo}. These particular type errors have nothing to do
5418 with the language in use, and usually arise from expressions, such as
5419 the one described above, which make little sense to evaluate anyway.
5421 Each language defines to what degree it is strict about type. For
5422 instance, both Modula-2 and C require the arguments to arithmetical
5423 operators to be numbers. In C, enumerated types and pointers can be
5424 represented as numbers, so that they are valid arguments to mathematical
5425 operators. @xref{Support, ,Supported languages}, for further
5426 details on specific languages.
5428 @value{GDBN} provides some additional commands for controlling the type checker:
5431 @kindex set check type
5432 @kindex show check type
5434 @item set check type auto
5435 Set type checking on or off based on the current working language.
5436 @xref{Support, ,Supported languages}, for the default settings for
5439 @item set check type on
5440 @itemx set check type off
5441 Set type checking on or off, overriding the default setting for the
5442 current working language. Issue a warning if the setting does not
5443 match the language default. If any type mismatches occur in
5444 evaluating an expression while typechecking is on, @value{GDBN} prints a
5445 message and aborts evaluation of the expression.
5447 @item set check type warn
5448 Cause the type checker to issue warnings, but to always attempt to
5449 evaluate the expression. Evaluating the expression may still
5450 be impossible for other reasons. For example, @value{GDBN} cannot add
5451 numbers and structures.
5454 Show the current setting of the type checker, and whether or not @value{GDBN}
5455 is setting it automatically.
5458 @cindex range checking
5459 @cindex checks, range
5460 @node Range Checking
5461 @subsection An overview of range checking
5463 In some languages (such as Modula-2), it is an error to exceed the
5464 bounds of a type; this is enforced with run-time checks. Such range
5465 checking is meant to ensure program correctness by making sure
5466 computations do not overflow, or indices on an array element access do
5467 not exceed the bounds of the array.
5469 For expressions you use in @value{GDBN} commands, you can tell
5470 @value{GDBN} to treat range errors in one of three ways: ignore them,
5471 always treat them as errors and abandon the expression, or issue
5472 warnings but evaluate the expression anyway.
5474 A range error can result from numerical overflow, from exceeding an
5475 array index bound, or when you type a constant that is not a member
5476 of any type. Some languages, however, do not treat overflows as an
5477 error. In many implementations of C, mathematical overflow causes the
5478 result to ``wrap around'' to lower values---for example, if @var{m} is
5479 the largest integer value, and @var{s} is the smallest, then
5482 @var{m} + 1 @result{} @var{s}
5485 This, too, is specific to individual languages, and in some cases
5486 specific to individual compilers or machines. @xref{Support, ,
5487 Supported languages}, for further details on specific languages.
5489 @value{GDBN} provides some additional commands for controlling the range checker:
5492 @kindex set check range
5493 @kindex show check range
5495 @item set check range auto
5496 Set range checking on or off based on the current working language.
5497 @xref{Support, ,Supported languages}, for the default settings for
5500 @item set check range on
5501 @itemx set check range off
5502 Set range checking on or off, overriding the default setting for the
5503 current working language. A warning is issued if the setting does not
5504 match the language default. If a range error occurs, then a message
5505 is printed and evaluation of the expression is aborted.
5507 @item set check range warn
5508 Output messages when the @value{GDBN} range checker detects a range error,
5509 but attempt to evaluate the expression anyway. Evaluating the
5510 expression may still be impossible for other reasons, such as accessing
5511 memory that the process does not own (a typical example from many Unix
5515 Show the current setting of the range checker, and whether or not it is
5516 being set automatically by @value{GDBN}.
5520 @section Supported languages
5522 @value{GDBN} supports C, C++, Fortran, Java, Chill, assembly, and Modula-2.
5523 @c This is false ...
5524 Some @value{GDBN} features may be used in expressions regardless of the
5525 language you use: the @value{GDBN} @code{@@} and @code{::} operators,
5526 and the @samp{@{type@}addr} construct (@pxref{Expressions,
5527 ,Expressions}) can be used with the constructs of any supported
5530 The following sections detail to what degree each source language is
5531 supported by @value{GDBN}. These sections are not meant to be language
5532 tutorials or references, but serve only as a reference guide to what the
5533 @value{GDBN} expression parser accepts, and what input and output
5534 formats should look like for different languages. There are many good
5535 books written on each of these languages; please look to these for a
5536 language reference or tutorial.
5540 * Modula-2:: Modula-2
5545 @subsection C and C++
5548 @cindex expressions in C or C++
5550 Since C and C++ are so closely related, many features of @value{GDBN} apply
5551 to both languages. Whenever this is the case, we discuss those languages
5556 @cindex @sc{gnu} C++
5557 The C++ debugging facilities are jointly implemented by the C++
5558 compiler and @value{GDBN}. Therefore, to debug your C++ code
5559 effectively, you must compile your C++ programs with a supported
5560 C++ compiler, such as @sc{gnu} @code{g++}, or the HP ANSI C++
5561 compiler (@code{aCC}).
5563 For best results when using @sc{gnu} C++, use the stabs debugging
5564 format. You can select that format explicitly with the @code{g++}
5565 command-line options @samp{-gstabs} or @samp{-gstabs+}. See
5566 @ref{Debugging Options,,Options for Debugging Your Program or @sc{gnu}
5567 CC, gcc.info, Using @sc{gnu} CC}, for more information.
5570 * C Operators:: C and C++ operators
5571 * C Constants:: C and C++ constants
5572 * C plus plus expressions:: C++ expressions
5573 * C Defaults:: Default settings for C and C++
5574 * C Checks:: C and C++ type and range checks
5575 * Debugging C:: @value{GDBN} and C
5576 * Debugging C plus plus:: @value{GDBN} features for C++
5580 @subsubsection C and C++ operators
5582 @cindex C and C++ operators
5584 Operators must be defined on values of specific types. For instance,
5585 @code{+} is defined on numbers, but not on structures. Operators are
5586 often defined on groups of types.
5588 For the purposes of C and C++, the following definitions hold:
5593 @emph{Integral types} include @code{int} with any of its storage-class
5594 specifiers; @code{char}; @code{enum}; and, for C++, @code{bool}.
5597 @emph{Floating-point types} include @code{float} and @code{double}.
5600 @emph{Pointer types} include all types defined as @code{(@var{type} *)}.
5603 @emph{Scalar types} include all of the above.
5608 The following operators are supported. They are listed here
5609 in order of increasing precedence:
5613 The comma or sequencing operator. Expressions in a comma-separated list
5614 are evaluated from left to right, with the result of the entire
5615 expression being the last expression evaluated.
5618 Assignment. The value of an assignment expression is the value
5619 assigned. Defined on scalar types.
5622 Used in an expression of the form @w{@code{@var{a} @var{op}= @var{b}}},
5623 and translated to @w{@code{@var{a} = @var{a op b}}}.
5624 @w{@code{@var{op}=}} and @code{=} have the same precendence.
5625 @var{op} is any one of the operators @code{|}, @code{^}, @code{&},
5626 @code{<<}, @code{>>}, @code{+}, @code{-}, @code{*}, @code{/}, @code{%}.
5629 The ternary operator. @code{@var{a} ? @var{b} : @var{c}} can be thought
5630 of as: if @var{a} then @var{b} else @var{c}. @var{a} should be of an
5634 Logical @sc{or}. Defined on integral types.
5637 Logical @sc{and}. Defined on integral types.
5640 Bitwise @sc{or}. Defined on integral types.
5643 Bitwise exclusive-@sc{or}. Defined on integral types.
5646 Bitwise @sc{and}. Defined on integral types.
5649 Equality and inequality. Defined on scalar types. The value of these
5650 expressions is 0 for false and non-zero for true.
5652 @item <@r{, }>@r{, }<=@r{, }>=
5653 Less than, greater than, less than or equal, greater than or equal.
5654 Defined on scalar types. The value of these expressions is 0 for false
5655 and non-zero for true.
5658 left shift, and right shift. Defined on integral types.
5661 The @value{GDBN} ``artificial array'' operator (@pxref{Expressions, ,Expressions}).
5664 Addition and subtraction. Defined on integral types, floating-point types and
5667 @item *@r{, }/@r{, }%
5668 Multiplication, division, and modulus. Multiplication and division are
5669 defined on integral and floating-point types. Modulus is defined on
5673 Increment and decrement. When appearing before a variable, the
5674 operation is performed before the variable is used in an expression;
5675 when appearing after it, the variable's value is used before the
5676 operation takes place.
5679 Pointer dereferencing. Defined on pointer types. Same precedence as
5683 Address operator. Defined on variables. Same precedence as @code{++}.
5685 For debugging C++, @value{GDBN} implements a use of @samp{&} beyond what is
5686 allowed in the C++ language itself: you can use @samp{&(&@var{ref})}
5687 (or, if you prefer, simply @samp{&&@var{ref}}) to examine the address
5688 where a C++ reference variable (declared with @samp{&@var{ref}}) is
5692 Negative. Defined on integral and floating-point types. Same
5693 precedence as @code{++}.
5696 Logical negation. Defined on integral types. Same precedence as
5700 Bitwise complement operator. Defined on integral types. Same precedence as
5705 Structure member, and pointer-to-structure member. For convenience,
5706 @value{GDBN} regards the two as equivalent, choosing whether to dereference a
5707 pointer based on the stored type information.
5708 Defined on @code{struct} and @code{union} data.
5711 Dereferences of pointers to members.
5714 Array indexing. @code{@var{a}[@var{i}]} is defined as
5715 @code{*(@var{a}+@var{i})}. Same precedence as @code{->}.
5718 Function parameter list. Same precedence as @code{->}.
5721 C++ scope resolution operator. Defined on @code{struct}, @code{union},
5722 and @code{class} types.
5725 Doubled colons also represent the @value{GDBN} scope operator
5726 (@pxref{Expressions, ,Expressions}). Same precedence as @code{::},
5730 If an operator is redefined in the user code, @value{GDBN} usually
5731 attempts to invoke the redefined version instead of using the operator's
5739 @subsubsection C and C++ constants
5741 @cindex C and C++ constants
5743 @value{GDBN} allows you to express the constants of C and C++ in the
5748 Integer constants are a sequence of digits. Octal constants are
5749 specified by a leading @samp{0} (i.e. zero), and hexadecimal constants by
5750 a leading @samp{0x} or @samp{0X}. Constants may also end with a letter
5751 @samp{l}, specifying that the constant should be treated as a
5755 Floating point constants are a sequence of digits, followed by a decimal
5756 point, followed by a sequence of digits, and optionally followed by an
5757 exponent. An exponent is of the form:
5758 @samp{@w{e@r{[[}+@r{]|}-@r{]}@var{nnn}}}, where @var{nnn} is another
5759 sequence of digits. The @samp{+} is optional for positive exponents.
5762 Enumerated constants consist of enumerated identifiers, or their
5763 integral equivalents.
5766 Character constants are a single character surrounded by single quotes
5767 (@code{'}), or a number---the ordinal value of the corresponding character
5768 (usually its @sc{ASCII} value). Within quotes, the single character may
5769 be represented by a letter or by @dfn{escape sequences}, which are of
5770 the form @samp{\@var{nnn}}, where @var{nnn} is the octal representation
5771 of the character's ordinal value; or of the form @samp{\@var{x}}, where
5772 @samp{@var{x}} is a predefined special character---for example,
5773 @samp{\n} for newline.
5776 String constants are a sequence of character constants surrounded
5777 by double quotes (@code{"}).
5780 Pointer constants are an integral value. You can also write pointers
5781 to constants using the C operator @samp{&}.
5784 Array constants are comma-separated lists surrounded by braces @samp{@{}
5785 and @samp{@}}; for example, @samp{@{1,2,3@}} is a three-element array of
5786 integers, @samp{@{@{1,2@}, @{3,4@}, @{5,6@}@}} is a three-by-two array,
5787 and @samp{@{&"hi", &"there", &"fred"@}} is a three-element array of pointers.
5791 * C plus plus expressions::
5798 @node C plus plus expressions
5799 @subsubsection C++ expressions
5801 @cindex expressions in C++
5802 @value{GDBN} expression handling can interpret most C++ expressions.
5804 @cindex C++ support, not in @sc{coff}
5805 @cindex @sc{coff} versus C++
5806 @cindex C++ and object formats
5807 @cindex object formats and C++
5808 @cindex a.out and C++
5809 @cindex @sc{ecoff} and C++
5810 @cindex @sc{xcoff} and C++
5811 @cindex @sc{elf}/stabs and C++
5812 @cindex @sc{elf}/@sc{dwarf} and C++
5813 @c FIXME!! GDB may eventually be able to debug C++ using DWARF; check
5814 @c periodically whether this has happened...
5816 @emph{Warning:} @value{GDBN} can only debug C++ code if you use the
5817 proper compiler. Typically, C++ debugging depends on the use of
5818 additional debugging information in the symbol table, and thus requires
5819 special support. In particular, if your compiler generates a.out, MIPS
5820 @sc{ecoff}, RS/6000 @sc{xcoff}, or @sc{elf} with stabs extensions to the
5821 symbol table, these facilities are all available. (With @sc{gnu} CC,
5822 you can use the @samp{-gstabs} option to request stabs debugging
5823 extensions explicitly.) Where the object code format is standard
5824 @sc{coff} or @sc{dwarf} in @sc{elf}, on the other hand, most of the C++
5825 support in @value{GDBN} does @emph{not} work.
5830 @cindex member functions
5832 Member function calls are allowed; you can use expressions like
5835 count = aml->GetOriginal(x, y)
5839 @cindex namespace in C++
5841 While a member function is active (in the selected stack frame), your
5842 expressions have the same namespace available as the member function;
5843 that is, @value{GDBN} allows implicit references to the class instance
5844 pointer @code{this} following the same rules as C++.
5846 @cindex call overloaded functions
5847 @cindex overloaded functions
5848 @cindex type conversions in C++
5850 You can call overloaded functions; @value{GDBN} resolves the function
5851 call to the right definition, with some restrictions. GDB does not
5852 perform overload resolution involving user-defined type conversions,
5853 calls to constructors, or instantiations of templates that do not exist
5854 in the program. It also cannot handle ellipsis argument lists or
5857 It does perform integral conversions and promotions, floating-point
5858 promotions, arithmetic conversions, pointer conversions, conversions of
5859 class objects to base classes, and standard conversions such as those of
5860 functions or arrays to pointers; it requires an exact match on the
5861 number of function arguments.
5863 Overload resolution is always performed, unless you have specified
5864 @code{set overload-resolution off}. @xref{Debugging C plus plus,
5865 ,@value{GDBN} features for C++}.
5867 You must specify@code{set overload-resolution off} in order to use an
5868 explicit function signature to call an overloaded function, as in
5870 p 'foo(char,int)'('x', 13)
5872 The @value{GDBN} command-completion facility can simplify this;
5873 @pxref{Completion, ,Command completion}.
5875 @cindex reference declarations
5877 @value{GDBN} understands variables declared as C++ references; you can use
5878 them in expressions just as you do in C++ source---they are automatically
5881 In the parameter list shown when @value{GDBN} displays a frame, the values of
5882 reference variables are not displayed (unlike other variables); this
5883 avoids clutter, since references are often used for large structures.
5884 The @emph{address} of a reference variable is always shown, unless
5885 you have specified @samp{set print address off}.
5888 @value{GDBN} supports the C++ name resolution operator @code{::}---your
5889 expressions can use it just as expressions in your program do. Since
5890 one scope may be defined in another, you can use @code{::} repeatedly if
5891 necessary, for example in an expression like
5892 @samp{@var{scope1}::@var{scope2}::@var{name}}. @value{GDBN} also allows
5893 resolving name scope by reference to source files, in both C and C++
5894 debugging (@pxref{Variables, ,Program variables}).
5897 In addition, when used with HP's C++ compiler, @value{GDBN} supports
5898 calling virtual functions correctly, printing out virtual bases of
5899 objects, calling functions in a base subobject, casting objects, and
5900 invoking user-defined operators.
5903 @subsubsection C and C++ defaults
5905 @cindex C and C++ defaults
5907 If you allow @value{GDBN} to set type and range checking automatically, they
5908 both default to @code{off} whenever the working language changes to
5909 C or C++. This happens regardless of whether you or @value{GDBN}
5910 selects the working language.
5912 If you allow @value{GDBN} to set the language automatically, it
5913 recognizes source files whose names end with @file{.c}, @file{.C}, or
5914 @file{.cc}, etc, and when @value{GDBN} enters code compiled from one of
5915 these files, it sets the working language to C or C++.
5916 @xref{Automatically, ,Having @value{GDBN} infer the source language},
5917 for further details.
5919 @c Type checking is (a) primarily motivated by Modula-2, and (b)
5920 @c unimplemented. If (b) changes, it might make sense to let this node
5921 @c appear even if Mod-2 does not, but meanwhile ignore it. roland 16jul93.
5924 @subsubsection C and C++ type and range checks
5926 @cindex C and C++ checks
5928 By default, when @value{GDBN} parses C or C++ expressions, type checking
5929 is not used. However, if you turn type checking on, @value{GDBN}
5930 considers two variables type equivalent if:
5934 The two variables are structured and have the same structure, union, or
5938 The two variables have the same type name, or types that have been
5939 declared equivalent through @code{typedef}.
5942 @c leaving this out because neither J Gilmore nor R Pesch understand it.
5945 The two @code{struct}, @code{union}, or @code{enum} variables are
5946 declared in the same declaration. (Note: this may not be true for all C
5951 Range checking, if turned on, is done on mathematical operations. Array
5952 indices are not checked, since they are often used to index a pointer
5953 that is not itself an array.
5956 @subsubsection @value{GDBN} and C
5958 The @code{set print union} and @code{show print union} commands apply to
5959 the @code{union} type. When set to @samp{on}, any @code{union} that is
5960 inside a @code{struct} or @code{class} is also printed. Otherwise, it
5961 appears as @samp{@{...@}}.
5963 The @code{@@} operator aids in the debugging of dynamic arrays, formed
5964 with pointers and a memory allocation function. @xref{Expressions,
5968 * Debugging C plus plus::
5971 @node Debugging C plus plus
5972 @subsubsection @value{GDBN} features for C++
5974 @cindex commands for C++
5976 Some @value{GDBN} commands are particularly useful with C++, and some are
5977 designed specifically for use with C++. Here is a summary:
5980 @cindex break in overloaded functions
5981 @item @r{breakpoint menus}
5982 When you want a breakpoint in a function whose name is overloaded,
5983 @value{GDBN} breakpoint menus help you specify which function definition
5984 you want. @xref{Breakpoint Menus,,Breakpoint menus}.
5986 @cindex overloading in C++
5987 @item rbreak @var{regex}
5988 Setting breakpoints using regular expressions is helpful for setting
5989 breakpoints on overloaded functions that are not members of any special
5991 @xref{Set Breaks, ,Setting breakpoints}.
5993 @cindex C++ exception handling
5996 Debug C++ exception handling using these commands. @xref{Set
5997 Catchpoints, , Setting catchpoints}.
6000 @item ptype @var{typename}
6001 Print inheritance relationships as well as other information for type
6003 @xref{Symbols, ,Examining the Symbol Table}.
6005 @cindex C++ symbol display
6006 @item set print demangle
6007 @itemx show print demangle
6008 @itemx set print asm-demangle
6009 @itemx show print asm-demangle
6010 Control whether C++ symbols display in their source form, both when
6011 displaying code as C++ source and when displaying disassemblies.
6012 @xref{Print Settings, ,Print settings}.
6014 @item set print object
6015 @itemx show print object
6016 Choose whether to print derived (actual) or declared types of objects.
6017 @xref{Print Settings, ,Print settings}.
6019 @item set print vtbl
6020 @itemx show print vtbl
6021 Control the format for printing virtual function tables.
6022 @xref{Print Settings, ,Print settings}.
6023 (The @code{vtbl} commands do not work on programs compiled with the HP
6024 ANSI C++ compiler (@code{aCC}).)
6026 @kindex set overload-resolution
6027 @cindex overloaded functions
6028 @item set overload-resolution on
6029 Enable overload resolution for C++ expression evaluation. The default
6030 is on. For overloaded functions, @value{GDBN} evaluates the arguments
6031 and searches for a function whose signature matches the argument types,
6032 using the standard C++ conversion rules (@pxref{C plus plus expressions, ,C++
6033 expressions} for details). If it cannot find a match, it emits a
6036 @item set overload-resolution off
6037 Disable overload resolution for C++ expression evaluation. For
6038 overloaded functions that are not class member functions, @value{GDBN}
6039 chooses the first function of the specified name that it finds in the
6040 symbol table, whether or not its arguments are of the correct type. For
6041 overloaded functions that are class member functions, @value{GDBN}
6042 searches for a function whose signature @emph{exactly} matches the
6045 @item @r{Overloaded symbol names}
6046 You can specify a particular definition of an overloaded symbol, using
6047 the same notation that is used to declare such symbols in C++: type
6048 @code{@var{symbol}(@var{types})} rather than just @var{symbol}. You can
6049 also use the @value{GDBN} command-line word completion facilities to list the
6050 available choices, or to finish the type list for you.
6051 @xref{Completion,, Command completion}, for details on how to do this.
6055 @subsection Modula-2
6059 The extensions made to @value{GDBN} to support Modula-2 only support
6060 output from the @sc{gnu} Modula-2 compiler (which is currently being
6061 developed). Other Modula-2 compilers are not currently supported, and
6062 attempting to debug executables produced by them is most likely
6063 to give an error as @value{GDBN} reads in the executable's symbol
6066 @cindex expressions in Modula-2
6068 * M2 Operators:: Built-in operators
6069 * Built-In Func/Proc:: Built-in functions and procedures
6070 * M2 Constants:: Modula-2 constants
6071 * M2 Defaults:: Default settings for Modula-2
6072 * Deviations:: Deviations from standard Modula-2
6073 * M2 Checks:: Modula-2 type and range checks
6074 * M2 Scope:: The scope operators @code{::} and @code{.}
6075 * GDB/M2:: @value{GDBN} and Modula-2
6079 @subsubsection Operators
6080 @cindex Modula-2 operators
6082 Operators must be defined on values of specific types. For instance,
6083 @code{+} is defined on numbers, but not on structures. Operators are
6084 often defined on groups of types. For the purposes of Modula-2, the
6085 following definitions hold:
6090 @emph{Integral types} consist of @code{INTEGER}, @code{CARDINAL}, and
6094 @emph{Character types} consist of @code{CHAR} and its subranges.
6097 @emph{Floating-point types} consist of @code{REAL}.
6100 @emph{Pointer types} consist of anything declared as @code{POINTER TO
6104 @emph{Scalar types} consist of all of the above.
6107 @emph{Set types} consist of @code{SET} and @code{BITSET} types.
6110 @emph{Boolean types} consist of @code{BOOLEAN}.
6114 The following operators are supported, and appear in order of
6115 increasing precedence:
6119 Function argument or array index separator.
6122 Assignment. The value of @var{var} @code{:=} @var{value} is
6126 Less than, greater than on integral, floating-point, or enumerated
6130 Less than, greater than, less than or equal to, greater than or equal to
6131 on integral, floating-point and enumerated types, or set inclusion on
6132 set types. Same precedence as @code{<}.
6134 @item =@r{, }<>@r{, }#
6135 Equality and two ways of expressing inequality, valid on scalar types.
6136 Same precedence as @code{<}. In @value{GDBN} scripts, only @code{<>} is
6137 available for inequality, since @code{#} conflicts with the script
6141 Set membership. Defined on set types and the types of their members.
6142 Same precedence as @code{<}.
6145 Boolean disjunction. Defined on boolean types.
6148 Boolean conjuction. Defined on boolean types.
6151 The @value{GDBN} ``artificial array'' operator (@pxref{Expressions, ,Expressions}).
6154 Addition and subtraction on integral and floating-point types, or union
6155 and difference on set types.
6158 Multiplication on integral and floating-point types, or set intersection
6162 Division on floating-point types, or symmetric set difference on set
6163 types. Same precedence as @code{*}.
6166 Integer division and remainder. Defined on integral types. Same
6167 precedence as @code{*}.
6170 Negative. Defined on @code{INTEGER} and @code{REAL} data.
6173 Pointer dereferencing. Defined on pointer types.
6176 Boolean negation. Defined on boolean types. Same precedence as
6180 @code{RECORD} field selector. Defined on @code{RECORD} data. Same
6181 precedence as @code{^}.
6184 Array indexing. Defined on @code{ARRAY} data. Same precedence as @code{^}.
6187 Procedure argument list. Defined on @code{PROCEDURE} objects. Same precedence
6191 @value{GDBN} and Modula-2 scope operators.
6195 @emph{Warning:} Sets and their operations are not yet supported, so @value{GDBN}
6196 treats the use of the operator @code{IN}, or the use of operators
6197 @code{+}, @code{-}, @code{*}, @code{/}, @code{=}, , @code{<>}, @code{#},
6198 @code{<=}, and @code{>=} on sets as an error.
6201 @cindex Modula-2 built-ins
6202 @node Built-In Func/Proc
6203 @subsubsection Built-in functions and procedures
6205 Modula-2 also makes available several built-in procedures and functions.
6206 In describing these, the following metavariables are used:
6211 represents an @code{ARRAY} variable.
6214 represents a @code{CHAR} constant or variable.
6217 represents a variable or constant of integral type.
6220 represents an identifier that belongs to a set. Generally used in the
6221 same function with the metavariable @var{s}. The type of @var{s} should
6222 be @code{SET OF @var{mtype}} (where @var{mtype} is the type of @var{m}).
6225 represents a variable or constant of integral or floating-point type.
6228 represents a variable or constant of floating-point type.
6234 represents a variable.
6237 represents a variable or constant of one of many types. See the
6238 explanation of the function for details.
6241 All Modula-2 built-in procedures also return a result, described below.
6245 Returns the absolute value of @var{n}.
6248 If @var{c} is a lower case letter, it returns its upper case
6249 equivalent, otherwise it returns its argument
6252 Returns the character whose ordinal value is @var{i}.
6255 Decrements the value in the variable @var{v}. Returns the new value.
6257 @item DEC(@var{v},@var{i})
6258 Decrements the value in the variable @var{v} by @var{i}. Returns the
6261 @item EXCL(@var{m},@var{s})
6262 Removes the element @var{m} from the set @var{s}. Returns the new
6265 @item FLOAT(@var{i})
6266 Returns the floating point equivalent of the integer @var{i}.
6269 Returns the index of the last member of @var{a}.
6272 Increments the value in the variable @var{v}. Returns the new value.
6274 @item INC(@var{v},@var{i})
6275 Increments the value in the variable @var{v} by @var{i}. Returns the
6278 @item INCL(@var{m},@var{s})
6279 Adds the element @var{m} to the set @var{s} if it is not already
6280 there. Returns the new set.
6283 Returns the maximum value of the type @var{t}.
6286 Returns the minimum value of the type @var{t}.
6289 Returns boolean TRUE if @var{i} is an odd number.
6292 Returns the ordinal value of its argument. For example, the ordinal
6293 value of a character is its ASCII value (on machines supporting the
6294 ASCII character set). @var{x} must be of an ordered type, which include
6295 integral, character and enumerated types.
6298 Returns the size of its argument. @var{x} can be a variable or a type.
6300 @item TRUNC(@var{r})
6301 Returns the integral part of @var{r}.
6303 @item VAL(@var{t},@var{i})
6304 Returns the member of the type @var{t} whose ordinal value is @var{i}.
6308 @emph{Warning:} Sets and their operations are not yet supported, so
6309 @value{GDBN} treats the use of procedures @code{INCL} and @code{EXCL} as
6313 @cindex Modula-2 constants
6315 @subsubsection Constants
6317 @value{GDBN} allows you to express the constants of Modula-2 in the following
6323 Integer constants are simply a sequence of digits. When used in an
6324 expression, a constant is interpreted to be type-compatible with the
6325 rest of the expression. Hexadecimal integers are specified by a
6326 trailing @samp{H}, and octal integers by a trailing @samp{B}.
6329 Floating point constants appear as a sequence of digits, followed by a
6330 decimal point and another sequence of digits. An optional exponent can
6331 then be specified, in the form @samp{E@r{[}+@r{|}-@r{]}@var{nnn}}, where
6332 @samp{@r{[}+@r{|}-@r{]}@var{nnn}} is the desired exponent. All of the
6333 digits of the floating point constant must be valid decimal (base 10)
6337 Character constants consist of a single character enclosed by a pair of
6338 like quotes, either single (@code{'}) or double (@code{"}). They may
6339 also be expressed by their ordinal value (their ASCII value, usually)
6340 followed by a @samp{C}.
6343 String constants consist of a sequence of characters enclosed by a
6344 pair of like quotes, either single (@code{'}) or double (@code{"}).
6345 Escape sequences in the style of C are also allowed. @xref{C
6346 Constants, ,C and C++ constants}, for a brief explanation of escape
6350 Enumerated constants consist of an enumerated identifier.
6353 Boolean constants consist of the identifiers @code{TRUE} and
6357 Pointer constants consist of integral values only.
6360 Set constants are not yet supported.
6364 @subsubsection Modula-2 defaults
6365 @cindex Modula-2 defaults
6367 If type and range checking are set automatically by @value{GDBN}, they
6368 both default to @code{on} whenever the working language changes to
6369 Modula-2. This happens regardless of whether you, or @value{GDBN},
6370 selected the working language.
6372 If you allow @value{GDBN} to set the language automatically, then entering
6373 code compiled from a file whose name ends with @file{.mod} sets the
6374 working language to Modula-2. @xref{Automatically, ,Having @value{GDBN} set
6375 the language automatically}, for further details.
6378 @subsubsection Deviations from standard Modula-2
6379 @cindex Modula-2, deviations from
6381 A few changes have been made to make Modula-2 programs easier to debug.
6382 This is done primarily via loosening its type strictness:
6386 Unlike in standard Modula-2, pointer constants can be formed by
6387 integers. This allows you to modify pointer variables during
6388 debugging. (In standard Modula-2, the actual address contained in a
6389 pointer variable is hidden from you; it can only be modified
6390 through direct assignment to another pointer variable or expression that
6391 returned a pointer.)
6394 C escape sequences can be used in strings and characters to represent
6395 non-printable characters. @value{GDBN} prints out strings with these
6396 escape sequences embedded. Single non-printable characters are
6397 printed using the @samp{CHR(@var{nnn})} format.
6400 The assignment operator (@code{:=}) returns the value of its right-hand
6404 All built-in procedures both modify @emph{and} return their argument.
6408 @subsubsection Modula-2 type and range checks
6409 @cindex Modula-2 checks
6412 @emph{Warning:} in this release, @value{GDBN} does not yet perform type or
6415 @c FIXME remove warning when type/range checks added
6417 @value{GDBN} considers two Modula-2 variables type equivalent if:
6421 They are of types that have been declared equivalent via a @code{TYPE
6422 @var{t1} = @var{t2}} statement
6425 They have been declared on the same line. (Note: This is true of the
6426 @sc{gnu} Modula-2 compiler, but it may not be true of other compilers.)
6429 As long as type checking is enabled, any attempt to combine variables
6430 whose types are not equivalent is an error.
6432 Range checking is done on all mathematical operations, assignment, array
6433 index bounds, and all built-in functions and procedures.
6436 @subsubsection The scope operators @code{::} and @code{.}
6439 @cindex colon, doubled as scope operator
6442 @c Info cannot handle :: but TeX can.
6448 There are a few subtle differences between the Modula-2 scope operator
6449 (@code{.}) and the @value{GDBN} scope operator (@code{::}). The two have
6454 @var{module} . @var{id}
6455 @var{scope} :: @var{id}
6459 where @var{scope} is the name of a module or a procedure,
6460 @var{module} the name of a module, and @var{id} is any declared
6461 identifier within your program, except another module.
6463 Using the @code{::} operator makes @value{GDBN} search the scope
6464 specified by @var{scope} for the identifier @var{id}. If it is not
6465 found in the specified scope, then @value{GDBN} searches all scopes
6466 enclosing the one specified by @var{scope}.
6468 Using the @code{.} operator makes @value{GDBN} search the current scope for
6469 the identifier specified by @var{id} that was imported from the
6470 definition module specified by @var{module}. With this operator, it is
6471 an error if the identifier @var{id} was not imported from definition
6472 module @var{module}, or if @var{id} is not an identifier in
6476 @subsubsection @value{GDBN} and Modula-2
6478 Some @value{GDBN} commands have little use when debugging Modula-2 programs.
6479 Five subcommands of @code{set print} and @code{show print} apply
6480 specifically to C and C++: @samp{vtbl}, @samp{demangle},
6481 @samp{asm-demangle}, @samp{object}, and @samp{union}. The first four
6482 apply to C++, and the last to the C @code{union} type, which has no direct
6483 analogue in Modula-2.
6485 The @code{@@} operator (@pxref{Expressions, ,Expressions}), while available
6486 while using any language, is not useful with Modula-2. Its
6487 intent is to aid the debugging of @dfn{dynamic arrays}, which cannot be
6488 created in Modula-2 as they can in C or C++. However, because an
6489 address can be specified by an integral constant, the construct
6490 @samp{@{@var{type}@}@var{adrexp}} is still useful. (@pxref{Expressions, ,Expressions})
6492 @cindex @code{#} in Modula-2
6493 In @value{GDBN} scripts, the Modula-2 inequality operator @code{#} is
6494 interpreted as the beginning of a comment. Use @code{<>} instead.
6499 The extensions made to @value{GDBN} to support Chill only support output
6500 from the GNU Chill compiler. Other Chill compilers are not currently
6501 supported, and attempting to debug executables produced by them is most
6502 likely to give an error as @value{GDBN} reads in the executable's symbol
6505 This section covers the following Chill related topics and the features
6506 of @value{GDBN} which support these topics.
6509 * How modes are displayed:: How modes are displayed
6510 * Locations:: Locations and their accesses
6511 * Values and their Operations:: Values and their Operations
6512 * Chill type and range checks::
6516 @node How modes are displayed
6517 @subsubsection How modes are displayed
6519 The Chill Datatype- (Mode) support of @value{GDBN} is directly related
6520 with the functionality of the GNU Chill compiler, and therefore deviates
6521 slightly from the standard specification of the Chill language. The
6524 @item @r{@emph{Discrete modes:}}
6527 @emph{Integer Modes} which are predefined by @code{BYTE, UBYTE, INT,
6530 @emph{Boolean Mode} which is predefined by @code{BOOL},
6532 @emph{Character Mode} which is predefined by @code{CHAR},
6534 @emph{Set Mode} which is displayed by the keyword @code{SET}.
6536 (@value{GDBP}) ptype x
6537 type = SET (karli = 10, susi = 20, fritzi = 100)
6539 If the type is an unnumbered set the set element values are omitted.
6541 @emph{Range Mode} which is displayed by @code{type = <basemode>
6542 (<lower bound> : <upper bound>)}, where @code{<lower bound>, <upper
6543 bound>} can be of any discrete literal expression (e.g. set element
6547 @item @r{@emph{Powerset Mode:}}
6548 A Powerset Mode is displayed by the keyword @code{POWERSET} followed by
6549 the member mode of the powerset. The member mode can be any discrete mode.
6551 (@value{GDBP}) ptype x
6552 type = POWERSET SET (egon, hugo, otto)
6555 @item @r{@emph{Reference Modes:}}
6558 @emph{Bound Reference Mode} which is diplayed by the keyword @code{REF}
6559 followed by the mode name to which the reference is bound.
6561 @emph{Free Reference Mode} which is displayed by the keyword @code{PTR}.
6564 @item @r{@emph{Procedure mode}}
6565 The procedure mode is displayed by @code{type = PROC(<parameter list>)
6566 <return mode> EXCEPTIONS (<exception list>)}. The @code{<parameter
6567 list>} is a list of the parameter modes. @code{<return mode>} indicates
6568 the mode of the result of the procedure if any. The exceptionlist lists
6569 all possible exceptions which can be raised by the procedure.
6572 @item @r{@emph{Instance mode}}
6573 The instance mode is represented by a structure, which has a static
6574 type, and is therefore not really of interest.
6577 @item @r{@emph{Synchronization Modes:}}
6580 @emph{Event Mode} which is displayed by @code{EVENT (<event length>)},
6581 where @code{(<event length>)} is optional.
6583 @emph{Buffer Mode} which is displayed by @code{BUFFER (<buffer length>)
6584 <buffer element mode>}, where @code{(<buffer length>)} is optional.
6587 @item @r{@emph{Timing Modes:}}
6590 @emph{Duration Mode} which is predefined by @code{DURATION}
6592 @emph{Absolute Time Mode} which is predefined by @code{TIME}
6595 @item @r{@emph{Real Modes:}}
6596 Real Modes are predefined with @code{REAL} and @code{LONG_REAL}.
6598 @item @r{@emph{String Modes:}}
6601 @emph{Character String Mode} which is displayed by @code{CHARS(<string
6602 length>)}, followed by the keyword @code{VARYING} if the String Mode is
6605 @emph{Bit String Mode} which is displayed by @code{BOOLS(<string
6609 @item @r{@emph{Array Mode:}}
6610 The Array Mode is displayed by the keyword @code{ARRAY(<range>)}
6611 followed by the element mode (which may in turn be an array mode).
6613 (@value{GDBP}) ptype x
6616 SET (karli = 10, susi = 20, fritzi = 100)
6619 @item @r{@emph{Structure Mode}}
6620 The Structure mode is displayed by the keyword @code{STRUCT(<field
6621 list>)}. The @code{<field list>} consists of names and modes of fields
6622 of the structure. Variant structures have the keyword @code{CASE <field>
6623 OF <variant fields> ESAC} in their field list. Since the current version
6624 of the GNU Chill compiler doesn't implement tag processing (no runtime
6625 checks of variant fields, and therefore no debugging info), the output
6626 always displays all variant fields.
6628 (@value{GDBP}) ptype str
6643 @subsubsection Locations and their accesses
6645 A location in Chill is an object which can contain values.
6647 A value of a location is generally accessed by the (declared) name of
6648 the location. The output conforms to the specification of values in
6649 Chill programs. How values are specified
6650 is the topic of the next section.
6652 The pseudo-location @code{RESULT} (or @code{result}) can be used to
6653 display or change the result of a currently-active procedure:
6657 - does the same as the Chill action @code{RESULT EXPR} (which
6658 is not available in gdb).
6660 Values of reference mode locations are printed by @code{PTR(<hex
6661 value>)} in case of a free reference mode, and by @code{(REF <reference
6662 mode>) (<hex-value>)} in case of a bound reference. @code{<hex value>}
6663 represents the address where the reference points to. To access the
6664 value of the location referenced by the pointer, use the dereference
6665 operator `@code{->}'.
6667 Values of procedure mode locations are displayed by @code{@{ PROC
6668 (<argument modes> ) <return mode> @} <address> <name of procedure
6669 location>}. @code{<argument modes>} is a list of modes according to the
6670 parameter specification of the procedure and @code{<address>} shows the
6671 address of the entry point.
6674 Locations of instance modes are displayed just like a structure with two
6675 fields specifying the @emph{process type} and the @emph{copy number} of
6676 the investigated instance location@footnote{This comes from the current
6677 implementation of instances. They are implemented as a structure (no
6678 na). The output should be something like @code{[<name of the process>;
6679 <instance number>]}.}. The field names are @code{__proc_type} and
6682 Locations of synchronization modes are displayed like a structure with
6683 the field name @code{__event_data} in case of a event mode location, and
6684 like a structure with the field @code{__buffer_data} in case of a buffer
6685 mode location (refer to previous paragraph).
6687 Structure Mode locations are printed by @code{[.<field name>: <value>,
6688 ...]}. The @code{<field name>} corresponds to the structure mode
6689 definition and the layout of @code{<value>} varies depending of the mode
6690 of the field. If the investigated structure mode location is of variant
6691 structure mode the variant parts of the structure are enclosed in curled
6692 braces (`@code{@{@}}'). Fields enclosed by `@code{@{,@}}' are residing
6693 on the same memory location and represent the current values of the
6694 memory location in their specific modes. Since no tag processing is done
6695 all variants are displayed. A variant field is printed by
6696 @code{(<variant name>) = .<field name>: <value>}. (who implements the
6699 (@value{GDBP}) print str1 $4 = [.as: 0, .bs: karli, .<TAG>: { (karli) =
6700 [.cs: []], (susi) = [.ds: susi]}]
6704 Substructures of string mode-, array mode- or structure mode-values
6705 (e.g. array slices, fields of structure locations) are accessed using
6706 certain operations which are descibed in the next chapter.
6708 A location value may be interpreted as having a different mode using the
6709 location conversion. This mode conversion is written as @code{<mode
6710 name>(<location>)}. The user has to consider that the sizes of the modes
6711 have to be equal otherwise an error message occurs. Further no range
6712 checking of the location against the destination mode is performed and
6713 therefore the result can be quite confusing.
6715 (@value{GDBP}) print int (s(3 up 4)) XXX TO be filled in !! XXX
6718 @node Values and their Operations
6719 @subsubsection Values and their Operations
6721 Values are used to alter locations, to investigate complex structures in
6722 more detail or to filter relevant information out of a large amount of
6723 data. There are several (mode dependent) operations defined which enable
6724 such investigations. These operations are not only applicable to
6725 constant values but also to locations, which can become quite useful
6726 when debugging complex structures. During parsing the command line
6727 (e.g. evaluating an expression) @value{GDBN} treats location names as
6728 the values behind these locations.
6730 This subchapters describes how values have to be specified and which
6731 operations are legal to be used with such values.
6734 @item Literal Values
6735 Literal values are specified in the same manner as in GNU Chill programs.
6736 For detailed specification refer to the GNU Chill implementation Manual
6742 @emph{Integer Literals} are specified in the same manner as in Chill
6743 programs (refer z200/88 chpt 5.2.4.2)
6745 @emph{Boolean Literals} are defined by @code{TRUE} and @code{FALSE}.
6747 @emph{Character Literals} are defined by @code{'<character>'}. (e.g.
6750 @emph{Set Literals} are defined by a name which was specified in a set
6751 mode. The value delivered by a Set Literal is the set value. This is
6752 comparable to an enumaration in C/C++ language.
6754 @emph{Emptiness Literal} is predefined by @code{NULL}. The value of the
6755 emptiness literal delivers either the empty reference value, the empty
6756 procedure value or the empty instance value.
6759 @emph{Character String Literals} are defined by a sequence of characters
6760 enclosed in single- or double quotes. If a single- or double quote has
6761 to be part of the string literal it has to be stuffed (specified twice).
6763 @emph{Bitstring Literals} are specified in the same manner as in Chill
6764 programs (refer z200/88 chpt 5.2.4.8).
6766 @emph{Floating point literals} are specified in the same manner as in
6767 (gnu-)Chill programs (refer GNU Chill implementation Manual chapter 1.5).
6772 A tuple is specified by @code{<mode name>[<tuple>]}, where @code{<mode
6773 name>} can be omitted if the mode of the tuple is unambigous. This
6774 unambiguity is derived from the context of a evaluated expression.
6775 @code{<tuple>} can be one of the following:
6777 @item @emph{Powerset Tuple}
6778 @item @emph{Array Tuple}
6779 @item @emph{Structure Tuple}
6780 Powerset tuples, array tuples and structure tuples are specified in the
6781 same manner as in Chill programs refer z200/88 chpt 5.2.5.
6784 @item String Element Value
6785 A string element value is specified by @code{<string value>(<index>)},
6786 where @code{<index>} is a integer expression. It delivers a character
6787 value which is equivalent to the character indexed by @code{<index>} in
6790 @item String Slice Value
6791 A string slice value is specified by @code{<string value>(<slice
6792 spec>)}, where @code{<slice spec>} can be either a range of integer
6793 expressions or specified by @code{<start expr> up <size>}.
6794 @code{<size>} denotes the number of elements which the slice contains.
6795 The delivered value is a string value, which is part of the specified
6798 @item Array Element Values
6799 An array element value is specified by @code{<array value>(<expr>)} and
6800 delivers a array element value of the mode of the specified array.
6802 @item Array Slice Values
6803 An array slice is specified by @code{<array value>(<slice spec>)}, where
6804 @code{<slice spec>} can be either a range specified by expressions or by
6805 @code{<start expr> up <size>}. @code{<size>} denotes the number of
6806 arrayelements the slice contains. The delivered value is an array value
6807 which is part of the specified array.
6809 @item Structure Field Values
6810 A structure field value is derived by @code{<structure value>.<field
6811 name>}, where @code{<field name>} indcates the name of a field specified
6812 in the mode definition of the structure. The mode of the delivered value
6813 corresponds to this mode definition in the structure definition.
6815 @item Procedure Call Value
6816 The procedure call value is derived from the return value of the
6817 procedure@footnote{If a procedure call is used for instance in an
6818 expression, then this procedure is called with all its side
6819 effects. This can lead to confusing results if used carelessly.}.
6821 Values of duration mode locations are represented by ULONG literals.
6823 Values of time mode locations are represented by TIME(<secs>:<nsecs>).
6826 This is not implemented yet:
6827 @item Built-in Value
6829 The following built in functions are provided:
6840 @item @code{UPPER()}
6841 @item @code{LOWER()}
6842 @item @code{LENGTH()}
6846 @item @code{ARCSIN()}
6847 @item @code{ARCCOS()}
6848 @item @code{ARCTAN()}
6855 For a detailed description refer to the GNU Chill implementation manual
6859 @item Zero-adic Operator Value
6860 The zero-adic operator value is derived from the instance value for the
6861 current active process.
6863 @item Expression Values
6864 The value delivered by an expression is the result of the evaluation of
6865 the specified expression. If there are error conditions (mode
6866 incompatibility, etc.) the evaluation of expressions is aborted with a
6867 corresponding error message. Expressions may be paranthesised which
6868 causes the evaluation of this expression before any other expression
6869 which uses the result of the paranthesised expression. The following
6870 operators are supported by @value{GDBN}:
6872 @item @code{OR, ORIF, XOR}
6873 @item @code{AND, ANDIF}
6875 Logical operators defined over operands of boolean mode.
6877 Equality and inequality operators defined over all modes.
6880 Relational operators defined over predefined modes.
6882 @item @code{*, /, MOD, REM}
6883 Arithmetic operators defined over predefined modes.
6885 Change sign operator.
6887 String concatenation operator.
6889 String repetition operator.
6891 Referenced location operator which can be used either to take the
6892 address of a location (@code{->loc}), or to dereference a reference
6893 location (@code{loc->}).
6894 @item @code{OR, XOR}
6897 Powerset and bitstring operators.
6900 Powerset inclusion operators.
6902 Membership operator.
6906 @node Chill type and range checks
6907 @subsubsection Chill type and range checks
6909 @value{GDBN} considers two Chill variables mode equivalent if the sizes
6910 of the two modes are equal. This rule applies recursively to more
6911 complex datatypes which means that complex modes are treated
6912 eqivalent if all element modes (which also can be complex modes like
6913 structures, arrays, etc.) have the same size.
6915 Range checking is done on all mathematical operations, assignment, array
6916 index bounds and all built in procedures.
6918 Strong type checks are forced using the @value{GDBN} command @code{set
6919 check strong}. This enforces strong type and range checks on all
6920 operations where Chill constructs are used (expressions, built in
6921 functions, etc.) in respect to the semantics as defined in the z.200
6922 language specification.
6925 All checks can be disabled by the @value{GDBN} command @code{set check
6929 @c Deviations from the Chill Standard Z200/88
6930 see last paragraph ?
6933 @node Chill defaults
6934 @subsubsection Chill defaults
6936 If type and range checking are set automatically by @value{GDBN}, they
6937 both default to @code{on} whenever the working language changes to
6938 Chill. This happens regardless of whether you, or @value{GDBN},
6939 selected the working language.
6941 If you allow @value{GDBN} to set the language automatically, then entering
6942 code compiled from a file whose name ends with @file{.ch} sets the
6943 working language to Chill. @xref{Automatically, ,Having @value{GDBN} set
6944 the language automatically}, for further details.
6947 @chapter Examining the Symbol Table
6949 The commands described in this section allow you to inquire about the
6950 symbols (names of variables, functions and types) defined in your
6951 program. This information is inherent in the text of your program and
6952 does not change as your program executes. @value{GDBN} finds it in your
6953 program's symbol table, in the file indicated when you started @value{GDBN}
6954 (@pxref{File Options, ,Choosing files}), or by one of the
6955 file-management commands (@pxref{Files, ,Commands to specify files}).
6957 @cindex symbol names
6958 @cindex names of symbols
6959 @cindex quoting names
6960 Occasionally, you may need to refer to symbols that contain unusual
6961 characters, which @value{GDBN} ordinarily treats as word delimiters. The
6962 most frequent case is in referring to static variables in other
6963 source files (@pxref{Variables,,Program variables}). File names
6964 are recorded in object files as debugging symbols, but @value{GDBN} would
6965 ordinarily parse a typical file name, like @file{foo.c}, as the three words
6966 @samp{foo} @samp{.} @samp{c}. To allow @value{GDBN} to recognize
6967 @samp{foo.c} as a single symbol, enclose it in single quotes; for example,
6974 looks up the value of @code{x} in the scope of the file @file{foo.c}.
6977 @kindex info address
6978 @item info address @var{symbol}
6979 Describe where the data for @var{symbol} is stored. For a register
6980 variable, this says which register it is kept in. For a non-register
6981 local variable, this prints the stack-frame offset at which the variable
6984 Note the contrast with @samp{print &@var{symbol}}, which does not work
6985 at all for a register variable, and for a stack local variable prints
6986 the exact address of the current instantiation of the variable.
6989 @item whatis @var{exp}
6990 Print the data type of expression @var{exp}. @var{exp} is not
6991 actually evaluated, and any side-effecting operations (such as
6992 assignments or function calls) inside it do not take place.
6993 @xref{Expressions, ,Expressions}.
6996 Print the data type of @code{$}, the last value in the value history.
6999 @item ptype @var{typename}
7000 Print a description of data type @var{typename}. @var{typename} may be
7001 the name of a type, or for C code it may have the form @samp{class
7002 @var{class-name}}, @samp{struct @var{struct-tag}}, @samp{union
7003 @var{union-tag}} or @samp{enum @var{enum-tag}}.
7005 @item ptype @var{exp}
7007 Print a description of the type of expression @var{exp}. @code{ptype}
7008 differs from @code{whatis} by printing a detailed description, instead
7009 of just the name of the type.
7011 For example, for this variable declaration:
7014 struct complex @{double real; double imag;@} v;
7018 the two commands give this output:
7022 (@value{GDBP}) whatis v
7023 type = struct complex
7024 (@value{GDBP}) ptype v
7025 type = struct complex @{
7033 As with @code{whatis}, using @code{ptype} without an argument refers to
7034 the type of @code{$}, the last value in the value history.
7037 @item info types @var{regexp}
7039 Print a brief description of all types whose name matches @var{regexp}
7040 (or all types in your program, if you supply no argument). Each
7041 complete typename is matched as though it were a complete line; thus,
7042 @samp{i type value} gives information on all types in your program whose
7043 name includes the string @code{value}, but @samp{i type ^value$} gives
7044 information only on types whose complete name is @code{value}.
7046 This command differs from @code{ptype} in two ways: first, like
7047 @code{whatis}, it does not print a detailed description; second, it
7048 lists all source files where a type is defined.
7052 Show the name of the current source file---that is, the source file for
7053 the function containing the current point of execution---and the language
7056 @kindex info sources
7058 Print the names of all source files in your program for which there is
7059 debugging information, organized into two lists: files whose symbols
7060 have already been read, and files whose symbols will be read when needed.
7062 @kindex info functions
7063 @item info functions
7064 Print the names and data types of all defined functions.
7066 @item info functions @var{regexp}
7067 Print the names and data types of all defined functions
7068 whose names contain a match for regular expression @var{regexp}.
7069 Thus, @samp{info fun step} finds all functions whose names
7070 include @code{step}; @samp{info fun ^step} finds those whose names
7071 start with @code{step}.
7073 @kindex info variables
7074 @item info variables
7075 Print the names and data types of all variables that are declared
7076 outside of functions (i.e., excluding local variables).
7078 @item info variables @var{regexp}
7079 Print the names and data types of all variables (except for local
7080 variables) whose names contain a match for regular expression
7084 This was never implemented.
7085 @kindex info methods
7087 @itemx info methods @var{regexp}
7088 The @code{info methods} command permits the user to examine all defined
7089 methods within C++ program, or (with the @var{regexp} argument) a
7090 specific set of methods found in the various C++ classes. Many
7091 C++ classes provide a large number of methods. Thus, the output
7092 from the @code{ptype} command can be overwhelming and hard to use. The
7093 @code{info-methods} command filters the methods, printing only those
7094 which match the regular-expression @var{regexp}.
7097 @cindex reloading symbols
7098 Some systems allow individual object files that make up your program to
7099 be replaced without stopping and restarting your program. For example,
7100 in VxWorks you can simply recompile a defective object file and keep on
7101 running. If you are running on one of these systems, you can allow
7102 @value{GDBN} to reload the symbols for automatically relinked modules:
7105 @kindex set symbol-reloading
7106 @item set symbol-reloading on
7107 Replace symbol definitions for the corresponding source file when an
7108 object file with a particular name is seen again.
7110 @item set symbol-reloading off
7111 Do not replace symbol definitions when re-encountering object files of
7112 the same name. This is the default state; if you are not running on a
7113 system that permits automatically relinking modules, you should leave
7114 @code{symbol-reloading} off, since otherwise @value{GDBN} may discard symbols
7115 when linking large programs, that may contain several modules (from
7116 different directories or libraries) with the same name.
7118 @kindex show symbol-reloading
7119 @item show symbol-reloading
7120 Show the current @code{on} or @code{off} setting.
7123 @kindex set opaque-type-resolution
7124 @item set opaque-type-resolution on
7125 Tell @value{GDBN} to resolve opaque types. An opaque type is a type
7126 declared as a pointer to a @code{struct}, @code{class}, or
7127 @code{union}---for example, @code{struct MyType *}---that is used in one
7128 source file although the full declaration of @code{struct MyType} is in
7129 another source file. The default is on.
7131 A change in the setting of this subcommand will not take effect until
7132 the next time symbols for a file are loaded.
7134 @item set opaque-type-resolution off
7135 Tell @value{GDBN} not to resolve opaque types. In this case, the type
7136 is printed as follows:
7138 @{<no data fields>@}
7141 @kindex show opaque-type-resolution
7142 @item show opaque-type-resolution
7143 Show whether opaque types are resolved or not.
7145 @kindex maint print symbols
7147 @kindex maint print psymbols
7148 @cindex partial symbol dump
7149 @item maint print symbols @var{filename}
7150 @itemx maint print psymbols @var{filename}
7151 @itemx maint print msymbols @var{filename}
7152 Write a dump of debugging symbol data into the file @var{filename}.
7153 These commands are used to debug the @value{GDBN} symbol-reading code. Only
7154 symbols with debugging data are included. If you use @samp{maint print
7155 symbols}, @value{GDBN} includes all the symbols for which it has already
7156 collected full details: that is, @var{filename} reflects symbols for
7157 only those files whose symbols @value{GDBN} has read. You can use the
7158 command @code{info sources} to find out which files these are. If you
7159 use @samp{maint print psymbols} instead, the dump shows information about
7160 symbols that @value{GDBN} only knows partially---that is, symbols defined in
7161 files that @value{GDBN} has skimmed, but not yet read completely. Finally,
7162 @samp{maint print msymbols} dumps just the minimal symbol information
7163 required for each object file from which @value{GDBN} has read some symbols.
7164 @xref{Files, ,Commands to specify files}, for a discussion of how
7165 @value{GDBN} reads symbols (in the description of @code{symbol-file}).
7169 @chapter Altering Execution
7171 Once you think you have found an error in your program, you might want to
7172 find out for certain whether correcting the apparent error would lead to
7173 correct results in the rest of the run. You can find the answer by
7174 experiment, using the @value{GDBN} features for altering execution of the
7177 For example, you can store new values into variables or memory
7178 locations, give your program a signal, restart it at a different
7179 address, or even return prematurely from a function.
7182 * Assignment:: Assignment to variables
7183 * Jumping:: Continuing at a different address
7184 * Signaling:: Giving your program a signal
7185 * Returning:: Returning from a function
7186 * Calling:: Calling your program's functions
7187 * Patching:: Patching your program
7191 @section Assignment to variables
7194 @cindex setting variables
7195 To alter the value of a variable, evaluate an assignment expression.
7196 @xref{Expressions, ,Expressions}. For example,
7203 stores the value 4 into the variable @code{x}, and then prints the
7204 value of the assignment expression (which is 4).
7205 @xref{Languages, ,Using @value{GDBN} with Different Languages}, for more
7206 information on operators in supported languages.
7208 @kindex set variable
7209 @cindex variables, setting
7210 If you are not interested in seeing the value of the assignment, use the
7211 @code{set} command instead of the @code{print} command. @code{set} is
7212 really the same as @code{print} except that the expression's value is
7213 not printed and is not put in the value history (@pxref{Value History,
7214 ,Value history}). The expression is evaluated only for its effects.
7216 If the beginning of the argument string of the @code{set} command
7217 appears identical to a @code{set} subcommand, use the @code{set
7218 variable} command instead of just @code{set}. This command is identical
7219 to @code{set} except for its lack of subcommands. For example, if your
7220 program has a variable @code{width}, you get an error if you try to set
7221 a new value with just @samp{set width=13}, because @value{GDBN} has the
7222 command @code{set width}:
7225 (@value{GDBP}) whatis width
7227 (@value{GDBP}) p width
7229 (@value{GDBP}) set width=47
7230 Invalid syntax in expression.
7234 The invalid expression, of course, is @samp{=47}. In
7235 order to actually set the program's variable @code{width}, use
7238 (@value{GDBP}) set var width=47
7241 Because the @code{set} command has many subcommands that can conflict
7242 with the names of program variables, it is a good idea to use the
7243 @code{set variable} command instead of just @code{set}. For example, if
7244 your program has a variable @code{g}, you run into problems if you try
7245 to set a new value with just @samp{set g=4}, because @value{GDBN} has
7246 the command @code{set gnutarget}, abbreviated @code{set g}:
7250 (@value{GDBP}) whatis g
7254 (@value{GDBP}) set g=4
7258 The program being debugged has been started already.
7259 Start it from the beginning? (y or n) y
7260 Starting program: /home/smith/cc_progs/a.out
7261 "/home/smith/cc_progs/a.out": can't open to read symbols: Invalid bfd target.
7262 (@value{GDBP}) show g
7263 The current BFD target is "=4".
7268 The program variable @code{g} did not change, and you silently set the
7269 @code{gnutarget} to an invalid value. In order to set the variable
7273 (@value{GDBP}) set var g=4
7276 @value{GDBN} allows more implicit conversions in assignments than C; you can
7277 freely store an integer value into a pointer variable or vice versa,
7278 and you can convert any structure to any other structure that is the
7279 same length or shorter.
7280 @comment FIXME: how do structs align/pad in these conversions?
7281 @comment /doc@cygnus.com 18dec1990
7283 To store values into arbitrary places in memory, use the @samp{@{@dots{}@}}
7284 construct to generate a value of specified type at a specified address
7285 (@pxref{Expressions, ,Expressions}). For example, @code{@{int@}0x83040} refers
7286 to memory location @code{0x83040} as an integer (which implies a certain size
7287 and representation in memory), and
7290 set @{int@}0x83040 = 4
7294 stores the value 4 into that memory location.
7297 @section Continuing at a different address
7299 Ordinarily, when you continue your program, you do so at the place where
7300 it stopped, with the @code{continue} command. You can instead continue at
7301 an address of your own choosing, with the following commands:
7305 @item jump @var{linespec}
7306 Resume execution at line @var{linespec}. Execution stops again
7307 immediately if there is a breakpoint there. @xref{List, ,Printing
7308 source lines}, for a description of the different forms of
7309 @var{linespec}. It is common practice to use the @code{tbreak} command
7310 in conjunction with @code{jump}. @xref{Set Breaks, ,Setting
7313 The @code{jump} command does not change the current stack frame, or
7314 the stack pointer, or the contents of any memory location or any
7315 register other than the program counter. If line @var{linespec} is in
7316 a different function from the one currently executing, the results may
7317 be bizarre if the two functions expect different patterns of arguments or
7318 of local variables. For this reason, the @code{jump} command requests
7319 confirmation if the specified line is not in the function currently
7320 executing. However, even bizarre results are predictable if you are
7321 well acquainted with the machine-language code of your program.
7323 @item jump *@var{address}
7324 Resume execution at the instruction at address @var{address}.
7327 @c Doesn't work on HP-UX; have to set $pcoqh and $pcoqt.
7328 On many systems, you can get much the same effect as the @code{jump}
7329 command by storing a new value into the register @code{$pc}. The
7330 difference is that this does not start your program running; it only
7331 changes the address of where it @emph{will} run when you continue. For
7339 makes the next @code{continue} command or stepping command execute at
7340 address @code{0x485}, rather than at the address where your program stopped.
7341 @xref{Continuing and Stepping, ,Continuing and stepping}.
7343 The most common occasion to use the @code{jump} command is to back
7344 up---perhaps with more breakpoints set---over a portion of a program
7345 that has already executed, in order to examine its execution in more
7350 @section Giving your program a signal
7354 @item signal @var{signal}
7355 Resume execution where your program stopped, but immediately give it the
7356 signal @var{signal}. @var{signal} can be the name or the number of a
7357 signal. For example, on many systems @code{signal 2} and @code{signal
7358 SIGINT} are both ways of sending an interrupt signal.
7360 Alternatively, if @var{signal} is zero, continue execution without
7361 giving a signal. This is useful when your program stopped on account of
7362 a signal and would ordinary see the signal when resumed with the
7363 @code{continue} command; @samp{signal 0} causes it to resume without a
7366 @code{signal} does not repeat when you press @key{RET} a second time
7367 after executing the command.
7371 Invoking the @code{signal} command is not the same as invoking the
7372 @code{kill} utility from the shell. Sending a signal with @code{kill}
7373 causes @value{GDBN} to decide what to do with the signal depending on
7374 the signal handling tables (@pxref{Signals}). The @code{signal} command
7375 passes the signal directly to your program.
7379 @section Returning from a function
7382 @cindex returning from a function
7385 @itemx return @var{expression}
7386 You can cancel execution of a function call with the @code{return}
7387 command. If you give an
7388 @var{expression} argument, its value is used as the function's return
7392 When you use @code{return}, @value{GDBN} discards the selected stack frame
7393 (and all frames within it). You can think of this as making the
7394 discarded frame return prematurely. If you wish to specify a value to
7395 be returned, give that value as the argument to @code{return}.
7397 This pops the selected stack frame (@pxref{Selection, ,Selecting a
7398 frame}), and any other frames inside of it, leaving its caller as the
7399 innermost remaining frame. That frame becomes selected. The
7400 specified value is stored in the registers used for returning values
7403 The @code{return} command does not resume execution; it leaves the
7404 program stopped in the state that would exist if the function had just
7405 returned. In contrast, the @code{finish} command (@pxref{Continuing
7406 and Stepping, ,Continuing and stepping}) resumes execution until the
7407 selected stack frame returns naturally.
7410 @section Calling program functions
7412 @cindex calling functions
7415 @item call @var{expr}
7416 Evaluate the expression @var{expr} without displaying @code{void}
7420 You can use this variant of the @code{print} command if you want to
7421 execute a function from your program, but without cluttering the output
7422 with @code{void} returned values. If the result is not void, it
7423 is printed and saved in the value history.
7425 For the A29K, a user-controlled variable @code{call_scratch_address},
7426 specifies the location of a scratch area to be used when @value{GDBN}
7427 calls a function in the target. This is necessary because the usual
7428 method of putting the scratch area on the stack does not work in systems
7429 that have separate instruction and data spaces.
7432 @section Patching programs
7434 @cindex patching binaries
7435 @cindex writing into executables
7436 @cindex writing into corefiles
7438 By default, @value{GDBN} opens the file containing your program's
7439 executable code (or the corefile) read-only. This prevents accidental
7440 alterations to machine code; but it also prevents you from intentionally
7441 patching your program's binary.
7443 If you'd like to be able to patch the binary, you can specify that
7444 explicitly with the @code{set write} command. For example, you might
7445 want to turn on internal debugging flags, or even to make emergency
7451 @itemx set write off
7452 If you specify @samp{set write on}, @value{GDBN} opens executable and
7453 core files for both reading and writing; if you specify @samp{set write
7454 off} (the default), @value{GDBN} opens them read-only.
7456 If you have already loaded a file, you must load it again (using the
7457 @code{exec-file} or @code{core-file} command) after changing @code{set
7458 write}, for your new setting to take effect.
7462 Display whether executable files and core files are opened for writing
7467 @chapter @value{GDBN} Files
7469 @value{GDBN} needs to know the file name of the program to be debugged,
7470 both in order to read its symbol table and in order to start your
7471 program. To debug a core dump of a previous run, you must also tell
7472 @value{GDBN} the name of the core dump file.
7475 * Files:: Commands to specify files
7476 * Symbol Errors:: Errors reading symbol files
7480 @section Commands to specify files
7482 @cindex symbol table
7483 @cindex core dump file
7485 You may want to specify executable and core dump file names. The usual
7486 way to do this is at start-up time, using the arguments to
7487 @value{GDBN}'s start-up commands (@pxref{Invocation, , Getting In and
7488 Out of @value{GDBN}}).
7490 Occasionally it is necessary to change to a different file during a
7491 @value{GDBN} session. Or you may run @value{GDBN} and forget to specify
7492 a file you want to use. In these situations the @value{GDBN} commands
7493 to specify new files are useful.
7496 @cindex executable file
7498 @item file @var{filename}
7499 Use @var{filename} as the program to be debugged. It is read for its
7500 symbols and for the contents of pure memory. It is also the program
7501 executed when you use the @code{run} command. If you do not specify a
7502 directory and the file is not found in the @value{GDBN} working directory,
7503 @value{GDBN} uses the environment variable @code{PATH} as a list of
7504 directories to search, just as the shell does when looking for a program
7505 to run. You can change the value of this variable, for both @value{GDBN}
7506 and your program, using the @code{path} command.
7508 On systems with memory-mapped files, an auxiliary file
7509 @file{@var{filename}.syms} may hold symbol table information for
7510 @var{filename}. If so, @value{GDBN} maps in the symbol table from
7511 @file{@var{filename}.syms}, starting up more quickly. See the
7512 descriptions of the file options @samp{-mapped} and @samp{-readnow}
7513 (available on the command line, and with the commands @code{file},
7514 @code{symbol-file}, or @code{add-symbol-file}, described below),
7515 for more information.
7518 @code{file} with no argument makes @value{GDBN} discard any information it
7519 has on both executable file and the symbol table.
7522 @item exec-file @r{[} @var{filename} @r{]}
7523 Specify that the program to be run (but not the symbol table) is found
7524 in @var{filename}. @value{GDBN} searches the environment variable @code{PATH}
7525 if necessary to locate your program. Omitting @var{filename} means to
7526 discard information on the executable file.
7529 @item symbol-file @r{[} @var{filename} @r{]}
7530 Read symbol table information from file @var{filename}. @code{PATH} is
7531 searched when necessary. Use the @code{file} command to get both symbol
7532 table and program to run from the same file.
7534 @code{symbol-file} with no argument clears out @value{GDBN} information on your
7535 program's symbol table.
7537 The @code{symbol-file} command causes @value{GDBN} to forget the contents
7538 of its convenience variables, the value history, and all breakpoints and
7539 auto-display expressions. This is because they may contain pointers to
7540 the internal data recording symbols and data types, which are part of
7541 the old symbol table data being discarded inside @value{GDBN}.
7543 @code{symbol-file} does not repeat if you press @key{RET} again after
7546 When @value{GDBN} is configured for a particular environment, it
7547 understands debugging information in whatever format is the standard
7548 generated for that environment; you may use either a @sc{gnu} compiler, or
7549 other compilers that adhere to the local conventions.
7550 Best results are usually obtained from @sc{gnu} compilers; for example,
7551 using @code{@value{GCC}} you can generate debugging information for
7554 For most kinds of object files, with the exception of old SVR3 systems
7555 using COFF, the @code{symbol-file} command does not normally read the
7556 symbol table in full right away. Instead, it scans the symbol table
7557 quickly to find which source files and which symbols are present. The
7558 details are read later, one source file at a time, as they are needed.
7560 The purpose of this two-stage reading strategy is to make @value{GDBN}
7561 start up faster. For the most part, it is invisible except for
7562 occasional pauses while the symbol table details for a particular source
7563 file are being read. (The @code{set verbose} command can turn these
7564 pauses into messages if desired. @xref{Messages/Warnings, ,Optional
7565 warnings and messages}.)
7567 We have not implemented the two-stage strategy for COFF yet. When the
7568 symbol table is stored in COFF format, @code{symbol-file} reads the
7569 symbol table data in full right away. Note that ``stabs-in-COFF''
7570 still does the two-stage strategy, since the debug info is actually
7574 @cindex reading symbols immediately
7575 @cindex symbols, reading immediately
7577 @cindex memory-mapped symbol file
7578 @cindex saving symbol table
7579 @item symbol-file @var{filename} @r{[} -readnow @r{]} @r{[} -mapped @r{]}
7580 @itemx file @var{filename} @r{[} -readnow @r{]} @r{[} -mapped @r{]}
7581 You can override the @value{GDBN} two-stage strategy for reading symbol
7582 tables by using the @samp{-readnow} option with any of the commands that
7583 load symbol table information, if you want to be sure @value{GDBN} has the
7584 entire symbol table available.
7586 If memory-mapped files are available on your system through the
7587 @code{mmap} system call, you can use another option, @samp{-mapped}, to
7588 cause @value{GDBN} to write the symbols for your program into a reusable
7589 file. Future @value{GDBN} debugging sessions map in symbol information
7590 from this auxiliary symbol file (if the program has not changed), rather
7591 than spending time reading the symbol table from the executable
7592 program. Using the @samp{-mapped} option has the same effect as
7593 starting @value{GDBN} with the @samp{-mapped} command-line option.
7595 You can use both options together, to make sure the auxiliary symbol
7596 file has all the symbol information for your program.
7598 The auxiliary symbol file for a program called @var{myprog} is called
7599 @samp{@var{myprog}.syms}. Once this file exists (so long as it is newer
7600 than the corresponding executable), @value{GDBN} always attempts to use
7601 it when you debug @var{myprog}; no special options or commands are
7604 The @file{.syms} file is specific to the host machine where you run
7605 @value{GDBN}. It holds an exact image of the internal @value{GDBN}
7606 symbol table. It cannot be shared across multiple host platforms.
7608 @c FIXME: for now no mention of directories, since this seems to be in
7609 @c flux. 13mar1992 status is that in theory GDB would look either in
7610 @c current dir or in same dir as myprog; but issues like competing
7611 @c GDB's, or clutter in system dirs, mean that in practice right now
7612 @c only current dir is used. FFish says maybe a special GDB hierarchy
7613 @c (eg rooted in val of env var GDBSYMS) could exist for mappable symbol
7618 @item core-file @r{[} @var{filename} @r{]}
7619 Specify the whereabouts of a core dump file to be used as the ``contents
7620 of memory''. Traditionally, core files contain only some parts of the
7621 address space of the process that generated them; @value{GDBN} can access the
7622 executable file itself for other parts.
7624 @code{core-file} with no argument specifies that no core file is
7627 Note that the core file is ignored when your program is actually running
7628 under @value{GDBN}. So, if you have been running your program and you
7629 wish to debug a core file instead, you must kill the subprocess in which
7630 the program is running. To do this, use the @code{kill} command
7631 (@pxref{Kill Process, ,Killing the child process}).
7633 @kindex add-symbol-file
7634 @cindex dynamic linking
7635 @item add-symbol-file @var{filename} @var{address}
7636 @itemx add-symbol-file @var{filename} @var{address} @r{[} -readnow @r{]} @r{[} -mapped @r{]}
7637 The @code{add-symbol-file} command reads additional symbol table information
7638 from the file @var{filename}. You would use this command when @var{filename}
7639 has been dynamically loaded (by some other means) into the program that
7640 is running. @var{address} should be the memory address at which the
7641 file has been loaded; @value{GDBN} cannot figure this out for itself.
7642 You can specify @var{address} as an expression.
7644 The symbol table of the file @var{filename} is added to the symbol table
7645 originally read with the @code{symbol-file} command. You can use the
7646 @code{add-symbol-file} command any number of times; the new symbol data thus
7647 read keeps adding to the old. To discard all old symbol data instead,
7648 use the @code{symbol-file} command.
7650 @code{add-symbol-file} does not repeat if you press @key{RET} after using it.
7652 You can use the @samp{-mapped} and @samp{-readnow} options just as with
7653 the @code{symbol-file} command, to change how @value{GDBN} manages the symbol
7654 table information for @var{filename}.
7656 @kindex add-shared-symbol-file
7657 @item add-shared-symbol-file
7658 The @code{add-shared-symbol-file} command can be used only under Harris' CXUX
7659 operating system for the Motorola 88k. @value{GDBN} automatically looks for
7660 shared libraries, however if @value{GDBN} does not find yours, you can run
7661 @code{add-shared-symbol-file}. It takes no arguments.
7665 The @code{section} command changes the base address of section SECTION of
7666 the exec file to ADDR. This can be used if the exec file does not contain
7667 section addresses, (such as in the a.out format), or when the addresses
7668 specified in the file itself are wrong. Each section must be changed
7669 separately. The ``info files'' command lists all the sections and their
7676 @code{info files} and @code{info target} are synonymous; both print the
7677 current target (@pxref{Targets, ,Specifying a Debugging Target}),
7678 including the names of the executable and core dump files currently in
7679 use by @value{GDBN}, and the files from which symbols were loaded. The
7680 command @code{help target} lists all possible targets rather than
7685 All file-specifying commands allow both absolute and relative file names
7686 as arguments. @value{GDBN} always converts the file name to an absolute file
7687 name and remembers it that way.
7689 @cindex shared libraries
7690 @value{GDBN} supports HP-UX, SunOS, SVr4, Irix 5, and IBM RS/6000 shared
7693 @value{GDBN} automatically loads symbol definitions from shared libraries
7694 when you use the @code{run} command, or when you examine a core file.
7695 (Before you issue the @code{run} command, @value{GDBN} does not understand
7696 references to a function in a shared library, however---unless you are
7697 debugging a core file).
7699 On HP-UX, if the program loads a library explicitly, @value{GDBN}
7700 automatically loads the symbols at the time of the @code{shl_load} call.
7702 @c FIXME: some @value{GDBN} release may permit some refs to undef
7703 @c FIXME...symbols---eg in a break cmd---assuming they are from a shared
7704 @c FIXME...lib; check this from time to time when updating manual
7707 @kindex info sharedlibrary
7710 @itemx info sharedlibrary
7711 Print the names of the shared libraries which are currently loaded.
7713 @kindex sharedlibrary
7715 @item sharedlibrary @var{regex}
7716 @itemx share @var{regex}
7718 Load shared object library symbols for files matching a
7719 Unix regular expression.
7720 As with files loaded automatically, it only loads shared libraries
7721 required by your program for a core file or after typing @code{run}. If
7722 @var{regex} is omitted all shared libraries required by your program are
7726 On HP-UX systems, @value{GDBN} detects the loading of a shared library
7727 and automatically reads in symbols from the newly loaded library, up to
7728 a threshold that is initially set but that you can modify if you wish.
7730 Beyond that threshold, symbols from shared libraries must be explicitly
7731 loaded. To load these symbols, use the command @code{sharedlibrary}
7732 @var{filename}. The base address of the shared library is determined
7733 automatically by @value{GDBN} and need not be specified.
7735 To display or set the threshold, use the commands:
7738 @kindex set auto-solib-add
7739 @item set auto-solib-add @var{threshold}
7740 Set the autoloading size threshold, in megabytes. If @var{threshold} is
7741 nonzero, symbols from all shared object libraries will be loaded
7742 automatically when the inferior begins execution or when the dynamic
7743 linker informs @value{GDBN} that a new library has been loaded, until
7744 the symbol table of the program and libraries exceeds this threshold.
7745 Otherwise, symbols must be loaded manually, using the
7746 @code{sharedlibrary} command. The default threshold is 100 megabytes.
7748 @kindex show auto-solib-add
7749 @item show auto-solib-add
7750 Display the current autoloading size threshold, in megabytes.
7754 @section Errors reading symbol files
7756 While reading a symbol file, @value{GDBN} occasionally encounters problems,
7757 such as symbol types it does not recognize, or known bugs in compiler
7758 output. By default, @value{GDBN} does not notify you of such problems, since
7759 they are relatively common and primarily of interest to people
7760 debugging compilers. If you are interested in seeing information
7761 about ill-constructed symbol tables, you can either ask @value{GDBN} to print
7762 only one message about each such type of problem, no matter how many
7763 times the problem occurs; or you can ask @value{GDBN} to print more messages,
7764 to see how many times the problems occur, with the @code{set
7765 complaints} command (@pxref{Messages/Warnings, ,Optional warnings and
7768 The messages currently printed, and their meanings, include:
7771 @item inner block not inside outer block in @var{symbol}
7773 The symbol information shows where symbol scopes begin and end
7774 (such as at the start of a function or a block of statements). This
7775 error indicates that an inner scope block is not fully contained
7776 in its outer scope blocks.
7778 @value{GDBN} circumvents the problem by treating the inner block as if it had
7779 the same scope as the outer block. In the error message, @var{symbol}
7780 may be shown as ``@code{(don't know)}'' if the outer block is not a
7783 @item block at @var{address} out of order
7785 The symbol information for symbol scope blocks should occur in
7786 order of increasing addresses. This error indicates that it does not
7789 @value{GDBN} does not circumvent this problem, and has trouble
7790 locating symbols in the source file whose symbols it is reading. (You
7791 can often determine what source file is affected by specifying
7792 @code{set verbose on}. @xref{Messages/Warnings, ,Optional warnings and
7795 @item bad block start address patched
7797 The symbol information for a symbol scope block has a start address
7798 smaller than the address of the preceding source line. This is known
7799 to occur in the SunOS 4.1.1 (and earlier) C compiler.
7801 @value{GDBN} circumvents the problem by treating the symbol scope block as
7802 starting on the previous source line.
7804 @item bad string table offset in symbol @var{n}
7807 Symbol number @var{n} contains a pointer into the string table which is
7808 larger than the size of the string table.
7810 @value{GDBN} circumvents the problem by considering the symbol to have the
7811 name @code{foo}, which may cause other problems if many symbols end up
7814 @item unknown symbol type @code{0x@var{nn}}
7816 The symbol information contains new data types that @value{GDBN} does
7817 not yet know how to read. @code{0x@var{nn}} is the symbol type of the
7818 misunderstood information, in hexadecimal.
7820 @value{GDBN} circumvents the error by ignoring this symbol information.
7821 This usually allows you to debug your program, though certain symbols
7822 are not accessible. If you encounter such a problem and feel like
7823 debugging it, you can debug @code{@value{GDBP}} with itself, breakpoint
7824 on @code{complain}, then go up to the function @code{read_dbx_symtab}
7825 and examine @code{*bufp} to see the symbol.
7827 @item stub type has NULL name
7829 @value{GDBN} could not find the full definition for a struct or class.
7831 @item const/volatile indicator missing (ok if using g++ v1.x), got@dots{}
7832 The symbol information for a C++ member function is missing some
7833 information that recent versions of the compiler should have output for
7836 @item info mismatch between compiler and debugger
7838 @value{GDBN} could not parse a type specification output by the compiler.
7843 @chapter Specifying a Debugging Target
7845 @cindex debugging target
7848 A @dfn{target} is the execution environment occupied by your program.
7850 Often, @value{GDBN} runs in the same host environment as your program;
7851 in that case, the debugging target is specified as a side effect when
7852 you use the @code{file} or @code{core} commands. When you need more
7853 flexibility---for example, running @value{GDBN} on a physically separate
7854 host, or controlling a standalone system over a serial port or a
7855 realtime system over a TCP/IP connection---you can use the @code{target}
7856 command to specify one of the target types configured for @value{GDBN}
7857 (@pxref{Target Commands, ,Commands for managing targets}).
7860 * Active Targets:: Active targets
7861 * Target Commands:: Commands for managing targets
7862 * Byte Order:: Choosing target byte order
7863 * Remote:: Remote debugging
7864 * KOD:: Kernel Object Display
7868 @node Active Targets
7869 @section Active targets
7871 @cindex stacking targets
7872 @cindex active targets
7873 @cindex multiple targets
7875 There are three classes of targets: processes, core files, and
7876 executable files. @value{GDBN} can work concurrently on up to three
7877 active targets, one in each class. This allows you to (for example)
7878 start a process and inspect its activity without abandoning your work on
7881 For example, if you execute @samp{gdb a.out}, then the executable file
7882 @code{a.out} is the only active target. If you designate a core file as
7883 well---presumably from a prior run that crashed and coredumped---then
7884 @value{GDBN} has two active targets and uses them in tandem, looking
7885 first in the corefile target, then in the executable file, to satisfy
7886 requests for memory addresses. (Typically, these two classes of target
7887 are complementary, since core files contain only a program's
7888 read-write memory---variables and so on---plus machine status, while
7889 executable files contain only the program text and initialized data.)
7891 When you type @code{run}, your executable file becomes an active process
7892 target as well. When a process target is active, all @value{GDBN}
7893 commands requesting memory addresses refer to that target; addresses in
7894 an active core file or executable file target are obscured while the
7895 process target is active.
7897 Use the @code{core-file} and @code{exec-file} commands to select a new
7898 core file or executable target (@pxref{Files, ,Commands to specify
7899 files}). To specify as a target a process that is already running, use
7900 the @code{attach} command (@pxref{Attach, ,Debugging an already-running
7903 @node Target Commands
7904 @section Commands for managing targets
7907 @item target @var{type} @var{parameters}
7908 Connects the @value{GDBN} host environment to a target machine or
7909 process. A target is typically a protocol for talking to debugging
7910 facilities. You use the argument @var{type} to specify the type or
7911 protocol of the target machine.
7913 Further @var{parameters} are interpreted by the target protocol, but
7914 typically include things like device names or host names to connect
7915 with, process numbers, and baud rates.
7917 The @code{target} command does not repeat if you press @key{RET} again
7918 after executing the command.
7922 Displays the names of all targets available. To display targets
7923 currently selected, use either @code{info target} or @code{info files}
7924 (@pxref{Files, ,Commands to specify files}).
7926 @item help target @var{name}
7927 Describe a particular target, including any parameters necessary to
7930 @kindex set gnutarget
7931 @item set gnutarget @var{args}
7932 @value{GDBN} uses its own library BFD to read your files. @value{GDBN}
7933 knows whether it is reading an @dfn{executable},
7934 a @dfn{core}, or a @dfn{.o} file; however, you can specify the file format
7935 with the @code{set gnutarget} command. Unlike most @code{target} commands,
7936 with @code{gnutarget} the @code{target} refers to a program, not a machine.
7938 @emph{Warning:} To specify a file format with @code{set gnutarget},
7939 you must know the actual BFD name.
7941 @noindent @xref{Files, , Commands to specify files}.
7943 @kindex show gnutarget
7944 @item show gnutarget
7945 Use the @code{show gnutarget} command to display what file format
7946 @code{gnutarget} is set to read. If you have not set @code{gnutarget},
7947 @value{GDBN} will determine the file format for each file automatically,
7948 and @code{show gnutarget} displays @samp{The current BDF target is "auto"}.
7951 Here are some common targets (available, or not, depending on the GDB
7956 @item target exec @var{program}
7957 An executable file. @samp{target exec @var{program}} is the same as
7958 @samp{exec-file @var{program}}.
7961 @item target core @var{filename}
7962 A core dump file. @samp{target core @var{filename}} is the same as
7963 @samp{core-file @var{filename}}.
7965 @kindex target remote
7966 @item target remote @var{dev}
7967 Remote serial target in GDB-specific protocol. The argument @var{dev}
7968 specifies what serial device to use for the connection (e.g.
7969 @file{/dev/ttya}). @xref{Remote, ,Remote debugging}. @code{target remote}
7970 now supports the @code{load} command. This is only useful if you have
7971 some other way of getting the stub to the target system, and you can put
7972 it somewhere in memory where it won't get clobbered by the download.
7976 Builtin CPU simulator. GDB includes simulators for most architectures.
7983 works; however, you cannot assume that a specific memory map, device
7984 drivers, or even basic I/O is available, although some simulator do
7985 provide these. For info about any processor-specific simulator details,
7986 see the appropriate section in @ref{Embedded Processors, ,Embedded
7991 Some configurations may include these targets as well:
7996 @item target nrom @var{dev}
7997 NetROM ROM emulator. This target only supports downloading.
8001 Different targets are available on different configurations of @value{GDBN};
8002 your configuration may have more or fewer targets.
8004 Many remote targets require you to download the executable's code
8005 once you've successfully established a connection.
8009 @kindex load @var{filename}
8010 @item load @var{filename}
8011 Depending on what remote debugging facilities are configured into
8012 @value{GDBN}, the @code{load} command may be available. Where it exists, it
8013 is meant to make @var{filename} (an executable) available for debugging
8014 on the remote system---by downloading, or dynamic linking, for example.
8015 @code{load} also records the @var{filename} symbol table in @value{GDBN}, like
8016 the @code{add-symbol-file} command.
8018 If your @value{GDBN} does not have a @code{load} command, attempting to
8019 execute it gets the error message ``@code{You can't do that when your
8020 target is @dots{}}''
8022 The file is loaded at whatever address is specified in the executable.
8023 For some object file formats, you can specify the load address when you
8024 link the program; for other formats, like a.out, the object file format
8025 specifies a fixed address.
8026 @c FIXME! This would be a good place for an xref to the GNU linker doc.
8028 @code{load} does not repeat if you press @key{RET} again after using it.
8032 @section Choosing target byte order
8034 @cindex choosing target byte order
8035 @cindex target byte order
8036 @kindex set endian big
8037 @kindex set endian little
8038 @kindex set endian auto
8041 Some types of processors, such as the MIPS, PowerPC, and Hitachi SH,
8042 offer the ability to run either big-endian or little-endian byte
8043 orders. Usually the executable or symbol will include a bit to
8044 designate the endian-ness, and you will not need to worry about
8045 which to use. However, you may still find it useful to adjust
8046 GDB's idea of processor endian-ness manually.
8049 @kindex set endian big
8050 @item set endian big
8051 Instruct @value{GDBN} to assume the target is big-endian.
8053 @kindex set endian little
8054 @item set endian little
8055 Instruct @value{GDBN} to assume the target is little-endian.
8057 @kindex set endian auto
8058 @item set endian auto
8059 Instruct @value{GDBN} to use the byte order associated with the
8063 Display @value{GDBN}'s current idea of the target byte order.
8067 Note that these commands merely adjust interpretation of symbolic
8068 data on the host, and that they have absolutely no effect on the
8072 @section Remote debugging
8073 @cindex remote debugging
8075 If you are trying to debug a program running on a machine that cannot run
8076 @value{GDBN} in the usual way, it is often useful to use remote debugging.
8077 For example, you might use remote debugging on an operating system kernel,
8078 or on a small system which does not have a general purpose operating system
8079 powerful enough to run a full-featured debugger.
8081 Some configurations of @value{GDBN} have special serial or TCP/IP interfaces
8082 to make this work with particular debugging targets. In addition,
8083 @value{GDBN} comes with a generic serial protocol (specific to @value{GDBN},
8084 but not specific to any particular target system) which you can use if you
8085 write the remote stubs---the code that runs on the remote system to
8086 communicate with @value{GDBN}.
8088 Other remote targets may be available in your
8089 configuration of @value{GDBN}; use @code{help target} to list them.
8092 * Remote Serial:: @value{GDBN} remote serial protocol
8096 @subsection The @value{GDBN} remote serial protocol
8098 @cindex remote serial debugging, overview
8099 To debug a program running on another machine (the debugging
8100 @dfn{target} machine), you must first arrange for all the usual
8101 prerequisites for the program to run by itself. For example, for a C
8106 A startup routine to set up the C runtime environment; these usually
8107 have a name like @file{crt0}. The startup routine may be supplied by
8108 your hardware supplier, or you may have to write your own.
8111 You probably need a C subroutine library to support your program's
8112 subroutine calls, notably managing input and output.
8115 A way of getting your program to the other machine---for example, a
8116 download program. These are often supplied by the hardware
8117 manufacturer, but you may have to write your own from hardware
8121 The next step is to arrange for your program to use a serial port to
8122 communicate with the machine where @value{GDBN} is running (the @dfn{host}
8123 machine). In general terms, the scheme looks like this:
8127 @value{GDBN} already understands how to use this protocol; when everything
8128 else is set up, you can simply use the @samp{target remote} command
8129 (@pxref{Targets,,Specifying a Debugging Target}).
8131 @item On the target,
8132 you must link with your program a few special-purpose subroutines that
8133 implement the @value{GDBN} remote serial protocol. The file containing these
8134 subroutines is called a @dfn{debugging stub}.
8136 On certain remote targets, you can use an auxiliary program
8137 @code{gdbserver} instead of linking a stub into your program.
8138 @xref{Server,,Using the @code{gdbserver} program}, for details.
8141 The debugging stub is specific to the architecture of the remote
8142 machine; for example, use @file{sparc-stub.c} to debug programs on
8145 @cindex remote serial stub list
8146 These working remote stubs are distributed with @value{GDBN}:
8154 For Intel 386 and compatible architectures.
8158 @cindex Motorola 680x0
8160 For Motorola 680x0 architectures.
8166 For Hitachi SH architectures.
8169 @kindex sparc-stub.c
8171 For @sc{sparc} architectures.
8174 @kindex sparcl-stub.c
8177 For Fujitsu @sc{sparclite} architectures.
8181 The @file{README} file in the @value{GDBN} distribution may list other
8182 recently added stubs.
8185 * Stub Contents:: What the stub can do for you
8186 * Bootstrapping:: What you must do for the stub
8187 * Debug Session:: Putting it all together
8188 * Protocol:: Definition of the communication protocol
8189 * Server:: Using the `gdbserver' program
8190 * NetWare:: Using the `gdbserve.nlm' program
8194 @subsubsection What the stub can do for you
8196 @cindex remote serial stub
8197 The debugging stub for your architecture supplies these three
8201 @item set_debug_traps
8202 @kindex set_debug_traps
8203 @cindex remote serial stub, initialization
8204 This routine arranges for @code{handle_exception} to run when your
8205 program stops. You must call this subroutine explicitly near the
8206 beginning of your program.
8208 @item handle_exception
8209 @kindex handle_exception
8210 @cindex remote serial stub, main routine
8211 This is the central workhorse, but your program never calls it
8212 explicitly---the setup code arranges for @code{handle_exception} to
8213 run when a trap is triggered.
8215 @code{handle_exception} takes control when your program stops during
8216 execution (for example, on a breakpoint), and mediates communications
8217 with @value{GDBN} on the host machine. This is where the communications
8218 protocol is implemented; @code{handle_exception} acts as the @value{GDBN}
8219 representative on the target machine; it begins by sending summary
8220 information on the state of your program, then continues to execute,
8221 retrieving and transmitting any information @value{GDBN} needs, until you
8222 execute a @value{GDBN} command that makes your program resume; at that point,
8223 @code{handle_exception} returns control to your own code on the target
8227 @cindex @code{breakpoint} subroutine, remote
8228 Use this auxiliary subroutine to make your program contain a
8229 breakpoint. Depending on the particular situation, this may be the only
8230 way for @value{GDBN} to get control. For instance, if your target
8231 machine has some sort of interrupt button, you won't need to call this;
8232 pressing the interrupt button transfers control to
8233 @code{handle_exception}---in effect, to @value{GDBN}. On some machines,
8234 simply receiving characters on the serial port may also trigger a trap;
8235 again, in that situation, you don't need to call @code{breakpoint} from
8236 your own program---simply running @samp{target remote} from the host
8237 @value{GDBN} session gets control.
8239 Call @code{breakpoint} if none of these is true, or if you simply want
8240 to make certain your program stops at a predetermined point for the
8241 start of your debugging session.
8245 @subsubsection What you must do for the stub
8247 @cindex remote stub, support routines
8248 The debugging stubs that come with @value{GDBN} are set up for a particular
8249 chip architecture, but they have no information about the rest of your
8250 debugging target machine.
8252 First of all you need to tell the stub how to communicate with the
8256 @item int getDebugChar()
8257 @kindex getDebugChar
8258 Write this subroutine to read a single character from the serial port.
8259 It may be identical to @code{getchar} for your target system; a
8260 different name is used to allow you to distinguish the two if you wish.
8262 @item void putDebugChar(int)
8263 @kindex putDebugChar
8264 Write this subroutine to write a single character to the serial port.
8265 It may be identical to @code{putchar} for your target system; a
8266 different name is used to allow you to distinguish the two if you wish.
8269 @cindex control C, and remote debugging
8270 @cindex interrupting remote targets
8271 If you want @value{GDBN} to be able to stop your program while it is
8272 running, you need to use an interrupt-driven serial driver, and arrange
8273 for it to stop when it receives a @code{^C} (@samp{\003}, the control-C
8274 character). That is the character which @value{GDBN} uses to tell the
8275 remote system to stop.
8277 Getting the debugging target to return the proper status to @value{GDBN}
8278 probably requires changes to the standard stub; one quick and dirty way
8279 is to just execute a breakpoint instruction (the ``dirty'' part is that
8280 @value{GDBN} reports a @code{SIGTRAP} instead of a @code{SIGINT}).
8282 Other routines you need to supply are:
8285 @item void exceptionHandler (int @var{exception_number}, void *@var{exception_address})
8286 @kindex exceptionHandler
8287 Write this function to install @var{exception_address} in the exception
8288 handling tables. You need to do this because the stub does not have any
8289 way of knowing what the exception handling tables on your target system
8290 are like (for example, the processor's table might be in @sc{rom},
8291 containing entries which point to a table in @sc{ram}).
8292 @var{exception_number} is the exception number which should be changed;
8293 its meaning is architecture-dependent (for example, different numbers
8294 might represent divide by zero, misaligned access, etc). When this
8295 exception occurs, control should be transferred directly to
8296 @var{exception_address}, and the processor state (stack, registers,
8297 and so on) should be just as it is when a processor exception occurs. So if
8298 you want to use a jump instruction to reach @var{exception_address}, it
8299 should be a simple jump, not a jump to subroutine.
8301 For the 386, @var{exception_address} should be installed as an interrupt
8302 gate so that interrupts are masked while the handler runs. The gate
8303 should be at privilege level 0 (the most privileged level). The
8304 @sc{sparc} and 68k stubs are able to mask interrupts themselves without
8305 help from @code{exceptionHandler}.
8307 @item void flush_i_cache()
8308 @kindex flush_i_cache
8309 (sparc and sparclite only) Write this subroutine to flush the
8310 instruction cache, if any, on your target machine. If there is no
8311 instruction cache, this subroutine may be a no-op.
8313 On target machines that have instruction caches, @value{GDBN} requires this
8314 function to make certain that the state of your program is stable.
8318 You must also make sure this library routine is available:
8321 @item void *memset(void *, int, int)
8323 This is the standard library function @code{memset} that sets an area of
8324 memory to a known value. If you have one of the free versions of
8325 @code{libc.a}, @code{memset} can be found there; otherwise, you must
8326 either obtain it from your hardware manufacturer, or write your own.
8329 If you do not use the GNU C compiler, you may need other standard
8330 library subroutines as well; this varies from one stub to another,
8331 but in general the stubs are likely to use any of the common library
8332 subroutines which @code{gcc} generates as inline code.
8336 @subsubsection Putting it all together
8338 @cindex remote serial debugging summary
8339 In summary, when your program is ready to debug, you must follow these
8344 Make sure you have the supporting low-level routines
8345 (@pxref{Bootstrapping,,What you must do for the stub}):
8347 @code{getDebugChar}, @code{putDebugChar},
8348 @code{flush_i_cache}, @code{memset}, @code{exceptionHandler}.
8352 Insert these lines near the top of your program:
8360 For the 680x0 stub only, you need to provide a variable called
8361 @code{exceptionHook}. Normally you just use:
8364 void (*exceptionHook)() = 0;
8367 but if before calling @code{set_debug_traps}, you set it to point to a
8368 function in your program, that function is called when
8369 @code{@value{GDBN}} continues after stopping on a trap (for example, bus
8370 error). The function indicated by @code{exceptionHook} is called with
8371 one parameter: an @code{int} which is the exception number.
8374 Compile and link together: your program, the @value{GDBN} debugging stub for
8375 your target architecture, and the supporting subroutines.
8378 Make sure you have a serial connection between your target machine and
8379 the @value{GDBN} host, and identify the serial port on the host.
8382 @c The "remote" target now provides a `load' command, so we should
8383 @c document that. FIXME.
8384 Download your program to your target machine (or get it there by
8385 whatever means the manufacturer provides), and start it.
8388 To start remote debugging, run @value{GDBN} on the host machine, and specify
8389 as an executable file the program that is running in the remote machine.
8390 This tells @value{GDBN} how to find your program's symbols and the contents
8393 @cindex serial line, @code{target remote}
8394 Then establish communication using the @code{target remote} command.
8395 Its argument specifies how to communicate with the target
8396 machine---either via a devicename attached to a direct serial line, or a
8397 TCP port (usually to a terminal server which in turn has a serial line
8398 to the target). For example, to use a serial line connected to the
8399 device named @file{/dev/ttyb}:
8402 target remote /dev/ttyb
8405 @cindex TCP port, @code{target remote}
8406 To use a TCP connection, use an argument of the form
8407 @code{@var{host}:port}. For example, to connect to port 2828 on a
8408 terminal server named @code{manyfarms}:
8411 target remote manyfarms:2828
8415 Now you can use all the usual commands to examine and change data and to
8416 step and continue the remote program.
8418 To resume the remote program and stop debugging it, use the @code{detach}
8421 @cindex interrupting remote programs
8422 @cindex remote programs, interrupting
8423 Whenever @value{GDBN} is waiting for the remote program, if you type the
8424 interrupt character (often @key{C-C}), @value{GDBN} attempts to stop the
8425 program. This may or may not succeed, depending in part on the hardware
8426 and the serial drivers the remote system uses. If you type the
8427 interrupt character once again, @value{GDBN} displays this prompt:
8430 Interrupted while waiting for the program.
8431 Give up (and stop debugging it)? (y or n)
8434 If you type @kbd{y}, @value{GDBN} abandons the remote debugging session.
8435 (If you decide you want to try again later, you can use @samp{target
8436 remote} again to connect once more.) If you type @kbd{n}, @value{GDBN}
8437 goes back to waiting.
8440 @subsubsection Communication protocol
8442 @cindex debugging stub, example
8443 @cindex remote stub, example
8444 @cindex stub example, remote debugging
8445 The stub files provided with @value{GDBN} implement the target side of the
8446 communication protocol, and the @value{GDBN} side is implemented in the
8447 @value{GDBN} source file @file{remote.c}. Normally, you can simply allow
8448 these subroutines to communicate, and ignore the details. (If you're
8449 implementing your own stub file, you can still ignore the details: start
8450 with one of the existing stub files. @file{sparc-stub.c} is the best
8451 organized, and therefore the easiest to read.)
8453 However, there may be occasions when you need to know something about
8454 the protocol---for example, if there is only one serial port to your
8455 target machine, you might want your program to do something special if
8456 it recognizes a packet meant for @value{GDBN}.
8458 In the examples below, @samp{<-} and @samp{->} are used to indicate
8459 transmitted and received data respectfully.
8461 @cindex protocol, @value{GDBN} remote serial
8462 @cindex serial protocol, @value{GDBN} remote
8463 @cindex remote serial protocol
8464 All @value{GDBN} commands and responses (other than acknowledgments)
8465 are sent as a @var{packet}. A @var{packet} is introduced with the
8466 character @samp{$}, this is followed by an optional two-digit
8467 @var{sequence-id} and the character @samp{:}, the actual
8468 @var{packet-data}, and the terminating character @samp{#} followed by a
8469 two-digit @var{checksum}:
8472 @code{$}@var{packet-data}@code{#}@var{checksum}
8475 or, with the optional @var{sequence-id}:
8477 @code{$}@var{sequence-id}@code{:}@var{packet-data}@code{#}@var{checksum}
8480 @cindex checksum, for @value{GDBN} remote
8482 The two-digit @var{checksum} is computed as the modulo 256 sum of all
8483 characters between the leading @samp{$} and the trailing @samp{#} (that
8484 consisting of both the optional @var{sequence-id}@code{:} and the actual
8487 @cindex sequence-id, for @value{GDBN} remote
8489 The two-digit @var{sequence-id}, when present, is returned with the
8490 acknowledgment. Beyond that its meaning is poorly defined.
8491 @value{GDBN} is not known to output @var{sequence-id}s.
8493 When either the host or the target machine receives a packet, the first
8494 response expected is an acknowledgment: either @samp{+} (to indicate
8495 the package was received correctly) or @samp{-} (to request
8499 <- @code{$}@var{packet-data}@code{#}@var{checksum}
8503 If the received packet included a @var{sequence-id} than that is
8504 appended to a positive acknowledgment:
8507 <- @code{$}@var{sequence-id}@code{:}@var{packet-data}@code{#}@var{checksum}
8508 -> @code{+}@var{sequence-id}
8511 The host (@value{GDBN}) sends @var{command}s, and the target (the
8512 debugging stub incorporated in your program) sends a @var{response}. In
8513 the case of step and continue @var{command}s, the response is only sent
8514 when the operation has completed (the target has again stopped).
8516 @var{packet-data} consists of a sequence of characters with the
8517 exception of @samp{#} and @samp{$} (see @samp{X} packet for an
8518 exception). @samp{:} can not appear as the third character in a packet.
8519 Fields within the packet should be separated using @samp{,} and @samp{;}
8520 (unfortunately some packets chose to use @samp{:}). Except where
8521 otherwise noted all numbers are represented in HEX with leading zeros
8524 Response @var{data} can be run-length encoded to save space. A @samp{*}
8525 means that the next character is an ASCII encoding giving a repeat count
8526 which stands for that many repetitions of the character preceding the
8527 @samp{*}. The encoding is @code{n+29}, yielding a printable character
8528 where @code{n >=3} (which is where rle starts to win). Don't use an
8536 means the same as "0000".
8538 The error response, returned for some packets includes a two character
8539 error number. That number is not well defined.
8541 For any @var{command} not supported by the stub, an empty response
8542 (@samp{$#00}) should be returned. That way it is possible to extend the
8543 protocol. A newer @value{GDBN} can tell if a packet is supported based
8546 Below is a complete list of all currently defined @var{command}s and
8547 their corresponding response @var{data}:
8549 @multitable @columnfractions .30 .30 .40
8554 @item extended ops @emph{(optional)}
8557 Use the extended remote protocol. Sticky -- only needs to be set once.
8558 The extended remote protocol support the @samp{R} packet.
8562 Stubs that support the extended remote protocol return @samp{} which,
8563 unfortunately, is identical to the response returned by stubs that do not
8564 support protocol extensions.
8569 Reply the current reason for stopping. This is the same reply as is
8570 generated for step or cont : @code{S}@var{AA} where @var{AA} is the
8575 @tab Reserved for future use
8577 @item set program arguments @strong{(reserved)} @emph{(optional)}
8578 @tab @code{A}@var{arglen}@code{,}@var{argnum}@code{,}@var{arg}@code{,...}
8580 Initialized @samp{argv[]} array passed into program. @var{arglen}
8581 specifies the number of bytes in the hex encoded byte stream @var{arg}.
8583 @tab reply @code{OK}
8585 @tab reply @code{E}@var{NN}
8587 @item set baud @strong{(deprecated)}
8588 @tab @code{b}@var{baud}
8590 Change the serial line speed to @var{baud}. JTC: @emph{When does the
8591 transport layer state change? When it's received, or after the ACK is
8592 transmitted. In either case, there are problems if the command or the
8593 acknowledgment packet is dropped.} Stan: @emph{If people really wanted
8594 to add something like this, and get it working for the first time, they
8595 ought to modify ser-unix.c to send some kind of out-of-band message to a
8596 specially-setup stub and have the switch happen "in between" packets, so
8597 that from remote protocol's point of view, nothing actually
8600 @item set breakpoint @strong{(deprecated)}
8601 @tab @code{B}@var{addr},@var{mode}
8603 Set (@var{mode} is @samp{S}) or clear (@var{mode} is @samp{C}) a
8604 breakpoint at @var{addr}. @emph{This has been replaced by the @samp{Z} and
8608 @tab @code{c}@var{addr}
8610 @var{addr} is address to resume. If @var{addr} is omitted, resume at
8616 @item continue with signal @emph{(optional)}
8617 @tab @code{C}@var{sig}@code{;}@var{addr}
8619 Continue with signal @var{sig} (hex signal number). If
8620 @code{;}@var{addr} is omitted, resume at same address.
8625 @item toggle debug @emph{(optional)}
8628 toggle debug flag (see 386 & 68k stubs)
8630 @item detach @emph{(optional)}
8636 @tab Reserved for future use
8640 @tab Reserved for future use
8644 @tab Reserved for future use
8648 @tab Reserved for future use
8650 @item read registers
8652 @tab Read general registers.
8654 @tab reply @var{XX...}
8656 Each byte of register data is described by two hex digits. The bytes
8657 with the register are transmitted in target byte order. The size of
8658 each register and their position within the @samp{g} @var{packet} is
8659 determined by the @var{REGISTER_RAW_SIZE} and @var{REGISTER_NAME}
8662 @tab @code{E}@var{NN}
8666 @tab @code{G}@var{XX...}
8668 See @samp{g} for a description of the @var{XX...} data.
8670 @tab reply @code{OK}
8673 @tab reply @code{E}@var{NN}
8678 @tab Reserved for future use
8680 @item set thread @emph{(optional)}
8681 @tab @code{H}@var{c}@var{t...}
8683 Set thread for subsequent operations. @var{c} = @samp{c} for thread
8684 used in step and continue; @var{t...} can be -1 for all threads.
8685 @var{c} = @samp{g} for thread used in other operations. If zero, pick a
8688 @tab reply @code{OK}
8691 @tab reply @code{E}@var{NN}
8694 @item cycle step @strong{(draft)} @emph{(optional)}
8695 @tab @code{i}@var{addr}@code{,}@var{nnn}
8697 Step the remote target by a single clock cycle. If @code{,}@var{nnn} is
8698 present, cycle step @var{nnn} cycles. If @var{addr} is present, cycle
8699 step starting at that address.
8701 @item signal then cycle step @strong{(reserved)} @emph{(optional)}
8704 See @samp{i} and @samp{S} for likely syntax and semantics.
8708 @tab Reserved for future use
8712 @tab Reserved for future use
8714 @item kill request @emph{(optional)}
8720 @tab Reserved for future use
8724 @tab Reserved for future use
8727 @tab @code{m}@var{addr}@code{,}@var{length}
8729 Read @var{length} bytes of memory starting at address @var{addr}.
8731 @tab reply @var{XX...}
8733 @var{XX...} is mem contents. Can be fewer bytes than requested if able to
8734 read only part of the data.
8736 @tab reply @code{E}@var{NN}
8737 @tab @var{NN} is errno
8740 @tab @code{M}@var{addr},@var{length}@code{:}@var{XX...}
8742 Write @var{length} bytes of memory starting at address @var{addr}.
8743 @var{XX...} is the data.
8745 @tab reply @code{OK}
8748 @tab reply @code{E}@var{NN}
8750 for an error (this includes the case where only part of the data was
8755 @tab Reserved for future use
8759 @tab Reserved for future use
8763 @tab Reserved for future use
8767 @tab Reserved for future use
8769 @item read reg @strong{(reserved)}
8770 @tab @code{p}@var{n...}
8774 @tab return @var{r....}
8775 @tab The hex encoded value of the register in target byte order.
8777 @item write reg @emph{(optional)}
8778 @tab @code{P}@var{n...}@code{=}@var{r...}
8780 Write register @var{n...} with value @var{r...}, which contains two hex
8781 digits for each byte in the register (target byte order).
8783 @tab reply @code{OK}
8786 @tab reply @code{E}@var{NN}
8789 @item general query @emph{(optional)}
8790 @tab @code{q}@var{query}
8792 Request info about @var{query}. In general @value{GDBN} @var{query}'s
8793 have a leading upper case letter. Custom vendor queries should use a
8794 leading lower case letter and a company prefix, ex: @samp{qfsf.var}.
8795 @var{query} may optionally be followed by a @samp{,} or @samp{;}
8796 separated list. Stubs should ensure that they fully match any
8799 @tab reply @code{XX...}
8800 @tab Hex encoded data from query. The reply can not be empty.
8802 @tab reply @code{E}@var{NN}
8806 @tab Indicating an unrecognized @var{query}.
8808 @item current thread
8809 @tab @code{q}@code{C}
8810 @tab Return the current thread id.
8812 @tab reply @code{QC}@var{pid}
8814 Where @var{pid} is a HEX encoded 16 bit process id.
8817 @tab Any other reply implies the old pid.
8819 @item compute CRC of memory block
8820 @tab @code{q}@code{CRC:}@var{addr}@code{,}@var{length}
8823 @tab reply @code{E}@var{NN}
8824 @tab An error (such as memory fault)
8826 @tab reply @code{C}@var{CRC32}
8827 @tab A 32 bit cyclic redundancy check of the specified memory region.
8829 @item query @var{LIST} or @var{threadLIST}
8830 @tab @code{q}@code{L}@var{startflag}@var{threadcount}@var{nextthread}
8832 Obtain thread information from RTOS. @var{startflag} is one hex digit;
8833 @var{threadcount} is two hex digits; and @var{nextthread} is 16 hex
8838 See @code{remote.c:parse_threadlist_response()}.
8840 @item query sect offs
8841 @tab @code{q}@code{Offsets}
8842 @tab Get section offsets.
8844 @tab reply @code{Text=}@var{xxx}@code{;Data=}@var{yyy}@code{;Bss=}@var{zzz}
8846 @item thread info request
8847 @tab @code{q}@code{P}@var{mode}@var{threadid}
8849 Returns information on @var{threadid}. Where: @var{mode} is a hex
8850 encoded 32 bit mode; @var{threadid} is a hex encoded 64 bit thread ID.
8854 See @code{remote.c:remote_unpack_thread_info_response()}.
8856 @item remote command
8857 @tab @code{q}@code{Rcmd,}@var{COMMAND}
8859 @var{COMMAND} (hex encoded) is passed to the local interpreter for
8860 execution. Invalid commands should be reported using the output string.
8861 Before the final result packet, the target may also respond with a
8862 number of intermediate @code{O}@var{OUTPUT} console output
8863 packets. @emph{Implementors should note that providing access to a
8864 stubs's interpreter may have security implications}.
8866 @tab reply @code{OK}
8868 A command response with no output.
8870 @tab reply @var{OUTPUT}
8872 A command response with the hex encoded output string @var{OUTPUT}.
8874 @tab reply @code{E}@var{NN}
8876 Indicate a badly formed request.
8881 When @samp{q}@samp{Rcmd} is not recognized.
8883 @item general set @emph{(optional)}
8884 @tab @code{Q}@var{var}@code{=}@var{val}
8886 Set value of @var{var} to @var{val}. See @samp{q} for a discussing of
8889 @item reset @emph{(optional)}
8891 @tab reset -- see sparc stub.
8893 @item remote restart @emph{(optional)}
8894 @tab @code{R}@var{XX}
8896 Restart the remote server. @var{XX} while needed has no clear
8899 @item step @emph{(optional)}
8900 @tab @code{s}@var{addr}
8902 @var{addr} is address to resume. If @var{addr} is omitted, resume at
8908 @item step with signal @emph{(optional)}
8909 @tab @code{S}@var{sig}@code{;}@var{addr}
8911 Like @samp{C} but step not continue.
8916 @item search @emph{(optional)}
8917 @tab @code{t}@var{addr}@code{:}@var{PP}@code{,}@var{MM}
8919 Search backwards starting at address @var{addr} for a match with pattern
8920 @var{PP} and mask @var{MM}. @var{PP} and @var{MM} are 4
8921 bytes. @var{addr} must be at least 3 digits.
8923 @item thread alive @emph{(optional)}
8924 @tab @code{T}@var{XX}
8925 @tab Find out if the thread XX is alive.
8927 @tab reply @code{OK}
8928 @tab thread is still alive
8930 @tab reply @code{E}@var{NN}
8935 @tab Reserved for future use
8939 @tab Reserved for future use
8943 @tab Reserved for future use
8947 @tab Reserved for future use
8951 @tab Reserved for future use
8955 @tab Reserved for future use
8959 @tab Reserved for future use
8961 @item write mem (binary) @emph{(optional)}
8962 @tab @code{X}@var{addr}@code{,}@var{length}@var{:}@var{XX...}
8964 @var{addr} is address, @var{length} is number of bytes, @var{XX...} is
8967 @tab reply @code{OK}
8970 @tab reply @code{E}@var{NN}
8975 @tab Reserved for future use
8979 @tab Reserved for future use
8981 @item remove break or watchpoint @strong{(draft)} @emph{(optional)}
8982 @tab @code{z}@var{t}@code{,}@var{addr}@code{,}@var{length}
8986 @item insert break or watchpoint @strong{(draft)} @emph{(optional)}
8987 @tab @code{Z}@var{t}@code{,}@var{addr}@code{,}@var{length}
8989 @var{t} is type: @samp{0} - software breakpoint, @samp{1} - hardware
8990 breakpoint, @samp{2} - write watchpoint, @samp{3} - read watchpoint,
8991 @samp{4} - access watchpoint; @var{addr} is address; @var{length} is in
8992 bytes. For a software breakpoint, @var{length} specifies the size of
8993 the instruction to be patched. For hardware breakpoints and watchpoints
8994 @var{length} specifies the memory region to be monitored.
8996 @tab reply @code{E}@var{NN}
8999 @tab reply @code{OK}
9003 @tab If not supported.
9007 @tab Reserved for future use
9011 In the case of the @samp{C}, @samp{c}, @samp{S} and @samp{s} packets,
9012 there is no immediate response. The reply, described below, comes when
9015 @multitable @columnfractions .4 .6
9017 @item @code{S}@var{AA}
9018 @tab @var{AA} is the signal number
9020 @item @code{T}@var{AA}@var{n...}@code{:}@var{r...}@code{;}@var{n...}@code{:}@var{r...}@code{;}@var{n...}@code{:}@var{r...}@code{;}
9022 @var{AA} = two hex digit signal number; @var{n...} = register number
9023 (hex), @var{r...} = target byte ordered register contents, size defined
9024 by @code{REGISTER_RAW_SIZE}; @var{n...} = @samp{thread}, @var{r...} =
9025 thread process ID, this is a hex integer; @var{n...} = other string not
9026 starting with valid hex digit. @value{GDBN} should ignore this
9027 @var{n...}, @var{r...} pair and go on to the next. This way we can
9028 extend the protocol.
9030 @item @code{W}@var{AA}
9032 The process exited, and @var{AA} is the exit status. This is only
9033 applicable for certains sorts of targets.
9035 @item @code{X}@var{AA}
9037 The process terminated with signal @var{AA}.
9039 @item @code{N}@var{AA}@code{;}@var{tttttttt}@code{;}@var{dddddddd}@code{;}@var{bbbbbbbb} @strong{(obsolete)}
9041 @var{AA} = signal number; @var{tttttttt} = address of symbol "_start";
9042 @var{dddddddd} = base of data section; @var{bbbbbbbb} = base of bss
9043 section. @emph{Note: only used by Cisco Systems targets. The difference
9044 between this reply and the "qOffsets" query is that the 'N' packet may
9045 arrive spontaneously whereas the 'qOffsets' is a query initiated by the
9048 @item @code{O}@var{XX...}
9050 @var{XX...} is hex encoding of ASCII data. This can happen at any time
9051 while the program is running and the debugger should continue to wait
9056 Example sequence of a target being re-started. Notice how the restart
9057 does not get any direct output:
9062 @emph{target restarts}
9065 -> @code{T001:1234123412341234}
9069 Example sequence of a target being stepped by a single instruction:
9077 -> @code{T001:1234123412341234}
9085 @kindex set remotedebug
9086 @kindex show remotedebug
9087 @cindex packets, reporting on stdout
9088 @cindex serial connections, debugging
9089 If you have trouble with the serial connection, you can use the command
9090 @code{set remotedebug}. This makes @value{GDBN} report on all packets sent
9091 back and forth across the serial line to the remote machine. The
9092 packet-debugging information is printed on the @value{GDBN} standard output
9093 stream. @code{set remotedebug off} turns it off, and @code{show
9094 remotedebug} shows you its current state.
9097 @subsubsection Using the @code{gdbserver} program
9100 @cindex remote connection without stubs
9101 @code{gdbserver} is a control program for Unix-like systems, which
9102 allows you to connect your program with a remote @value{GDBN} via
9103 @code{target remote}---but without linking in the usual debugging stub.
9105 @code{gdbserver} is not a complete replacement for the debugging stubs,
9106 because it requires essentially the same operating-system facilities
9107 that @value{GDBN} itself does. In fact, a system that can run
9108 @code{gdbserver} to connect to a remote @value{GDBN} could also run
9109 @value{GDBN} locally! @code{gdbserver} is sometimes useful nevertheless,
9110 because it is a much smaller program than @value{GDBN} itself. It is
9111 also easier to port than all of @value{GDBN}, so you may be able to get
9112 started more quickly on a new system by using @code{gdbserver}.
9113 Finally, if you develop code for real-time systems, you may find that
9114 the tradeoffs involved in real-time operation make it more convenient to
9115 do as much development work as possible on another system, for example
9116 by cross-compiling. You can use @code{gdbserver} to make a similar
9117 choice for debugging.
9119 @value{GDBN} and @code{gdbserver} communicate via either a serial line
9120 or a TCP connection, using the standard @value{GDBN} remote serial
9124 @item On the target machine,
9125 you need to have a copy of the program you want to debug.
9126 @code{gdbserver} does not need your program's symbol table, so you can
9127 strip the program if necessary to save space. @value{GDBN} on the host
9128 system does all the symbol handling.
9130 To use the server, you must tell it how to communicate with @value{GDBN};
9131 the name of your program; and the arguments for your program. The
9135 target> gdbserver @var{comm} @var{program} [ @var{args} @dots{} ]
9138 @var{comm} is either a device name (to use a serial line) or a TCP
9139 hostname and portnumber. For example, to debug Emacs with the argument
9140 @samp{foo.txt} and communicate with @value{GDBN} over the serial port
9144 target> gdbserver /dev/com1 emacs foo.txt
9147 @code{gdbserver} waits passively for the host @value{GDBN} to communicate
9150 To use a TCP connection instead of a serial line:
9153 target> gdbserver host:2345 emacs foo.txt
9156 The only difference from the previous example is the first argument,
9157 specifying that you are communicating with the host @value{GDBN} via
9158 TCP. The @samp{host:2345} argument means that @code{gdbserver} is to
9159 expect a TCP connection from machine @samp{host} to local TCP port 2345.
9160 (Currently, the @samp{host} part is ignored.) You can choose any number
9161 you want for the port number as long as it does not conflict with any
9162 TCP ports already in use on the target system (for example, @code{23} is
9163 reserved for @code{telnet}).@footnote{If you choose a port number that
9164 conflicts with another service, @code{gdbserver} prints an error message
9165 and exits.} You must use the same port number with the host @value{GDBN}
9166 @code{target remote} command.
9168 @item On the @value{GDBN} host machine,
9169 you need an unstripped copy of your program, since @value{GDBN} needs
9170 symbols and debugging information. Start up @value{GDBN} as usual,
9171 using the name of the local copy of your program as the first argument.
9172 (You may also need the @w{@samp{--baud}} option if the serial line is
9173 running at anything other than 9600 bps.) After that, use @code{target
9174 remote} to establish communications with @code{gdbserver}. Its argument
9175 is either a device name (usually a serial device, like
9176 @file{/dev/ttyb}), or a TCP port descriptor in the form
9177 @code{@var{host}:@var{PORT}}. For example:
9180 (@value{GDBP}) target remote /dev/ttyb
9184 communicates with the server via serial line @file{/dev/ttyb}, and
9187 (@value{GDBP}) target remote the-target:2345
9191 communicates via a TCP connection to port 2345 on host @w{@file{the-target}}.
9192 For TCP connections, you must start up @code{gdbserver} prior to using
9193 the @code{target remote} command. Otherwise you may get an error whose
9194 text depends on the host system, but which usually looks something like
9195 @samp{Connection refused}.
9199 @subsubsection Using the @code{gdbserve.nlm} program
9201 @kindex gdbserve.nlm
9202 @code{gdbserve.nlm} is a control program for NetWare systems, which
9203 allows you to connect your program with a remote @value{GDBN} via
9204 @code{target remote}.
9206 @value{GDBN} and @code{gdbserve.nlm} communicate via a serial line,
9207 using the standard @value{GDBN} remote serial protocol.
9210 @item On the target machine,
9211 you need to have a copy of the program you want to debug.
9212 @code{gdbserve.nlm} does not need your program's symbol table, so you
9213 can strip the program if necessary to save space. @value{GDBN} on the
9214 host system does all the symbol handling.
9216 To use the server, you must tell it how to communicate with
9217 @value{GDBN}; the name of your program; and the arguments for your
9218 program. The syntax is:
9221 load gdbserve [ BOARD=@var{board} ] [ PORT=@var{port} ]
9222 [ BAUD=@var{baud} ] @var{program} [ @var{args} @dots{} ]
9225 @var{board} and @var{port} specify the serial line; @var{baud} specifies
9226 the baud rate used by the connection. @var{port} and @var{node} default
9227 to 0, @var{baud} defaults to 9600 bps.
9229 For example, to debug Emacs with the argument @samp{foo.txt}and
9230 communicate with @value{GDBN} over serial port number 2 or board 1
9231 using a 19200 bps connection:
9234 load gdbserve BOARD=1 PORT=2 BAUD=19200 emacs foo.txt
9237 @item On the @value{GDBN} host machine,
9238 you need an unstripped copy of your program, since @value{GDBN} needs
9239 symbols and debugging information. Start up @value{GDBN} as usual,
9240 using the name of the local copy of your program as the first argument.
9241 (You may also need the @w{@samp{--baud}} option if the serial line is
9242 running at anything other than 9600 bps. After that, use @code{target
9243 remote} to establish communications with @code{gdbserve.nlm}. Its
9244 argument is a device name (usually a serial device, like
9245 @file{/dev/ttyb}). For example:
9248 (@value{GDBP}) target remote /dev/ttyb
9252 communications with the server via serial line @file{/dev/ttyb}.
9256 @section Kernel Object Display
9258 @cindex kernel object display
9259 @cindex kernel object
9262 Some targets support kernel object display. Using this facility,
9263 @value{GDBN} communicates specially with the underlying operating system
9264 and can display information about operating system-level objects such as
9265 mutexes and other synchronization objects. Exactly which objects can be
9266 displayed is determined on a per-OS basis.
9268 Use the @code{set os} command to set the operating system. This tells
9269 @value{GDBN} which kernel object display module to initialize:
9275 If @code{set os} succeeds, @value{GDBN} will display some information
9276 about the operating system, and will create a new @code{info} command
9277 which can be used to query the target. The @code{info} command is named
9278 after the operating system:
9282 List of Cisco Kernel Objects
9284 any Any and all objects
9287 Further subcommands can be used to query about particular objects known
9290 There is currently no way to determine whether a given operating system
9291 is supported other than to try it.
9294 @node Configurations
9295 @chapter Configuration-Specific Information
9297 While nearly all @value{GDBN} commands are available for all native and
9298 cross versions of the debugger, there are some exceptions. This chapter
9299 describes things that are only available in certain configurations.
9301 There are three major categories of configurations: native
9302 configurations, where the host and target are the same, embedded
9303 operating system configurations, which are usually the same for several
9304 different processor architectures, and bare embedded processors, which
9305 are quite different from each other.
9310 * Embedded Processors::
9317 This section describes details specific to particular native
9322 * SVR4 Process Information:: SVR4 process information
9328 On HP-UX systems, if you refer to a function or variable name that
9329 begins with a dollar sign, @value{GDBN} searches for a user or system
9330 name first, before it searches for a convenience variable.
9332 @node SVR4 Process Information
9333 @subsection SVR4 process information
9336 @cindex process image
9338 Many versions of SVR4 provide a facility called @samp{/proc} that can be
9339 used to examine the image of a running process using file-system
9340 subroutines. If @value{GDBN} is configured for an operating system with
9341 this facility, the command @code{info proc} is available to report on
9342 several kinds of information about the process running your program.
9343 @code{info proc} works only on SVR4 systems that include the
9344 @code{procfs} code. This includes OSF/1 (Digital Unix), Solaris, Irix,
9345 and Unixware, but not HP-UX or Linux, for example.
9350 Summarize available information about the process.
9352 @kindex info proc mappings
9353 @item info proc mappings
9354 Report on the address ranges accessible in the program, with information
9355 on whether your program may read, write, or execute each range.
9357 @kindex info proc times
9358 @item info proc times
9359 Starting time, user CPU time, and system CPU time for your program and
9362 @kindex info proc id
9364 Report on the process IDs related to your program: its own process ID,
9365 the ID of its parent, the process group ID, and the session ID.
9367 @kindex info proc status
9368 @item info proc status
9369 General information on the state of the process. If the process is
9370 stopped, this report includes the reason for stopping, and any signal
9374 Show all the above information about the process.
9378 @section Embedded Operating Systems
9380 This section describes configurations involving the debugging of
9381 embedded operating systems that are available for several different
9385 * VxWorks:: Using @value{GDBN} with VxWorks
9388 @value{GDBN} includes the ability to debug programs running on
9389 various real-time operating systems.
9392 @subsection Using @value{GDBN} with VxWorks
9398 @kindex target vxworks
9399 @item target vxworks @var{machinename}
9400 A VxWorks system, attached via TCP/IP. The argument @var{machinename}
9401 is the target system's machine name or IP address.
9405 On VxWorks, @code{load} links @var{filename} dynamically on the
9406 current target system as well as adding its symbols in @value{GDBN}.
9408 @value{GDBN} enables developers to spawn and debug tasks running on networked
9409 VxWorks targets from a Unix host. Already-running tasks spawned from
9410 the VxWorks shell can also be debugged. @value{GDBN} uses code that runs on
9411 both the Unix host and on the VxWorks target. The program
9412 @code{gdb} is installed and executed on the Unix host. (It may be
9413 installed with the name @code{vxgdb}, to distinguish it from a
9414 @value{GDBN} for debugging programs on the host itself.)
9417 @item VxWorks-timeout @var{args}
9418 @kindex vxworks-timeout
9419 All VxWorks-based targets now support the option @code{vxworks-timeout}.
9420 This option is set by the user, and @var{args} represents the number of
9421 seconds @value{GDBN} waits for responses to rpc's. You might use this if
9422 your VxWorks target is a slow software simulator or is on the far side
9423 of a thin network line.
9426 The following information on connecting to VxWorks was current when
9427 this manual was produced; newer releases of VxWorks may use revised
9431 To use @value{GDBN} with VxWorks, you must rebuild your VxWorks kernel
9432 to include the remote debugging interface routines in the VxWorks
9433 library @file{rdb.a}. To do this, define @code{INCLUDE_RDB} in the
9434 VxWorks configuration file @file{configAll.h} and rebuild your VxWorks
9435 kernel. The resulting kernel contains @file{rdb.a}, and spawns the
9436 source debugging task @code{tRdbTask} when VxWorks is booted. For more
9437 information on configuring and remaking VxWorks, see the manufacturer's
9439 @c VxWorks, see the @cite{VxWorks Programmer's Guide}.
9441 Once you have included @file{rdb.a} in your VxWorks system image and set
9442 your Unix execution search path to find @value{GDBN}, you are ready to
9443 run @value{GDBN}. From your Unix host, run @code{gdb} (or @code{vxgdb},
9444 depending on your installation).
9446 @value{GDBN} comes up showing the prompt:
9453 * VxWorks Connection:: Connecting to VxWorks
9454 * VxWorks Download:: VxWorks download
9455 * VxWorks Attach:: Running tasks
9458 @node VxWorks Connection
9459 @subsubsection Connecting to VxWorks
9461 The @value{GDBN} command @code{target} lets you connect to a VxWorks target on the
9462 network. To connect to a target whose host name is ``@code{tt}'', type:
9465 (vxgdb) target vxworks tt
9469 @value{GDBN} displays messages like these:
9472 Attaching remote machine across net...
9477 @value{GDBN} then attempts to read the symbol tables of any object modules
9478 loaded into the VxWorks target since it was last booted. @value{GDBN} locates
9479 these files by searching the directories listed in the command search
9480 path (@pxref{Environment, ,Your program's environment}); if it fails
9481 to find an object file, it displays a message such as:
9484 prog.o: No such file or directory.
9487 When this happens, add the appropriate directory to the search path with
9488 the @value{GDBN} command @code{path}, and execute the @code{target}
9491 @node VxWorks Download
9492 @subsubsection VxWorks download
9494 @cindex download to VxWorks
9495 If you have connected to the VxWorks target and you want to debug an
9496 object that has not yet been loaded, you can use the @value{GDBN}
9497 @code{load} command to download a file from Unix to VxWorks
9498 incrementally. The object file given as an argument to the @code{load}
9499 command is actually opened twice: first by the VxWorks target in order
9500 to download the code, then by @value{GDBN} in order to read the symbol
9501 table. This can lead to problems if the current working directories on
9502 the two systems differ. If both systems have NFS mounted the same
9503 filesystems, you can avoid these problems by using absolute paths.
9504 Otherwise, it is simplest to set the working directory on both systems
9505 to the directory in which the object file resides, and then to reference
9506 the file by its name, without any path. For instance, a program
9507 @file{prog.o} may reside in @file{@var{vxpath}/vw/demo/rdb} in VxWorks
9508 and in @file{@var{hostpath}/vw/demo/rdb} on the host. To load this
9509 program, type this on VxWorks:
9512 -> cd "@var{vxpath}/vw/demo/rdb"
9515 Then, in @value{GDBN}, type:
9518 (vxgdb) cd @var{hostpath}/vw/demo/rdb
9522 @value{GDBN} displays a response similar to this:
9525 Reading symbol data from wherever/vw/demo/rdb/prog.o... done.
9528 You can also use the @code{load} command to reload an object module
9529 after editing and recompiling the corresponding source file. Note that
9530 this makes @value{GDBN} delete all currently-defined breakpoints,
9531 auto-displays, and convenience variables, and to clear the value
9532 history. (This is necessary in order to preserve the integrity of
9533 debugger data structures that reference the target system's symbol
9536 @node VxWorks Attach
9537 @subsubsection Running tasks
9539 @cindex running VxWorks tasks
9540 You can also attach to an existing task using the @code{attach} command as
9544 (vxgdb) attach @var{task}
9548 where @var{task} is the VxWorks hexadecimal task ID. The task can be running
9549 or suspended when you attach to it. Running tasks are suspended at
9550 the time of attachment.
9552 @node Embedded Processors
9553 @section Embedded Processors
9555 This section goes into details specific to particular embedded
9559 * A29K Embedded:: AMD A29K Embedded
9561 * H8/300:: Hitachi H8/300
9562 * H8/500:: Hitachi H8/500
9564 * M32R/D:: Mitsubishi M32R/D
9565 * M68K:: Motorola M68K
9566 * M88K:: Motorola M88K
9567 * MIPS Embedded:: MIPS Embedded
9568 * PA:: HP PA Embedded
9571 * Sparclet:: Tsqware Sparclet
9572 * Sparclite:: Fujitsu Sparclite
9573 * ST2000:: Tandem ST2000
9574 * Z8000:: Zilog Z8000
9578 @subsection AMD A29K Embedded
9583 * Comms (EB29K):: Communications setup
9584 * gdb-EB29K:: EB29K cross-debugging
9585 * Remote Log:: Remote log
9590 @kindex target adapt
9591 @item target adapt @var{dev}
9592 Adapt monitor for A29K.
9594 @kindex target amd-eb
9595 @item target amd-eb @var{dev} @var{speed} @var{PROG}
9597 Remote PC-resident AMD EB29K board, attached over serial lines.
9598 @var{dev} is the serial device, as for @code{target remote};
9599 @var{speed} allows you to specify the linespeed; and @var{PROG} is the
9600 name of the program to be debugged, as it appears to DOS on the PC.
9601 @xref{A29K EB29K, ,EBMON protocol for AMD29K}.
9606 @subsubsection A29K UDI
9609 @cindex AMD29K via UDI
9611 @value{GDBN} supports AMD's UDI (``Universal Debugger Interface'')
9612 protocol for debugging the a29k processor family. To use this
9613 configuration with AMD targets running the MiniMON monitor, you need the
9614 program @code{MONTIP}, available from AMD at no charge. You can also
9615 use @value{GDBN} with the UDI-conformant a29k simulator program
9616 @code{ISSTIP}, also available from AMD.
9619 @item target udi @var{keyword}
9621 Select the UDI interface to a remote a29k board or simulator, where
9622 @var{keyword} is an entry in the AMD configuration file @file{udi_soc}.
9623 This file contains keyword entries which specify parameters used to
9624 connect to a29k targets. If the @file{udi_soc} file is not in your
9625 working directory, you must set the environment variable @samp{UDICONF}
9630 @subsubsection EBMON protocol for AMD29K
9633 @cindex running 29K programs
9635 AMD distributes a 29K development board meant to fit in a PC, together
9636 with a DOS-hosted monitor program called @code{EBMON}. As a shorthand
9637 term, this development system is called the ``EB29K''. To use
9638 @value{GDBN} from a Unix system to run programs on the EB29K board, you
9639 must first connect a serial cable between the PC (which hosts the EB29K
9640 board) and a serial port on the Unix system. In the following, we
9641 assume you've hooked the cable between the PC's @file{COM1} port and
9642 @file{/dev/ttya} on the Unix system.
9645 @subsubsection Communications setup
9647 The next step is to set up the PC's port, by doing something like this
9651 C:\> MODE com1:9600,n,8,1,none
9655 This example---run on an MS DOS 4.0 system---sets the PC port to 9600
9656 bps, no parity, eight data bits, one stop bit, and no ``retry'' action;
9657 you must match the communications parameters when establishing the Unix
9658 end of the connection as well.
9659 @c FIXME: Who knows what this "no retry action" crud from the DOS manual may
9660 @c mean? It's optional; leave it out? ---doc@cygnus.com, 25feb91
9662 To give control of the PC to the Unix side of the serial line, type
9663 the following at the DOS console:
9670 (Later, if you wish to return control to the DOS console, you can use
9671 the command @code{CTTY con}---but you must send it over the device that
9672 had control, in our example over the @file{COM1} serial line).
9674 From the Unix host, use a communications program such as @code{tip} or
9675 @code{cu} to communicate with the PC; for example,
9678 cu -s 9600 -l /dev/ttya
9682 The @code{cu} options shown specify, respectively, the linespeed and the
9683 serial port to use. If you use @code{tip} instead, your command line
9684 may look something like the following:
9691 Your system may require a different name where we show
9692 @file{/dev/ttya} as the argument to @code{tip}. The communications
9693 parameters, including which port to use, are associated with the
9694 @code{tip} argument in the ``remote'' descriptions file---normally the
9695 system table @file{/etc/remote}.
9696 @c FIXME: What if anything needs doing to match the "n,8,1,none" part of
9697 @c the DOS side's comms setup? cu can support -o (odd
9698 @c parity), -e (even parity)---apparently no settings for no parity or
9699 @c for character size. Taken from stty maybe...? John points out tip
9700 @c can set these as internal variables, eg ~s parity=none; man stty
9701 @c suggests that it *might* work to stty these options with stdin or
9702 @c stdout redirected... ---doc@cygnus.com, 25feb91
9705 Using the @code{tip} or @code{cu} connection, change the DOS working
9706 directory to the directory containing a copy of your 29K program, then
9707 start the PC program @code{EBMON} (an EB29K control program supplied
9708 with your board by AMD). You should see an initial display from
9709 @code{EBMON} similar to the one that follows, ending with the
9710 @code{EBMON} prompt @samp{#}---
9715 G:\> CD \usr\joe\work29k
9717 G:\USR\JOE\WORK29K> EBMON
9718 Am29000 PC Coprocessor Board Monitor, version 3.0-18
9719 Copyright 1990 Advanced Micro Devices, Inc.
9720 Written by Gibbons and Associates, Inc.
9722 Enter '?' or 'H' for help
9724 PC Coprocessor Type = EB29K
9726 Memory Base = 0xd0000
9728 Data Memory Size = 2048KB
9729 Available I-RAM Range = 0x8000 to 0x1fffff
9730 Available D-RAM Range = 0x80002000 to 0x801fffff
9733 Register Stack Size = 0x800
9734 Memory Stack Size = 0x1800
9737 Am29027 Available = No
9738 Byte Write Available = Yes
9743 Then exit the @code{cu} or @code{tip} program (done in the example by
9744 typing @code{~.} at the @code{EBMON} prompt). @code{EBMON} keeps
9745 running, ready for @value{GDBN} to take over.
9747 For this example, we've assumed what is probably the most convenient
9748 way to make sure the same 29K program is on both the PC and the Unix
9749 system: a PC/NFS connection that establishes ``drive @code{G:}'' on the
9750 PC as a file system on the Unix host. If you do not have PC/NFS or
9751 something similar connecting the two systems, you must arrange some
9752 other way---perhaps floppy-disk transfer---of getting the 29K program
9753 from the Unix system to the PC; @value{GDBN} does @emph{not} download it over the
9757 @subsubsection EB29K cross-debugging
9759 Finally, @code{cd} to the directory containing an image of your 29K
9760 program on the Unix system, and start @value{GDBN}---specifying as argument the
9761 name of your 29K program:
9769 Now you can use the @code{target} command:
9772 target amd-eb /dev/ttya 9600 MYFOO
9773 @c FIXME: test above 'target amd-eb' as spelled, with caps! caps are meant to
9774 @c emphasize that this is the name as seen by DOS (since I think DOS is
9775 @c single-minded about case of letters). ---doc@cygnus.com, 25feb91
9779 In this example, we've assumed your program is in a file called
9780 @file{myfoo}. Note that the filename given as the last argument to
9781 @code{target amd-eb} should be the name of the program as it appears to DOS.
9782 In our example this is simply @code{MYFOO}, but in general it can include
9783 a DOS path, and depending on your transfer mechanism may not resemble
9784 the name on the Unix side.
9786 At this point, you can set any breakpoints you wish; when you are ready
9787 to see your program run on the 29K board, use the @value{GDBN} command
9790 To stop debugging the remote program, use the @value{GDBN} @code{detach}
9793 To return control of the PC to its console, use @code{tip} or @code{cu}
9794 once again, after your @value{GDBN} session has concluded, to attach to
9795 @code{EBMON}. You can then type the command @code{q} to shut down
9796 @code{EBMON}, returning control to the DOS command-line interpreter.
9797 Type @code{CTTY con} to return command input to the main DOS console,
9798 and type @kbd{~.} to leave @code{tip} or @code{cu}.
9801 @subsubsection Remote log
9803 @cindex log file for EB29K
9805 The @code{target amd-eb} command creates a file @file{eb.log} in the
9806 current working directory, to help debug problems with the connection.
9807 @file{eb.log} records all the output from @code{EBMON}, including echoes
9808 of the commands sent to it. Running @samp{tail -f} on this file in
9809 another window often helps to understand trouble with @code{EBMON}, or
9810 unexpected events on the PC side of the connection.
9818 @item target rdi @var{dev}
9819 ARM Angel monitor, via RDI library interface to ADP protocol. You may
9820 use this target to communicate with both boards running the Angel
9821 monitor, or with the EmbeddedICE JTAG debug device.
9824 @item target rdp @var{dev}
9830 @subsection Hitachi H8/300
9835 @item target hms @var{dev}
9836 A Hitachi SH, H8/300, or H8/500 board, attached via serial line to your host.
9837 Use special commands @code{device} and @code{speed} to control the serial
9838 line and the communications speed used.
9840 @kindex target e7000
9841 @item target e7000 @var{dev}
9842 E7000 emulator for Hitachi H8 and SH.
9846 @item target sh3 @var{dev}
9847 @item target sh3e @var{dev}
9848 Hitachi SH-3 and SH-3E target systems.
9852 @cindex download to H8/300 or H8/500
9853 @cindex H8/300 or H8/500 download
9854 @cindex download to Hitachi SH
9855 @cindex Hitachi SH download
9856 When you select remote debugging to a Hitachi SH, H8/300, or H8/500
9857 board, the @code{load} command downloads your program to the Hitachi
9858 board and also opens it as the current executable target for
9859 @value{GDBN} on your host (like the @code{file} command).
9861 @value{GDBN} needs to know these things to talk to your
9862 Hitachi SH, H8/300, or H8/500:
9866 that you want to use @samp{target hms}, the remote debugging interface
9867 for Hitachi microprocessors, or @samp{target e7000}, the in-circuit
9868 emulator for the Hitachi SH and the Hitachi 300H. (@samp{target hms} is
9869 the default when GDB is configured specifically for the Hitachi SH,
9873 what serial device connects your host to your Hitachi board (the first
9874 serial device available on your host is the default).
9877 what speed to use over the serial device.
9881 * Hitachi Boards:: Connecting to Hitachi boards.
9882 * Hitachi ICE:: Using the E7000 In-Circuit Emulator.
9883 * Hitachi Special:: Special @value{GDBN} commands for Hitachi micros.
9886 @node Hitachi Boards
9887 @subsubsection Connecting to Hitachi boards
9889 @c only for Unix hosts
9891 @cindex serial device, Hitachi micros
9892 Use the special @code{@value{GDBP}} command @samp{device @var{port}} if you
9893 need to explicitly set the serial device. The default @var{port} is the
9894 first available port on your host. This is only necessary on Unix
9895 hosts, where it is typically something like @file{/dev/ttya}.
9898 @cindex serial line speed, Hitachi micros
9899 @code{@value{GDBP}} has another special command to set the communications
9900 speed: @samp{speed @var{bps}}. This command also is only used from Unix
9901 hosts; on DOS hosts, set the line speed as usual from outside GDB with
9902 the DOS @kbd{mode} command (for instance, @w{@samp{mode
9903 com2:9600,n,8,1,p}} for a 9600 bps connection).
9905 The @samp{device} and @samp{speed} commands are available only when you
9906 use a Unix host to debug your Hitachi microprocessor programs. If you
9908 @value{GDBN} depends on an auxiliary terminate-and-stay-resident program
9909 called @code{asynctsr} to communicate with the development board
9910 through a PC serial port. You must also use the DOS @code{mode} command
9911 to set up the serial port on the DOS side.
9913 The following sample session illustrates the steps needed to start a
9914 program under @value{GDBN} control on an H8/300. The example uses a
9915 sample H8/300 program called @file{t.x}. The procedure is the same for
9916 the Hitachi SH and the H8/500.
9918 First hook up your development board. In this example, we use a
9919 board attached to serial port @code{COM2}; if you use a different serial
9920 port, substitute its name in the argument of the @code{mode} command.
9921 When you call @code{asynctsr}, the auxiliary comms program used by the
9922 degugger, you give it just the numeric part of the serial port's name;
9923 for example, @samp{asyncstr 2} below runs @code{asyncstr} on
9927 C:\H8300\TEST> asynctsr 2
9928 C:\H8300\TEST> mode com2:9600,n,8,1,p
9930 Resident portion of MODE loaded
9932 COM2: 9600, n, 8, 1, p
9937 @emph{Warning:} We have noticed a bug in PC-NFS that conflicts with
9938 @code{asynctsr}. If you also run PC-NFS on your DOS host, you may need to
9939 disable it, or even boot without it, to use @code{asynctsr} to control
9940 your development board.
9944 Now that serial communications are set up, and the development board is
9945 connected, you can start up @value{GDBN}. Call @code{@value{GDBP}} with
9946 the name of your program as the argument. @code{@value{GDBP}} prompts
9947 you, as usual, with the prompt @samp{(@value{GDBP})}. Use two special
9948 commands to begin your debugging session: @samp{target hms} to specify
9949 cross-debugging to the Hitachi board, and the @code{load} command to
9950 download your program to the board. @code{load} displays the names of
9951 the program's sections, and a @samp{*} for each 2K of data downloaded.
9952 (If you want to refresh @value{GDBN} data on symbols or on the
9953 executable file without downloading, use the @value{GDBN} commands
9954 @code{file} or @code{symbol-file}. These commands, and @code{load}
9955 itself, are described in @ref{Files,,Commands to specify files}.)
9958 (eg-C:\H8300\TEST) @value{GDBP} t.x
9959 GDB is free software and you are welcome to distribute copies
9960 of it under certain conditions; type "show copying" to see
9962 There is absolutely no warranty for GDB; type "show warranty"
9964 GDB @value{GDBVN}, Copyright 1992 Free Software Foundation, Inc...
9966 Connected to remote H8/300 HMS system.
9968 .text : 0x8000 .. 0xabde ***********
9969 .data : 0xabde .. 0xad30 *
9970 .stack : 0xf000 .. 0xf014 *
9973 At this point, you're ready to run or debug your program. From here on,
9974 you can use all the usual @value{GDBN} commands. The @code{break} command
9975 sets breakpoints; the @code{run} command starts your program;
9976 @code{print} or @code{x} display data; the @code{continue} command
9977 resumes execution after stopping at a breakpoint. You can use the
9978 @code{help} command at any time to find out more about @value{GDBN} commands.
9980 Remember, however, that @emph{operating system} facilities aren't
9981 available on your development board; for example, if your program hangs,
9982 you can't send an interrupt---but you can press the @sc{reset} switch!
9984 Use the @sc{reset} button on the development board
9987 to interrupt your program (don't use @kbd{ctl-C} on the DOS host---it has
9988 no way to pass an interrupt signal to the development board); and
9991 to return to the @value{GDBN} command prompt after your program finishes
9992 normally. The communications protocol provides no other way for @value{GDBN}
9993 to detect program completion.
9996 In either case, @value{GDBN} sees the effect of a @sc{reset} on the
9997 development board as a ``normal exit'' of your program.
10000 @subsubsection Using the E7000 in-circuit emulator
10002 @kindex target e7000
10003 You can use the E7000 in-circuit emulator to develop code for either the
10004 Hitachi SH or the H8/300H. Use one of these forms of the @samp{target
10005 e7000} command to connect @value{GDBN} to your E7000:
10008 @item target e7000 @var{port} @var{speed}
10009 Use this form if your E7000 is connected to a serial port. The
10010 @var{port} argument identifies what serial port to use (for example,
10011 @samp{com2}). The third argument is the line speed in bits per second
10012 (for example, @samp{9600}).
10014 @item target e7000 @var{hostname}
10015 If your E7000 is installed as a host on a TCP/IP network, you can just
10016 specify its hostname; @value{GDBN} uses @code{telnet} to connect.
10019 @node Hitachi Special
10020 @subsubsection Special @value{GDBN} commands for Hitachi micros
10022 Some @value{GDBN} commands are available only for the H8/300:
10026 @kindex set machine
10027 @kindex show machine
10028 @item set machine h8300
10029 @itemx set machine h8300h
10030 Condition @value{GDBN} for one of the two variants of the H8/300
10031 architecture with @samp{set machine}. You can use @samp{show machine}
10032 to check which variant is currently in effect.
10041 @kindex set memory @var{mod}
10042 @cindex memory models, H8/500
10043 @item set memory @var{mod}
10045 Specify which H8/500 memory model (@var{mod}) you are using with
10046 @samp{set memory}; check which memory model is in effect with @samp{show
10047 memory}. The accepted values for @var{mod} are @code{small},
10048 @code{big}, @code{medium}, and @code{compact}.
10053 @subsection Intel i960
10057 @kindex target mon960
10058 @item target mon960 @var{dev}
10059 MON960 monitor for Intel i960.
10061 @item target nindy @var{devicename}
10062 An Intel 960 board controlled by a Nindy Monitor. @var{devicename} is
10063 the name of the serial device to use for the connection, e.g.
10070 @dfn{Nindy} is a ROM Monitor program for Intel 960 target systems. When
10071 @value{GDBN} is configured to control a remote Intel 960 using Nindy, you can
10072 tell @value{GDBN} how to connect to the 960 in several ways:
10076 Through command line options specifying serial port, version of the
10077 Nindy protocol, and communications speed;
10080 By responding to a prompt on startup;
10083 By using the @code{target} command at any point during your @value{GDBN}
10084 session. @xref{Target Commands, ,Commands for managing targets}.
10086 @kindex target nindy
10087 @item target nindy @var{devicename}
10088 An Intel 960 board controlled by a Nindy Monitor. @var{devicename} is
10089 the name of the serial device to use for the connection, e.g.
10094 @cindex download to Nindy-960
10095 With the Nindy interface to an Intel 960 board, @code{load}
10096 downloads @var{filename} to the 960 as well as adding its symbols in
10100 * Nindy Startup:: Startup with Nindy
10101 * Nindy Options:: Options for Nindy
10102 * Nindy Reset:: Nindy reset command
10105 @node Nindy Startup
10106 @subsubsection Startup with Nindy
10108 If you simply start @code{@value{GDBP}} without using any command-line
10109 options, you are prompted for what serial port to use, @emph{before} you
10110 reach the ordinary @value{GDBN} prompt:
10113 Attach /dev/ttyNN -- specify NN, or "quit" to quit:
10117 Respond to the prompt with whatever suffix (after @samp{/dev/tty})
10118 identifies the serial port you want to use. You can, if you choose,
10119 simply start up with no Nindy connection by responding to the prompt
10120 with an empty line. If you do this and later wish to attach to Nindy,
10121 use @code{target} (@pxref{Target Commands, ,Commands for managing targets}).
10123 @node Nindy Options
10124 @subsubsection Options for Nindy
10126 These are the startup options for beginning your @value{GDBN} session with a
10127 Nindy-960 board attached:
10130 @item -r @var{port}
10131 Specify the serial port name of a serial interface to be used to connect
10132 to the target system. This option is only available when @value{GDBN} is
10133 configured for the Intel 960 target architecture. You may specify
10134 @var{port} as any of: a full pathname (e.g. @samp{-r /dev/ttya}), a
10135 device name in @file{/dev} (e.g. @samp{-r ttya}), or simply the unique
10136 suffix for a specific @code{tty} (e.g. @samp{-r a}).
10139 (An uppercase letter ``O'', not a zero.) Specify that @value{GDBN} should use
10140 the ``old'' Nindy monitor protocol to connect to the target system.
10141 This option is only available when @value{GDBN} is configured for the Intel 960
10142 target architecture.
10145 @emph{Warning:} if you specify @samp{-O}, but are actually trying to
10146 connect to a target system that expects the newer protocol, the connection
10147 fails, appearing to be a speed mismatch. @value{GDBN} repeatedly
10148 attempts to reconnect at several different line speeds. You can abort
10149 this process with an interrupt.
10153 Specify that @value{GDBN} should first send a @code{BREAK} signal to the target
10154 system, in an attempt to reset it, before connecting to a Nindy target.
10157 @emph{Warning:} Many target systems do not have the hardware that this
10158 requires; it only works with a few boards.
10162 The standard @samp{-b} option controls the line speed used on the serial
10167 @subsubsection Nindy reset command
10172 For a Nindy target, this command sends a ``break'' to the remote target
10173 system; this is only useful if the target has been equipped with a
10174 circuit to perform a hard reset (or some other interesting action) when
10175 a break is detected.
10180 @subsection Mitsubishi M32R/D
10184 @kindex target m32r
10185 @item target m32r @var{dev}
10186 Mitsubishi M32R/D ROM monitor.
10193 The Motorola m68k configuration includes ColdFire support, and
10194 target command for the following ROM monitors.
10198 @kindex target abug
10199 @item target abug @var{dev}
10200 ABug ROM monitor for M68K.
10202 @kindex target cpu32bug
10203 @item target cpu32bug @var{dev}
10204 CPU32BUG monitor, running on a CPU32 (M68K) board.
10206 @kindex target dbug
10207 @item target dbug @var{dev}
10208 dBUG ROM monitor for Motorola ColdFire.
10211 @item target est @var{dev}
10212 EST-300 ICE monitor, running on a CPU32 (M68K) board.
10214 @kindex target rom68k
10215 @item target rom68k @var{dev}
10216 ROM 68K monitor, running on an M68K IDP board.
10220 If @value{GDBN} is configured with @code{m68*-ericsson-*}, it will
10221 instead have only a single special target command:
10225 @kindex target es1800
10226 @item target es1800 @var{dev}
10227 ES-1800 emulator for M68K.
10235 @kindex target rombug
10236 @item target rombug @var{dev}
10237 ROMBUG ROM monitor for OS/9000.
10247 @item target bug @var{dev}
10248 BUG monitor, running on a MVME187 (m88k) board.
10252 @node MIPS Embedded
10253 @subsection MIPS Embedded
10255 @cindex MIPS boards
10256 @value{GDBN} can use the MIPS remote debugging protocol to talk to a
10257 MIPS board attached to a serial line. This is available when
10258 you configure @value{GDBN} with @samp{--target=mips-idt-ecoff}.
10261 Use these @value{GDBN} commands to specify the connection to your target board:
10264 @item target mips @var{port}
10265 @kindex target mips @var{port}
10266 To run a program on the board, start up @code{@value{GDBP}} with the
10267 name of your program as the argument. To connect to the board, use the
10268 command @samp{target mips @var{port}}, where @var{port} is the name of
10269 the serial port connected to the board. If the program has not already
10270 been downloaded to the board, you may use the @code{load} command to
10271 download it. You can then use all the usual @value{GDBN} commands.
10273 For example, this sequence connects to the target board through a serial
10274 port, and loads and runs a program called @var{prog} through the
10278 host$ @value{GDBP} @var{prog}
10279 GDB is free software and @dots{}
10280 (gdb) target mips /dev/ttyb
10281 (gdb) load @var{prog}
10285 @item target mips @var{hostname}:@var{portnumber}
10286 On some @value{GDBN} host configurations, you can specify a TCP
10287 connection (for instance, to a serial line managed by a terminal
10288 concentrator) instead of a serial port, using the syntax
10289 @samp{@var{hostname}:@var{portnumber}}.
10291 @item target pmon @var{port}
10292 @kindex target pmon @var{port}
10295 @item target ddb @var{port}
10296 @kindex target ddb @var{port}
10297 NEC's DDB variant of PMON for Vr4300.
10299 @item target lsi @var{port}
10300 @kindex target lsi @var{port}
10301 LSI variant of PMON.
10303 @kindex target r3900
10304 @item target r3900 @var{dev}
10305 Densan DVE-R3900 ROM monitor for Toshiba R3900 Mips.
10307 @kindex target array
10308 @item target array @var{dev}
10309 Array Tech LSI33K RAID controller board.
10315 @value{GDBN} also supports these special commands for MIPS targets:
10318 @item set processor @var{args}
10319 @itemx show processor
10320 @kindex set processor @var{args}
10321 @kindex show processor
10322 Use the @code{set processor} command to set the type of MIPS
10323 processor when you want to access processor-type-specific registers.
10324 For example, @code{set processor @var{r3041}} tells @value{GDBN}
10325 to use the CPO registers appropriate for the 3041 chip.
10326 Use the @code{show processor} command to see what MIPS processor @value{GDBN}
10327 is using. Use the @code{info reg} command to see what registers
10328 @value{GDBN} is using.
10330 @item set mipsfpu double
10331 @itemx set mipsfpu single
10332 @itemx set mipsfpu none
10333 @itemx show mipsfpu
10334 @kindex set mipsfpu
10335 @kindex show mipsfpu
10336 @cindex MIPS remote floating point
10337 @cindex floating point, MIPS remote
10338 If your target board does not support the MIPS floating point
10339 coprocessor, you should use the command @samp{set mipsfpu none} (if you
10340 need this, you may wish to put the command in your @value{GDBINIT}
10341 file). This tells @value{GDBN} how to find the return value of
10342 functions which return floating point values. It also allows
10343 @value{GDBN} to avoid saving the floating point registers when calling
10344 functions on the board. If you are using a floating point coprocessor
10345 with only single precision floating point support, as on the @sc{r4650}
10346 processor, use the command @samp{set mipsfpu single}. The default
10347 double precision floating point coprocessor may be selected using
10348 @samp{set mipsfpu double}.
10350 In previous versions the only choices were double precision or no
10351 floating point, so @samp{set mipsfpu on} will select double precision
10352 and @samp{set mipsfpu off} will select no floating point.
10354 As usual, you can inquire about the @code{mipsfpu} variable with
10355 @samp{show mipsfpu}.
10357 @item set remotedebug @var{n}
10358 @itemx show remotedebug
10359 @kindex set remotedebug
10360 @kindex show remotedebug
10361 @cindex @code{remotedebug}, MIPS protocol
10362 @cindex MIPS @code{remotedebug} protocol
10363 @c FIXME! For this to be useful, you must know something about the MIPS
10364 @c FIXME...protocol. Where is it described?
10365 You can see some debugging information about communications with the board
10366 by setting the @code{remotedebug} variable. If you set it to @code{1} using
10367 @samp{set remotedebug 1}, every packet is displayed. If you set it
10368 to @code{2}, every character is displayed. You can check the current value
10369 at any time with the command @samp{show remotedebug}.
10371 @item set timeout @var{seconds}
10372 @itemx set retransmit-timeout @var{seconds}
10373 @itemx show timeout
10374 @itemx show retransmit-timeout
10375 @cindex @code{timeout}, MIPS protocol
10376 @cindex @code{retransmit-timeout}, MIPS protocol
10377 @kindex set timeout
10378 @kindex show timeout
10379 @kindex set retransmit-timeout
10380 @kindex show retransmit-timeout
10381 You can control the timeout used while waiting for a packet, in the MIPS
10382 remote protocol, with the @code{set timeout @var{seconds}} command. The
10383 default is 5 seconds. Similarly, you can control the timeout used while
10384 waiting for an acknowledgement of a packet with the @code{set
10385 retransmit-timeout @var{seconds}} command. The default is 3 seconds.
10386 You can inspect both values with @code{show timeout} and @code{show
10387 retransmit-timeout}. (These commands are @emph{only} available when
10388 @value{GDBN} is configured for @samp{--target=mips-idt-ecoff}.)
10390 The timeout set by @code{set timeout} does not apply when @value{GDBN}
10391 is waiting for your program to stop. In that case, @value{GDBN} waits
10392 forever because it has no way of knowing how long the program is going
10393 to run before stopping.
10397 @subsection PowerPC
10401 @kindex target dink32
10402 @item target dink32 @var{dev}
10403 DINK32 ROM monitor.
10405 @kindex target ppcbug
10406 @item target ppcbug @var{dev}
10407 @kindex target ppcbug1
10408 @item target ppcbug1 @var{dev}
10409 PPCBUG ROM monitor for PowerPC.
10412 @item target sds @var{dev}
10413 SDS monitor, running on a PowerPC board (such as Motorola's ADS).
10418 @subsection HP PA Embedded
10422 @kindex target op50n
10423 @item target op50n @var{dev}
10424 OP50N monitor, running on an OKI HPPA board.
10426 @kindex target w89k
10427 @item target w89k @var{dev}
10428 W89K monitor, running on a Winbond HPPA board.
10433 @subsection Hitachi SH
10438 @item target hms @var{dev}
10439 A Hitachi SH board attached via serial line to your host. Use special
10440 commands @code{device} and @code{speed} to control the serial line and
10441 the communications speed used.
10443 @kindex target e7000
10444 @item target e7000 @var{dev}
10445 E7000 emulator for Hitachi SH.
10448 @kindex target sh3e
10449 @item target sh3 @var{dev}
10450 @item target sh3e @var{dev}
10451 Hitachi SH-3 and SH-3E target systems.
10456 @subsection Tsqware Sparclet
10460 @value{GDBN} enables developers to debug tasks running on
10461 Sparclet targets from a Unix host.
10462 @value{GDBN} uses code that runs on
10463 both the Unix host and on the Sparclet target. The program
10464 @code{gdb} is installed and executed on the Unix host.
10467 @item timeout @var{args}
10468 @kindex remotetimeout
10469 @value{GDBN} now supports the option @code{remotetimeout}.
10470 This option is set by the user, and @var{args} represents the number of
10471 seconds @value{GDBN} waits for responses.
10475 When compiling for debugging, include the options "-g" to get debug
10476 information and "-Ttext" to relocate the program to where you wish to
10477 load it on the target. You may also want to add the options "-n" or
10478 "-N" in order to reduce the size of the sections.
10481 sparclet-aout-gcc prog.c -Ttext 0x12010000 -g -o prog -N
10484 You can use objdump to verify that the addresses are what you intended.
10487 sparclet-aout-objdump --headers --syms prog
10492 your Unix execution search path to find @value{GDBN}, you are ready to
10493 run @value{GDBN}. From your Unix host, run @code{gdb}
10494 (or @code{sparclet-aout-gdb}, depending on your installation).
10496 @value{GDBN} comes up showing the prompt:
10503 * Sparclet File:: Setting the file to debug
10504 * Sparclet Connection:: Connecting to Sparclet
10505 * Sparclet Download:: Sparclet download
10506 * Sparclet Execution:: Running and debugging
10509 @node Sparclet File
10510 @subsubsection Setting file to debug
10512 The @value{GDBN} command @code{file} lets you choose with program to debug.
10515 (gdbslet) file prog
10519 @value{GDBN} then attempts to read the symbol table of @file{prog}.
10520 @value{GDBN} locates
10521 the file by searching the directories listed in the command search
10523 If the file was compiled with debug information (option "-g"), source
10524 files will be searched as well.
10525 @value{GDBN} locates
10526 the source files by searching the directories listed in the directory search
10527 path (@pxref{Environment, ,Your program's environment}).
10529 to find a file, it displays a message such as:
10532 prog: No such file or directory.
10535 When this happens, add the appropriate directories to the search paths with
10536 the @value{GDBN} commands @code{path} and @code{dir}, and execute the
10537 @code{target} command again.
10539 @node Sparclet Connection
10540 @subsubsection Connecting to Sparclet
10542 The @value{GDBN} command @code{target} lets you connect to a Sparclet target.
10543 To connect to a target on serial port ``@code{ttya}'', type:
10546 (gdbslet) target sparclet /dev/ttya
10547 Remote target sparclet connected to /dev/ttya
10548 main () at ../prog.c:3
10552 @value{GDBN} displays messages like these:
10558 @node Sparclet Download
10559 @subsubsection Sparclet download
10561 @cindex download to Sparclet
10562 Once connected to the Sparclet target,
10563 you can use the @value{GDBN}
10564 @code{load} command to download the file from the host to the target.
10565 The file name and load offset should be given as arguments to the @code{load}
10567 Since the file format is aout, the program must be loaded to the starting
10568 address. You can use objdump to find out what this value is. The load
10569 offset is an offset which is added to the VMA (virtual memory address)
10570 of each of the file's sections.
10571 For instance, if the program
10572 @file{prog} was linked to text address 0x1201000, with data at 0x12010160
10573 and bss at 0x12010170, in @value{GDBN}, type:
10576 (gdbslet) load prog 0x12010000
10577 Loading section .text, size 0xdb0 vma 0x12010000
10580 If the code is loaded at a different address then what the program was linked
10581 to, you may need to use the @code{section} and @code{add-symbol-file} commands
10582 to tell @value{GDBN} where to map the symbol table.
10584 @node Sparclet Execution
10585 @subsubsection Running and debugging
10587 @cindex running and debugging Sparclet programs
10588 You can now begin debugging the task using @value{GDBN}'s execution control
10589 commands, @code{b}, @code{step}, @code{run}, etc. See the @value{GDBN}
10590 manual for the list of commands.
10594 Breakpoint 1 at 0x12010000: file prog.c, line 3.
10596 Starting program: prog
10597 Breakpoint 1, main (argc=1, argv=0xeffff21c) at prog.c:3
10598 3 char *symarg = 0;
10600 4 char *execarg = "hello!";
10605 @subsection Fujitsu Sparclite
10609 @kindex target sparclite
10610 @item target sparclite @var{dev}
10611 Fujitsu sparclite boards, used only for the purpose of loading.
10612 You must use an additional command to debug the program.
10613 For example: target remote @var{dev} using @value{GDBN} standard
10619 @subsection Tandem ST2000
10621 GDB may be used with a Tandem ST2000 phone switch, running Tandem's
10624 To connect your ST2000 to the host system, see the manufacturer's
10625 manual. Once the ST2000 is physically attached, you can run:
10628 target st2000 @var{dev} @var{speed}
10632 to establish it as your debugging environment. @var{dev} is normally
10633 the name of a serial device, such as @file{/dev/ttya}, connected to the
10634 ST2000 via a serial line. You can instead specify @var{dev} as a TCP
10635 connection (for example, to a serial line attached via a terminal
10636 concentrator) using the syntax @code{@var{hostname}:@var{portnumber}}.
10638 The @code{load} and @code{attach} commands are @emph{not} defined for
10639 this target; you must load your program into the ST2000 as you normally
10640 would for standalone operation. @value{GDBN} reads debugging information
10641 (such as symbols) from a separate, debugging version of the program
10642 available on your host computer.
10643 @c FIXME!! This is terribly vague; what little content is here is
10644 @c basically hearsay.
10646 @cindex ST2000 auxiliary commands
10647 These auxiliary @value{GDBN} commands are available to help you with the ST2000
10651 @item st2000 @var{command}
10652 @kindex st2000 @var{cmd}
10653 @cindex STDBUG commands (ST2000)
10654 @cindex commands to STDBUG (ST2000)
10655 Send a @var{command} to the STDBUG monitor. See the manufacturer's
10656 manual for available commands.
10659 @cindex connect (to STDBUG)
10660 Connect the controlling terminal to the STDBUG command monitor. When
10661 you are done interacting with STDBUG, typing either of two character
10662 sequences gets you back to the @value{GDBN} command prompt:
10663 @kbd{@key{RET}~.} (Return, followed by tilde and period) or
10664 @kbd{@key{RET}~@key{C-d}} (Return, followed by tilde and control-D).
10668 @subsection Zilog Z8000
10671 @cindex simulator, Z8000
10672 @cindex Zilog Z8000 simulator
10674 When configured for debugging Zilog Z8000 targets, @value{GDBN} includes
10677 For the Z8000 family, @samp{target sim} simulates either the Z8002 (the
10678 unsegmented variant of the Z8000 architecture) or the Z8001 (the
10679 segmented variant). The simulator recognizes which architecture is
10680 appropriate by inspecting the object code.
10683 @item target sim @var{args}
10686 Debug programs on a simulated CPU. If the simulator supports setup
10687 options, specify them via @var{args}.
10691 After specifying this target, you can debug programs for the simulated
10692 CPU in the same style as programs for your host computer; use the
10693 @code{file} command to load a new program image, the @code{run} command
10694 to run your program, and so on.
10696 As well as making available all the usual machine registers (see
10697 @code{info reg}), the Z8000 simulator provides three additional items
10698 of information as specially named registers:
10703 Counts clock-ticks in the simulator.
10706 Counts instructions run in the simulator.
10709 Execution time in 60ths of a second.
10713 You can refer to these values in @value{GDBN} expressions with the usual
10714 conventions; for example, @w{@samp{b fputc if $cycles>5000}} sets a
10715 conditional breakpoint that suspends only after at least 5000
10716 simulated clock ticks.
10718 @node Architectures
10719 @section Architectures
10721 This section describes characteristics of architectures that affect
10722 all uses of GDB with this architecture, both native and cross.
10735 @kindex set rstack_high_address
10736 @cindex AMD 29K register stack
10737 @cindex register stack, AMD29K
10738 @item set rstack_high_address @var{address}
10739 On AMD 29000 family processors, registers are saved in a separate
10740 ``register stack''. There is no way for @value{GDBN} to determine the
10741 extent of this stack. Normally, @value{GDBN} just assumes that the
10742 stack is ``large enough''. This may result in @value{GDBN} referencing
10743 memory locations that do not exist. If necessary, you can get around
10744 this problem by specifying the ending address of the register stack with
10745 the @code{set rstack_high_address} command. The argument should be an
10746 address, which you probably want to precede with @samp{0x} to specify in
10749 @kindex show rstack_high_address
10750 @item show rstack_high_address
10751 Display the current limit of the register stack, on AMD 29000 family
10759 See the following section.
10764 @cindex stack on Alpha
10765 @cindex stack on MIPS
10766 @cindex Alpha stack
10768 Alpha- and MIPS-based computers use an unusual stack frame, which
10769 sometimes requires @value{GDBN} to search backward in the object code to
10770 find the beginning of a function.
10772 @cindex response time, MIPS debugging
10773 To improve response time (especially for embedded applications, where
10774 @value{GDBN} may be restricted to a slow serial line for this search)
10775 you may want to limit the size of this search, using one of these
10779 @cindex @code{heuristic-fence-post} (Alpha,MIPS)
10780 @item set heuristic-fence-post @var{limit}
10781 Restrict @value{GDBN} to examining at most @var{limit} bytes in its
10782 search for the beginning of a function. A value of @var{0} (the
10783 default) means there is no limit. However, except for @var{0}, the
10784 larger the limit the more bytes @code{heuristic-fence-post} must search
10785 and therefore the longer it takes to run.
10787 @item show heuristic-fence-post
10788 Display the current limit.
10792 These commands are available @emph{only} when @value{GDBN} is configured
10793 for debugging programs on Alpha or MIPS processors.
10796 @node Controlling GDB
10797 @chapter Controlling @value{GDBN}
10799 You can alter the way @value{GDBN} interacts with you by using the
10800 @code{set} command. For commands controlling how @value{GDBN} displays
10801 data, @pxref{Print Settings, ,Print settings}; other settings are
10806 * Editing:: Command editing
10807 * History:: Command history
10808 * Screen Size:: Screen size
10809 * Numbers:: Numbers
10810 * Messages/Warnings:: Optional warnings and messages
10818 @value{GDBN} indicates its readiness to read a command by printing a string
10819 called the @dfn{prompt}. This string is normally @samp{(@value{GDBP})}. You
10820 can change the prompt string with the @code{set prompt} command. For
10821 instance, when debugging @value{GDBN} with @value{GDBN}, it is useful to change
10822 the prompt in one of the @value{GDBN} sessions so that you can always tell
10823 which one you are talking to.
10825 @emph{Note:} @code{set prompt} no longer adds a space for you after the
10826 prompt you set. This allows you to set a prompt which ends in a space
10827 or a prompt that does not.
10831 @item set prompt @var{newprompt}
10832 Directs @value{GDBN} to use @var{newprompt} as its prompt string henceforth.
10834 @kindex show prompt
10836 Prints a line of the form: @samp{Gdb's prompt is: @var{your-prompt}}
10840 @section Command editing
10842 @cindex command line editing
10844 @value{GDBN} reads its input commands via the @dfn{readline} interface. This
10845 @sc{gnu} library provides consistent behavior for programs which provide a
10846 command line interface to the user. Advantages are @sc{gnu} Emacs-style
10847 or @dfn{vi}-style inline editing of commands, @code{csh}-like history
10848 substitution, and a storage and recall of command history across
10849 debugging sessions.
10851 You may control the behavior of command line editing in @value{GDBN} with the
10852 command @code{set}.
10855 @kindex set editing
10858 @itemx set editing on
10859 Enable command line editing (enabled by default).
10861 @item set editing off
10862 Disable command line editing.
10864 @kindex show editing
10866 Show whether command line editing is enabled.
10870 @section Command history
10872 @value{GDBN} can keep track of the commands you type during your
10873 debugging sessions, so that you can be certain of precisely what
10874 happened. Use these commands to manage the @value{GDBN} command
10878 @cindex history substitution
10879 @cindex history file
10880 @kindex set history filename
10881 @kindex GDBHISTFILE
10882 @item set history filename @var{fname}
10883 Set the name of the @value{GDBN} command history file to @var{fname}.
10884 This is the file where @value{GDBN} reads an initial command history
10885 list, and where it writes the command history from this session when it
10886 exits. You can access this list through history expansion or through
10887 the history command editing characters listed below. This file defaults
10888 to the value of the environment variable @code{GDBHISTFILE}, or to
10889 @file{./.gdb_history} if this variable is not set.
10891 @cindex history save
10892 @kindex set history save
10893 @item set history save
10894 @itemx set history save on
10895 Record command history in a file, whose name may be specified with the
10896 @code{set history filename} command. By default, this option is disabled.
10898 @item set history save off
10899 Stop recording command history in a file.
10901 @cindex history size
10902 @kindex set history size
10903 @item set history size @var{size}
10904 Set the number of commands which @value{GDBN} keeps in its history list.
10905 This defaults to the value of the environment variable
10906 @code{HISTSIZE}, or to 256 if this variable is not set.
10909 @cindex history expansion
10910 History expansion assigns special meaning to the character @kbd{!}.
10911 @ifset have-readline-appendices
10912 @xref{Event Designators}.
10915 Since @kbd{!} is also the logical not operator in C, history expansion
10916 is off by default. If you decide to enable history expansion with the
10917 @code{set history expansion on} command, you may sometimes need to
10918 follow @kbd{!} (when it is used as logical not, in an expression) with
10919 a space or a tab to prevent it from being expanded. The readline
10920 history facilities do not attempt substitution on the strings
10921 @kbd{!=} and @kbd{!(}, even when history expansion is enabled.
10923 The commands to control history expansion are:
10926 @kindex set history expansion
10927 @item set history expansion on
10928 @itemx set history expansion
10929 Enable history expansion. History expansion is off by default.
10931 @item set history expansion off
10932 Disable history expansion.
10934 The readline code comes with more complete documentation of
10935 editing and history expansion features. Users unfamiliar with @sc{gnu} Emacs
10936 or @code{vi} may wish to read it.
10937 @ifset have-readline-appendices
10938 @xref{Command Line Editing}.
10942 @kindex show history
10944 @itemx show history filename
10945 @itemx show history save
10946 @itemx show history size
10947 @itemx show history expansion
10948 These commands display the state of the @value{GDBN} history parameters.
10949 @code{show history} by itself displays all four states.
10954 @kindex show commands
10955 @item show commands
10956 Display the last ten commands in the command history.
10958 @item show commands @var{n}
10959 Print ten commands centered on command number @var{n}.
10961 @item show commands +
10962 Print ten commands just after the commands last printed.
10966 @section Screen size
10967 @cindex size of screen
10968 @cindex pauses in output
10970 Certain commands to @value{GDBN} may produce large amounts of
10971 information output to the screen. To help you read all of it,
10972 @value{GDBN} pauses and asks you for input at the end of each page of
10973 output. Type @key{RET} when you want to continue the output, or @kbd{q}
10974 to discard the remaining output. Also, the screen width setting
10975 determines when to wrap lines of output. Depending on what is being
10976 printed, @value{GDBN} tries to break the line at a readable place,
10977 rather than simply letting it overflow onto the following line.
10979 Normally @value{GDBN} knows the size of the screen from the termcap data base
10980 together with the value of the @code{TERM} environment variable and the
10981 @code{stty rows} and @code{stty cols} settings. If this is not correct,
10982 you can override it with the @code{set height} and @code{set
10989 @kindex show height
10990 @item set height @var{lpp}
10992 @itemx set width @var{cpl}
10994 These @code{set} commands specify a screen height of @var{lpp} lines and
10995 a screen width of @var{cpl} characters. The associated @code{show}
10996 commands display the current settings.
10998 If you specify a height of zero lines, @value{GDBN} does not pause during
10999 output no matter how long the output is. This is useful if output is to a
11000 file or to an editor buffer.
11002 Likewise, you can specify @samp{set width 0} to prevent @value{GDBN}
11003 from wrapping its output.
11008 @cindex number representation
11009 @cindex entering numbers
11011 You can always enter numbers in octal, decimal, or hexadecimal in @value{GDBN} by
11012 the usual conventions: octal numbers begin with @samp{0}, decimal
11013 numbers end with @samp{.}, and hexadecimal numbers begin with @samp{0x}.
11014 Numbers that begin with none of these are, by default, entered in base
11015 10; likewise, the default display for numbers---when no particular
11016 format is specified---is base 10. You can change the default base for
11017 both input and output with the @code{set radix} command.
11020 @kindex set input-radix
11021 @item set input-radix @var{base}
11022 Set the default base for numeric input. Supported choices
11023 for @var{base} are decimal 8, 10, or 16. @var{base} must itself be
11024 specified either unambiguously or using the current default radix; for
11034 sets the base to decimal. On the other hand, @samp{set radix 10}
11035 leaves the radix unchanged no matter what it was.
11037 @kindex set output-radix
11038 @item set output-radix @var{base}
11039 Set the default base for numeric display. Supported choices
11040 for @var{base} are decimal 8, 10, or 16. @var{base} must itself be
11041 specified either unambiguously or using the current default radix.
11043 @kindex show input-radix
11044 @item show input-radix
11045 Display the current default base for numeric input.
11047 @kindex show output-radix
11048 @item show output-radix
11049 Display the current default base for numeric display.
11052 @node Messages/Warnings
11053 @section Optional warnings and messages
11055 By default, @value{GDBN} is silent about its inner workings. If you are running
11056 on a slow machine, you may want to use the @code{set verbose} command.
11057 This makes @value{GDBN} tell you when it does a lengthy internal operation, so
11058 you will not think it has crashed.
11060 Currently, the messages controlled by @code{set verbose} are those
11061 which announce that the symbol table for a source file is being read;
11062 see @code{symbol-file} in @ref{Files, ,Commands to specify files}.
11065 @kindex set verbose
11066 @item set verbose on
11067 Enables @value{GDBN} output of certain informational messages.
11069 @item set verbose off
11070 Disables @value{GDBN} output of certain informational messages.
11072 @kindex show verbose
11074 Displays whether @code{set verbose} is on or off.
11077 By default, if @value{GDBN} encounters bugs in the symbol table of an object
11078 file, it is silent; but if you are debugging a compiler, you may find
11079 this information useful (@pxref{Symbol Errors, ,Errors reading symbol files}).
11082 @kindex set complaints
11083 @item set complaints @var{limit}
11084 Permits @value{GDBN} to output @var{limit} complaints about each type of unusual
11085 symbols before becoming silent about the problem. Set @var{limit} to
11086 zero to suppress all complaints; set it to a large number to prevent
11087 complaints from being suppressed.
11089 @kindex show complaints
11090 @item show complaints
11091 Displays how many symbol complaints @value{GDBN} is permitted to produce.
11094 By default, @value{GDBN} is cautious, and asks what sometimes seems to be a
11095 lot of stupid questions to confirm certain commands. For example, if
11096 you try to run a program which is already running:
11100 The program being debugged has been started already.
11101 Start it from the beginning? (y or n)
11104 If you are willing to unflinchingly face the consequences of your own
11105 commands, you can disable this ``feature'':
11108 @kindex set confirm
11110 @cindex confirmation
11111 @cindex stupid questions
11112 @item set confirm off
11113 Disables confirmation requests.
11115 @item set confirm on
11116 Enables confirmation requests (the default).
11118 @kindex show confirm
11120 Displays state of confirmation requests.
11124 @chapter Canned Sequences of Commands
11126 Aside from breakpoint commands (@pxref{Break Commands, ,Breakpoint
11127 command lists}), @value{GDBN} provides two ways to store sequences of commands
11128 for execution as a unit: user-defined commands and command files.
11131 * Define:: User-defined commands
11132 * Hooks:: User-defined command hooks
11133 * Command Files:: Command files
11134 * Output:: Commands for controlled output
11138 @section User-defined commands
11140 @cindex user-defined command
11141 A @dfn{user-defined command} is a sequence of @value{GDBN} commands to which
11142 you assign a new name as a command. This is done with the @code{define}
11143 command. User commands may accept up to 10 arguments separated by whitespace.
11144 Arguments are accessed within the user command via @var{$arg0@dots{}$arg9}.
11149 print $arg0 + $arg1 + $arg2
11152 @noindent To execute the command use:
11158 @noindent This defines the command @code{adder}, which prints the sum of
11159 its three arguments. Note the arguments are text substitutions, so they may
11160 reference variables, use complex expressions, or even perform inferior
11165 @item define @var{commandname}
11166 Define a command named @var{commandname}. If there is already a command
11167 by that name, you are asked to confirm that you want to redefine it.
11169 The definition of the command is made up of other @value{GDBN} command lines,
11170 which are given following the @code{define} command. The end of these
11171 commands is marked by a line containing @code{end}.
11176 Takes a single argument, which is an expression to evaluate.
11177 It is followed by a series of commands that are executed
11178 only if the expression is true (nonzero).
11179 There can then optionally be a line @code{else}, followed
11180 by a series of commands that are only executed if the expression
11181 was false. The end of the list is marked by a line containing @code{end}.
11185 The syntax is similar to @code{if}: the command takes a single argument,
11186 which is an expression to evaluate, and must be followed by the commands to
11187 execute, one per line, terminated by an @code{end}.
11188 The commands are executed repeatedly as long as the expression
11192 @item document @var{commandname}
11193 Document the user-defined command @var{commandname}, so that it can be
11194 accessed by @code{help}. The command @var{commandname} must already be
11195 defined. This command reads lines of documentation just as @code{define}
11196 reads the lines of the command definition, ending with @code{end}.
11197 After the @code{document} command is finished, @code{help} on command
11198 @var{commandname} displays the documentation you have written.
11200 You may use the @code{document} command again to change the
11201 documentation of a command. Redefining the command with @code{define}
11202 does not change the documentation.
11204 @kindex help user-defined
11205 @item help user-defined
11206 List all user-defined commands, with the first line of the documentation
11211 @itemx show user @var{commandname}
11212 Display the @value{GDBN} commands used to define @var{commandname} (but not its
11213 documentation). If no @var{commandname} is given, display the
11214 definitions for all user-defined commands.
11217 When user-defined commands are executed, the
11218 commands of the definition are not printed. An error in any command
11219 stops execution of the user-defined command.
11221 If used interactively, commands that would ask for confirmation proceed
11222 without asking when used inside a user-defined command. Many @value{GDBN}
11223 commands that normally print messages to say what they are doing omit the
11224 messages when used in a user-defined command.
11227 @section User-defined command hooks
11228 @cindex command files
11230 You may define @emph{hooks}, which are a special kind of user-defined
11231 command. Whenever you run the command @samp{foo}, if the user-defined
11232 command @samp{hook-foo} exists, it is executed (with no arguments)
11233 before that command.
11235 In addition, a pseudo-command, @samp{stop} exists. Defining
11236 (@samp{hook-stop}) makes the associated commands execute every time
11237 execution stops in your program: before breakpoint commands are run,
11238 displays are printed, or the stack frame is printed.
11240 For example, to ignore @code{SIGALRM} signals while
11241 single-stepping, but treat them normally during normal execution,
11246 handle SIGALRM nopass
11250 handle SIGALRM pass
11253 define hook-continue
11254 handle SIGLARM pass
11258 You can define a hook for any single-word command in @value{GDBN}, but
11259 not for command aliases; you should define a hook for the basic command
11260 name, e.g. @code{backtrace} rather than @code{bt}.
11261 @c FIXME! So how does Joe User discover whether a command is an alias
11263 If an error occurs during the execution of your hook, execution of
11264 @value{GDBN} commands stops and @value{GDBN} issues a prompt
11265 (before the command that you actually typed had a chance to run).
11267 If you try to define a hook which does not match any known command, you
11268 get a warning from the @code{define} command.
11270 @node Command Files
11271 @section Command files
11273 @cindex command files
11274 A command file for @value{GDBN} is a file of lines that are @value{GDBN}
11275 commands. Comments (lines starting with @kbd{#}) may also be included.
11276 An empty line in a command file does nothing; it does not mean to repeat
11277 the last command, as it would from the terminal.
11280 @cindex @file{.gdbinit}
11281 When you start @value{GDBN}, it automatically executes commands from its
11282 @dfn{init files}. These are files named @file{.gdbinit} on Unix, or
11283 @file{gdb.ini} on DOS/Windows. @value{GDBN} reads the init file (if
11284 any) in your home directory, then processes command line options and
11285 operands, and then reads the init file (if any) in the current working
11286 directory. This is so the init file in your home directory can set
11287 options (such as @code{set complaints}) which affect the processing of
11288 the command line options and operands. The init files are not executed
11289 if you use the @samp{-nx} option; @pxref{Mode Options, ,Choosing modes}.
11291 @cindex init file name
11292 On some configurations of @value{GDBN}, the init file is known by a
11293 different name (these are typically environments where a specialized
11294 form of @value{GDBN} may need to coexist with other forms, hence a
11295 different name for the specialized version's init file). These are the
11296 environments with special init file names:
11301 VxWorks (Wind River Systems real-time OS): @samp{.vxgdbinit}
11303 @kindex .os68gdbinit
11305 OS68K (Enea Data Systems real-time OS): @samp{.os68gdbinit}
11309 ES-1800 (Ericsson Telecom AB M68000 emulator): @samp{.esgdbinit}
11312 You can also request the execution of a command file with the
11313 @code{source} command:
11317 @item source @var{filename}
11318 Execute the command file @var{filename}.
11321 The lines in a command file are executed sequentially. They are not
11322 printed as they are executed. An error in any command terminates execution
11323 of the command file.
11325 Commands that would ask for confirmation if used interactively proceed
11326 without asking when used in a command file. Many @value{GDBN} commands that
11327 normally print messages to say what they are doing omit the messages
11328 when called from command files.
11331 @section Commands for controlled output
11333 During the execution of a command file or a user-defined command, normal
11334 @value{GDBN} output is suppressed; the only output that appears is what is
11335 explicitly printed by the commands in the definition. This section
11336 describes three commands useful for generating exactly the output you
11341 @item echo @var{text}
11342 @c I do not consider backslash-space a standard C escape sequence
11343 @c because it is not in ANSI.
11344 Print @var{text}. Nonprinting characters can be included in
11345 @var{text} using C escape sequences, such as @samp{\n} to print a
11346 newline. @strong{No newline is printed unless you specify one.}
11347 In addition to the standard C escape sequences, a backslash followed
11348 by a space stands for a space. This is useful for displaying a
11349 string with spaces at the beginning or the end, since leading and
11350 trailing spaces are otherwise trimmed from all arguments.
11351 To print @samp{@w{ }and foo =@w{ }}, use the command
11352 @samp{echo \@w{ }and foo = \@w{ }}.
11354 A backslash at the end of @var{text} can be used, as in C, to continue
11355 the command onto subsequent lines. For example,
11358 echo This is some text\n\
11359 which is continued\n\
11360 onto several lines.\n
11363 produces the same output as
11366 echo This is some text\n
11367 echo which is continued\n
11368 echo onto several lines.\n
11372 @item output @var{expression}
11373 Print the value of @var{expression} and nothing but that value: no
11374 newlines, no @samp{$@var{nn} = }. The value is not entered in the
11375 value history either. @xref{Expressions, ,Expressions}, for more information
11378 @item output/@var{fmt} @var{expression}
11379 Print the value of @var{expression} in format @var{fmt}. You can use
11380 the same formats as for @code{print}. @xref{Output Formats,,Output
11381 formats}, for more information.
11384 @item printf @var{string}, @var{expressions}@dots{}
11385 Print the values of the @var{expressions} under the control of
11386 @var{string}. The @var{expressions} are separated by commas and may be
11387 either numbers or pointers. Their values are printed as specified by
11388 @var{string}, exactly as if your program were to execute the C
11392 printf (@var{string}, @var{expressions}@dots{});
11395 For example, you can print two values in hex like this:
11398 printf "foo, bar-foo = 0x%x, 0x%x\n", foo, bar-foo
11401 The only backslash-escape sequences that you can use in the format
11402 string are the simple ones that consist of backslash followed by a
11407 @chapter Using @value{GDBN} under @sc{gnu} Emacs
11410 @cindex @sc{gnu} Emacs
11411 A special interface allows you to use @sc{gnu} Emacs to view (and
11412 edit) the source files for the program you are debugging with
11415 To use this interface, use the command @kbd{M-x gdb} in Emacs. Give the
11416 executable file you want to debug as an argument. This command starts
11417 @value{GDBN} as a subprocess of Emacs, with input and output through a newly
11418 created Emacs buffer.
11419 @c (Do not use the @code{-tui} option to run @value{GDBN} from Emacs.)
11421 Using @value{GDBN} under Emacs is just like using @value{GDBN} normally except for two
11426 All ``terminal'' input and output goes through the Emacs buffer.
11429 This applies both to @value{GDBN} commands and their output, and to the input
11430 and output done by the program you are debugging.
11432 This is useful because it means that you can copy the text of previous
11433 commands and input them again; you can even use parts of the output
11436 All the facilities of Emacs' Shell mode are available for interacting
11437 with your program. In particular, you can send signals the usual
11438 way---for example, @kbd{C-c C-c} for an interrupt, @kbd{C-c C-z} for a
11443 @value{GDBN} displays source code through Emacs.
11446 Each time @value{GDBN} displays a stack frame, Emacs automatically finds the
11447 source file for that frame and puts an arrow (@samp{=>}) at the
11448 left margin of the current line. Emacs uses a separate buffer for
11449 source display, and splits the screen to show both your @value{GDBN} session
11452 Explicit @value{GDBN} @code{list} or search commands still produce output as
11453 usual, but you probably have no reason to use them from Emacs.
11456 @emph{Warning:} If the directory where your program resides is not your
11457 current directory, it can be easy to confuse Emacs about the location of
11458 the source files, in which case the auxiliary display buffer does not
11459 appear to show your source. @value{GDBN} can find programs by searching your
11460 environment's @code{PATH} variable, so the @value{GDBN} input and output
11461 session proceeds normally; but Emacs does not get enough information
11462 back from @value{GDBN} to locate the source files in this situation. To
11463 avoid this problem, either start @value{GDBN} mode from the directory where
11464 your program resides, or specify an absolute file name when prompted for the
11465 @kbd{M-x gdb} argument.
11467 A similar confusion can result if you use the @value{GDBN} @code{file} command to
11468 switch to debugging a program in some other location, from an existing
11469 @value{GDBN} buffer in Emacs.
11472 By default, @kbd{M-x gdb} calls the program called @file{gdb}. If
11473 you need to call @value{GDBN} by a different name (for example, if you keep
11474 several configurations around, with different names) you can set the
11475 Emacs variable @code{gdb-command-name}; for example,
11478 (setq gdb-command-name "mygdb")
11482 (preceded by @kbd{ESC ESC}, or typed in the @code{*scratch*} buffer, or
11483 in your @file{.emacs} file) makes Emacs call the program named
11484 ``@code{mygdb}'' instead.
11486 In the @value{GDBN} I/O buffer, you can use these special Emacs commands in
11487 addition to the standard Shell mode commands:
11491 Describe the features of Emacs' @value{GDBN} Mode.
11494 Execute to another source line, like the @value{GDBN} @code{step} command; also
11495 update the display window to show the current file and location.
11498 Execute to next source line in this function, skipping all function
11499 calls, like the @value{GDBN} @code{next} command. Then update the display window
11500 to show the current file and location.
11503 Execute one instruction, like the @value{GDBN} @code{stepi} command; update
11504 display window accordingly.
11506 @item M-x gdb-nexti
11507 Execute to next instruction, using the @value{GDBN} @code{nexti} command; update
11508 display window accordingly.
11511 Execute until exit from the selected stack frame, like the @value{GDBN}
11512 @code{finish} command.
11515 Continue execution of your program, like the @value{GDBN} @code{continue}
11518 @emph{Warning:} In Emacs v19, this command is @kbd{C-c C-p}.
11521 Go up the number of frames indicated by the numeric argument
11522 (@pxref{Arguments, , Numeric Arguments, Emacs, The @sc{gnu} Emacs Manual}),
11523 like the @value{GDBN} @code{up} command.
11525 @emph{Warning:} In Emacs v19, this command is @kbd{C-c C-u}.
11528 Go down the number of frames indicated by the numeric argument, like the
11529 @value{GDBN} @code{down} command.
11531 @emph{Warning:} In Emacs v19, this command is @kbd{C-c C-d}.
11534 Read the number where the cursor is positioned, and insert it at the end
11535 of the @value{GDBN} I/O buffer. For example, if you wish to disassemble code
11536 around an address that was displayed earlier, type @kbd{disassemble};
11537 then move the cursor to the address display, and pick up the
11538 argument for @code{disassemble} by typing @kbd{C-x &}.
11540 You can customize this further by defining elements of the list
11541 @code{gdb-print-command}; once it is defined, you can format or
11542 otherwise process numbers picked up by @kbd{C-x &} before they are
11543 inserted. A numeric argument to @kbd{C-x &} indicates that you
11544 wish special formatting, and also acts as an index to pick an element of the
11545 list. If the list element is a string, the number to be inserted is
11546 formatted using the Emacs function @code{format}; otherwise the number
11547 is passed as an argument to the corresponding list element.
11550 In any source file, the Emacs command @kbd{C-x SPC} (@code{gdb-break})
11551 tells @value{GDBN} to set a breakpoint on the source line point is on.
11553 If you accidentally delete the source-display buffer, an easy way to get
11554 it back is to type the command @code{f} in the @value{GDBN} buffer, to
11555 request a frame display; when you run under Emacs, this recreates
11556 the source buffer if necessary to show you the context of the current
11559 The source files displayed in Emacs are in ordinary Emacs buffers
11560 which are visiting the source files in the usual way. You can edit
11561 the files with these buffers if you wish; but keep in mind that @value{GDBN}
11562 communicates with Emacs in terms of line numbers. If you add or
11563 delete lines from the text, the line numbers that @value{GDBN} knows cease
11564 to correspond properly with the code.
11566 @c The following dropped because Epoch is nonstandard. Reactivate
11567 @c if/when v19 does something similar. ---doc@cygnus.com 19dec1990
11569 @kindex Emacs Epoch environment
11573 Version 18 of @sc{gnu} Emacs has a built-in window system
11574 called the @code{epoch}
11575 environment. Users of this environment can use a new command,
11576 @code{inspect} which performs identically to @code{print} except that
11577 each value is printed in its own window.
11581 @chapter Reporting Bugs in @value{GDBN}
11582 @cindex bugs in @value{GDBN}
11583 @cindex reporting bugs in @value{GDBN}
11585 Your bug reports play an essential role in making @value{GDBN} reliable.
11587 Reporting a bug may help you by bringing a solution to your problem, or it
11588 may not. But in any case the principal function of a bug report is to help
11589 the entire community by making the next version of @value{GDBN} work better. Bug
11590 reports are your contribution to the maintenance of @value{GDBN}.
11592 In order for a bug report to serve its purpose, you must include the
11593 information that enables us to fix the bug.
11596 * Bug Criteria:: Have you found a bug?
11597 * Bug Reporting:: How to report bugs
11601 @section Have you found a bug?
11602 @cindex bug criteria
11604 If you are not sure whether you have found a bug, here are some guidelines:
11607 @cindex fatal signal
11608 @cindex debugger crash
11609 @cindex crash of debugger
11611 If the debugger gets a fatal signal, for any input whatever, that is a
11612 @value{GDBN} bug. Reliable debuggers never crash.
11614 @cindex error on valid input
11616 If @value{GDBN} produces an error message for valid input, that is a
11617 bug. (Note that if you're cross debugging, the problem may also be
11618 somewhere in the connection to the target.)
11620 @cindex invalid input
11622 If @value{GDBN} does not produce an error message for invalid input,
11623 that is a bug. However, you should note that your idea of
11624 ``invalid input'' might be our idea of ``an extension'' or ``support
11625 for traditional practice''.
11628 If you are an experienced user of debugging tools, your suggestions
11629 for improvement of @value{GDBN} are welcome in any case.
11632 @node Bug Reporting
11633 @section How to report bugs
11634 @cindex bug reports
11635 @cindex @value{GDBN} bugs, reporting
11637 A number of companies and individuals offer support for @sc{gnu} products.
11638 If you obtained @value{GDBN} from a support organization, we recommend you
11639 contact that organization first.
11641 You can find contact information for many support companies and
11642 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
11644 @c should add a web page ref...
11646 In any event, we also recommend that you send bug reports for
11647 @value{GDBN} to this addresses:
11650 bug-gdb@@prep.ai.mit.edu
11653 @strong{Do not send bug reports to @samp{info-gdb}, or to
11654 @samp{help-gdb}, or to any newsgroups.} Most users of @value{GDBN} do
11655 not want to receive bug reports. Those that do have arranged to receive
11658 The mailing list @samp{bug-gdb} has a newsgroup @samp{gnu.gdb.bug} which
11659 serves as a repeater. The mailing list and the newsgroup carry exactly
11660 the same messages. Often people think of posting bug reports to the
11661 newsgroup instead of mailing them. This appears to work, but it has one
11662 problem which can be crucial: a newsgroup posting often lacks a mail
11663 path back to the sender. Thus, if we need to ask for more information,
11664 we may be unable to reach you. For this reason, it is better to send
11665 bug reports to the mailing list.
11667 As a last resort, send bug reports on paper to:
11670 @sc{gnu} Debugger Bugs
11671 Free Software Foundation Inc.
11672 59 Temple Place - Suite 330
11673 Boston, MA 02111-1307
11677 The fundamental principle of reporting bugs usefully is this:
11678 @strong{report all the facts}. If you are not sure whether to state a
11679 fact or leave it out, state it!
11681 Often people omit facts because they think they know what causes the
11682 problem and assume that some details do not matter. Thus, you might
11683 assume that the name of the variable you use in an example does not matter.
11684 Well, probably it does not, but one cannot be sure. Perhaps the bug is a
11685 stray memory reference which happens to fetch from the location where that
11686 name is stored in memory; perhaps, if the name were different, the contents
11687 of that location would fool the debugger into doing the right thing despite
11688 the bug. Play it safe and give a specific, complete example. That is the
11689 easiest thing for you to do, and the most helpful.
11691 Keep in mind that the purpose of a bug report is to enable us to fix the
11692 bug. It may be that the bug has been reported previously, but neither
11693 you nor we can know that unless your bug report is complete and
11696 Sometimes people give a few sketchy facts and ask, ``Does this ring a
11697 bell?'' Those bug reports are useless, and we urge everyone to
11698 @emph{refuse to respond to them} except to chide the sender to report
11701 To enable us to fix the bug, you should include all these things:
11705 The version of @value{GDBN}. @value{GDBN} announces it if you start
11706 with no arguments; you can also print it at any time using @code{show
11709 Without this, we will not know whether there is any point in looking for
11710 the bug in the current version of @value{GDBN}.
11713 The type of machine you are using, and the operating system name and
11717 What compiler (and its version) was used to compile @value{GDBN}---e.g.
11718 ``@value{GCC}--2.8.1''.
11721 What compiler (and its version) was used to compile the program you are
11722 debugging---e.g. ``@value{GCC}--2.8.1'', or ``HP92453-01 A.10.32.03 HP
11723 C Compiler''. For GCC, you can say @code{gcc --version} to get this
11724 information; for other compilers, see the documentation for those
11728 The command arguments you gave the compiler to compile your example and
11729 observe the bug. For example, did you use @samp{-O}? To guarantee
11730 you will not omit something important, list them all. A copy of the
11731 Makefile (or the output from make) is sufficient.
11733 If we were to try to guess the arguments, we would probably guess wrong
11734 and then we might not encounter the bug.
11737 A complete input script, and all necessary source files, that will
11741 A description of what behavior you observe that you believe is
11742 incorrect. For example, ``It gets a fatal signal.''
11744 Of course, if the bug is that @value{GDBN} gets a fatal signal, then we
11745 will certainly notice it. But if the bug is incorrect output, we might
11746 not notice unless it is glaringly wrong. You might as well not give us
11747 a chance to make a mistake.
11749 Even if the problem you experience is a fatal signal, you should still
11750 say so explicitly. Suppose something strange is going on, such as, your
11751 copy of @value{GDBN} is out of synch, or you have encountered a bug in
11752 the C library on your system. (This has happened!) Your copy might
11753 crash and ours would not. If you told us to expect a crash, then when
11754 ours fails to crash, we would know that the bug was not happening for
11755 us. If you had not told us to expect a crash, then we would not be able
11756 to draw any conclusion from our observations.
11759 If you wish to suggest changes to the @value{GDBN} source, send us context
11760 diffs. If you even discuss something in the @value{GDBN} source, refer to
11761 it by context, not by line number.
11763 The line numbers in our development sources will not match those in your
11764 sources. Your line numbers would convey no useful information to us.
11768 Here are some things that are not necessary:
11772 A description of the envelope of the bug.
11774 Often people who encounter a bug spend a lot of time investigating
11775 which changes to the input file will make the bug go away and which
11776 changes will not affect it.
11778 This is often time consuming and not very useful, because the way we
11779 will find the bug is by running a single example under the debugger
11780 with breakpoints, not by pure deduction from a series of examples.
11781 We recommend that you save your time for something else.
11783 Of course, if you can find a simpler example to report @emph{instead}
11784 of the original one, that is a convenience for us. Errors in the
11785 output will be easier to spot, running under the debugger will take
11786 less time, and so on.
11788 However, simplification is not vital; if you do not want to do this,
11789 report the bug anyway and send us the entire test case you used.
11792 A patch for the bug.
11794 A patch for the bug does help us if it is a good one. But do not omit
11795 the necessary information, such as the test case, on the assumption that
11796 a patch is all we need. We might see problems with your patch and decide
11797 to fix the problem another way, or we might not understand it at all.
11799 Sometimes with a program as complicated as @value{GDBN} it is very hard to
11800 construct an example that will make the program follow a certain path
11801 through the code. If you do not send us the example, we will not be able
11802 to construct one, so we will not be able to verify that the bug is fixed.
11804 And if we cannot understand what bug you are trying to fix, or why your
11805 patch should be an improvement, we will not install it. A test case will
11806 help us to understand.
11809 A guess about what the bug is or what it depends on.
11811 Such guesses are usually wrong. Even we cannot guess right about such
11812 things without first using the debugger to find the facts.
11815 @c The readline documentation is distributed with the readline code
11816 @c and consists of the two following files:
11818 @c inc-hist.texinfo
11819 @c Use -I with makeinfo to point to the appropriate directory,
11820 @c environment var TEXINPUTS with TeX.
11821 @include rluser.texinfo
11822 @include inc-hist.texinfo
11825 @node Formatting Documentation
11826 @appendix Formatting Documentation
11828 @cindex @value{GDBN} reference card
11829 @cindex reference card
11830 The @value{GDBN} 4 release includes an already-formatted reference card, ready
11831 for printing with PostScript or Ghostscript, in the @file{gdb}
11832 subdirectory of the main source directory@footnote{In
11833 @file{gdb-@value{GDBVN}/gdb/refcard.ps} of the version @value{GDBVN}
11834 release.}. If you can use PostScript or Ghostscript with your printer,
11835 you can print the reference card immediately with @file{refcard.ps}.
11837 The release also includes the source for the reference card. You
11838 can format it, using @TeX{}, by typing:
11844 The @value{GDBN} reference card is designed to print in @dfn{landscape}
11845 mode on US ``letter'' size paper;
11846 that is, on a sheet 11 inches wide by 8.5 inches
11847 high. You will need to specify this form of printing as an option to
11848 your @sc{dvi} output program.
11850 @cindex documentation
11852 All the documentation for @value{GDBN} comes as part of the machine-readable
11853 distribution. The documentation is written in Texinfo format, which is
11854 a documentation system that uses a single source file to produce both
11855 on-line information and a printed manual. You can use one of the Info
11856 formatting commands to create the on-line version of the documentation
11857 and @TeX{} (or @code{texi2roff}) to typeset the printed version.
11859 @value{GDBN} includes an already formatted copy of the on-line Info
11860 version of this manual in the @file{gdb} subdirectory. The main Info
11861 file is @file{gdb-@value{GDBVN}/gdb/gdb.info}, and it refers to
11862 subordinate files matching @samp{gdb.info*} in the same directory. If
11863 necessary, you can print out these files, or read them with any editor;
11864 but they are easier to read using the @code{info} subsystem in @sc{gnu}
11865 Emacs or the standalone @code{info} program, available as part of the
11866 @sc{gnu} Texinfo distribution.
11868 If you want to format these Info files yourself, you need one of the
11869 Info formatting programs, such as @code{texinfo-format-buffer} or
11872 If you have @code{makeinfo} installed, and are in the top level
11873 @value{GDBN} source directory (@file{gdb-@value{GDBVN}}, in the case of
11874 version @value{GDBVN}), you can make the Info file by typing:
11881 If you want to typeset and print copies of this manual, you need @TeX{},
11882 a program to print its @sc{dvi} output files, and @file{texinfo.tex}, the
11883 Texinfo definitions file.
11885 @TeX{} is a typesetting program; it does not print files directly, but
11886 produces output files called @sc{dvi} files. To print a typeset
11887 document, you need a program to print @sc{dvi} files. If your system
11888 has @TeX{} installed, chances are it has such a program. The precise
11889 command to use depends on your system; @kbd{lpr -d} is common; another
11890 (for PostScript devices) is @kbd{dvips}. The @sc{dvi} print command may
11891 require a file name without any extension or a @samp{.dvi} extension.
11893 @TeX{} also requires a macro definitions file called
11894 @file{texinfo.tex}. This file tells @TeX{} how to typeset a document
11895 written in Texinfo format. On its own, @TeX{} cannot either read or
11896 typeset a Texinfo file. @file{texinfo.tex} is distributed with GDB
11897 and is located in the @file{gdb-@var{version-number}/texinfo}
11900 If you have @TeX{} and a @sc{dvi} printer program installed, you can
11901 typeset and print this manual. First switch to the the @file{gdb}
11902 subdirectory of the main source directory (for example, to
11903 @file{gdb-@value{GDBVN}/gdb}) and type:
11909 Then give @file{gdb.dvi} to your @sc{dvi} printing program.
11911 @node Installing GDB
11912 @appendix Installing @value{GDBN}
11913 @cindex configuring @value{GDBN}
11914 @cindex installation
11916 @value{GDBN} comes with a @code{configure} script that automates the process
11917 of preparing @value{GDBN} for installation; you can then use @code{make} to
11918 build the @code{gdb} program.
11920 @c irrelevant in info file; it's as current as the code it lives with.
11921 @footnote{If you have a more recent version of @value{GDBN} than @value{GDBVN},
11922 look at the @file{README} file in the sources; we may have improved the
11923 installation procedures since publishing this manual.}
11926 The @value{GDBN} distribution includes all the source code you need for
11927 @value{GDBN} in a single directory, whose name is usually composed by
11928 appending the version number to @samp{gdb}.
11930 For example, the @value{GDBN} version @value{GDBVN} distribution is in the
11931 @file{gdb-@value{GDBVN}} directory. That directory contains:
11934 @item gdb-@value{GDBVN}/configure @r{(and supporting files)}
11935 script for configuring @value{GDBN} and all its supporting libraries
11937 @item gdb-@value{GDBVN}/gdb
11938 the source specific to @value{GDBN} itself
11940 @item gdb-@value{GDBVN}/bfd
11941 source for the Binary File Descriptor library
11943 @item gdb-@value{GDBVN}/include
11944 @sc{gnu} include files
11946 @item gdb-@value{GDBVN}/libiberty
11947 source for the @samp{-liberty} free software library
11949 @item gdb-@value{GDBVN}/opcodes
11950 source for the library of opcode tables and disassemblers
11952 @item gdb-@value{GDBVN}/readline
11953 source for the @sc{gnu} command-line interface
11955 @item gdb-@value{GDBVN}/glob
11956 source for the @sc{gnu} filename pattern-matching subroutine
11958 @item gdb-@value{GDBVN}/mmalloc
11959 source for the @sc{gnu} memory-mapped malloc package
11962 The simplest way to configure and build @value{GDBN} is to run @code{configure}
11963 from the @file{gdb-@var{version-number}} source directory, which in
11964 this example is the @file{gdb-@value{GDBVN}} directory.
11966 First switch to the @file{gdb-@var{version-number}} source directory
11967 if you are not already in it; then run @code{configure}. Pass the
11968 identifier for the platform on which @value{GDBN} will run as an
11974 cd gdb-@value{GDBVN}
11975 ./configure @var{host}
11980 where @var{host} is an identifier such as @samp{sun4} or
11981 @samp{decstation}, that identifies the platform where @value{GDBN} will run.
11982 (You can often leave off @var{host}; @code{configure} tries to guess the
11983 correct value by examining your system.)
11985 Running @samp{configure @var{host}} and then running @code{make} builds the
11986 @file{bfd}, @file{readline}, @file{mmalloc}, and @file{libiberty}
11987 libraries, then @code{gdb} itself. The configured source files, and the
11988 binaries, are left in the corresponding source directories.
11991 @code{configure} is a Bourne-shell (@code{/bin/sh}) script; if your
11992 system does not recognize this automatically when you run a different
11993 shell, you may need to run @code{sh} on it explicitly:
11996 sh configure @var{host}
11999 If you run @code{configure} from a directory that contains source
12000 directories for multiple libraries or programs, such as the
12001 @file{gdb-@value{GDBVN}} source directory for version @value{GDBVN}, @code{configure}
12002 creates configuration files for every directory level underneath (unless
12003 you tell it not to, with the @samp{--norecursion} option).
12005 You can run the @code{configure} script from any of the
12006 subordinate directories in the @value{GDBN} distribution if you only want to
12007 configure that subdirectory, but be sure to specify a path to it.
12009 For example, with version @value{GDBVN}, type the following to configure only
12010 the @code{bfd} subdirectory:
12014 cd gdb-@value{GDBVN}/bfd
12015 ../configure @var{host}
12019 You can install @code{@value{GDBP}} anywhere; it has no hardwired paths.
12020 However, you should make sure that the shell on your path (named by
12021 the @samp{SHELL} environment variable) is publicly readable. Remember
12022 that @value{GDBN} uses the shell to start your program---some systems refuse to
12023 let @value{GDBN} debug child processes whose programs are not readable.
12026 * Separate Objdir:: Compiling @value{GDBN} in another directory
12027 * Config Names:: Specifying names for hosts and targets
12028 * Configure Options:: Summary of options for configure
12031 @node Separate Objdir
12032 @section Compiling @value{GDBN} in another directory
12034 If you want to run @value{GDBN} versions for several host or target machines,
12035 you need a different @code{gdb} compiled for each combination of
12036 host and target. @code{configure} is designed to make this easy by
12037 allowing you to generate each configuration in a separate subdirectory,
12038 rather than in the source directory. If your @code{make} program
12039 handles the @samp{VPATH} feature (@sc{gnu} @code{make} does), running
12040 @code{make} in each of these directories builds the @code{gdb}
12041 program specified there.
12043 To build @code{gdb} in a separate directory, run @code{configure}
12044 with the @samp{--srcdir} option to specify where to find the source.
12045 (You also need to specify a path to find @code{configure}
12046 itself from your working directory. If the path to @code{configure}
12047 would be the same as the argument to @samp{--srcdir}, you can leave out
12048 the @samp{--srcdir} option; it is assumed.)
12050 For example, with version @value{GDBVN}, you can build @value{GDBN} in a
12051 separate directory for a Sun 4 like this:
12055 cd gdb-@value{GDBVN}
12058 ../gdb-@value{GDBVN}/configure sun4
12063 When @code{configure} builds a configuration using a remote source
12064 directory, it creates a tree for the binaries with the same structure
12065 (and using the same names) as the tree under the source directory. In
12066 the example, you'd find the Sun 4 library @file{libiberty.a} in the
12067 directory @file{gdb-sun4/libiberty}, and @value{GDBN} itself in
12068 @file{gdb-sun4/gdb}.
12070 One popular reason to build several @value{GDBN} configurations in separate
12071 directories is to configure @value{GDBN} for cross-compiling (where
12072 @value{GDBN} runs on one machine---the @dfn{host}---while debugging
12073 programs that run on another machine---the @dfn{target}).
12074 You specify a cross-debugging target by
12075 giving the @samp{--target=@var{target}} option to @code{configure}.
12077 When you run @code{make} to build a program or library, you must run
12078 it in a configured directory---whatever directory you were in when you
12079 called @code{configure} (or one of its subdirectories).
12081 The @code{Makefile} that @code{configure} generates in each source
12082 directory also runs recursively. If you type @code{make} in a source
12083 directory such as @file{gdb-@value{GDBVN}} (or in a separate configured
12084 directory configured with @samp{--srcdir=@var{dirname}/gdb-@value{GDBVN}}), you
12085 will build all the required libraries, and then build GDB.
12087 When you have multiple hosts or targets configured in separate
12088 directories, you can run @code{make} on them in parallel (for example,
12089 if they are NFS-mounted on each of the hosts); they will not interfere
12093 @section Specifying names for hosts and targets
12095 The specifications used for hosts and targets in the @code{configure}
12096 script are based on a three-part naming scheme, but some short predefined
12097 aliases are also supported. The full naming scheme encodes three pieces
12098 of information in the following pattern:
12101 @var{architecture}-@var{vendor}-@var{os}
12104 For example, you can use the alias @code{sun4} as a @var{host} argument,
12105 or as the value for @var{target} in a @code{--target=@var{target}}
12106 option. The equivalent full name is @samp{sparc-sun-sunos4}.
12108 The @code{configure} script accompanying @value{GDBN} does not provide
12109 any query facility to list all supported host and target names or
12110 aliases. @code{configure} calls the Bourne shell script
12111 @code{config.sub} to map abbreviations to full names; you can read the
12112 script, if you wish, or you can use it to test your guesses on
12113 abbreviations---for example:
12116 % sh config.sub i386-linux
12118 % sh config.sub alpha-linux
12119 alpha-unknown-linux-gnu
12120 % sh config.sub hp9k700
12122 % sh config.sub sun4
12123 sparc-sun-sunos4.1.1
12124 % sh config.sub sun3
12125 m68k-sun-sunos4.1.1
12126 % sh config.sub i986v
12127 Invalid configuration `i986v': machine `i986v' not recognized
12131 @code{config.sub} is also distributed in the @value{GDBN} source
12132 directory (@file{gdb-@value{GDBVN}}, for version @value{GDBVN}).
12134 @node Configure Options
12135 @section @code{configure} options
12137 Here is a summary of the @code{configure} options and arguments that
12138 are most often useful for building @value{GDBN}. @code{configure} also has
12139 several other options not listed here. @inforef{What Configure
12140 Does,,configure.info}, for a full explanation of @code{configure}.
12143 configure @r{[}--help@r{]}
12144 @r{[}--prefix=@var{dir}@r{]}
12145 @r{[}--exec-prefix=@var{dir}@r{]}
12146 @r{[}--srcdir=@var{dirname}@r{]}
12147 @r{[}--norecursion@r{]} @r{[}--rm@r{]}
12148 @r{[}--target=@var{target}@r{]}
12153 You may introduce options with a single @samp{-} rather than
12154 @samp{--} if you prefer; but you may abbreviate option names if you use
12159 Display a quick summary of how to invoke @code{configure}.
12161 @item --prefix=@var{dir}
12162 Configure the source to install programs and files under directory
12165 @item --exec-prefix=@var{dir}
12166 Configure the source to install programs under directory
12169 @c avoid splitting the warning from the explanation:
12171 @item --srcdir=@var{dirname}
12172 @strong{Warning: using this option requires @sc{gnu} @code{make}, or another
12173 @code{make} that implements the @code{VPATH} feature.}@*
12174 Use this option to make configurations in directories separate from the
12175 @value{GDBN} source directories. Among other things, you can use this to
12176 build (or maintain) several configurations simultaneously, in separate
12177 directories. @code{configure} writes configuration specific files in
12178 the current directory, but arranges for them to use the source in the
12179 directory @var{dirname}. @code{configure} creates directories under
12180 the working directory in parallel to the source directories below
12183 @item --norecursion
12184 Configure only the directory level where @code{configure} is executed; do not
12185 propagate configuration to subdirectories.
12187 @item --target=@var{target}
12188 Configure @value{GDBN} for cross-debugging programs running on the specified
12189 @var{target}. Without this option, @value{GDBN} is configured to debug
12190 programs that run on the same machine (@var{host}) as @value{GDBN} itself.
12192 There is no convenient way to generate a list of all available targets.
12194 @item @var{host} @dots{}
12195 Configure @value{GDBN} to run on the specified @var{host}.
12197 There is no convenient way to generate a list of all available hosts.
12200 There are many other options available as well, but they are generally
12201 needed for special purposes only.
12209 % I think something like @colophon should be in texinfo. In the
12211 \long\def\colophon{\hbox to0pt{}\vfill
12212 \centerline{The body of this manual is set in}
12213 \centerline{\fontname\tenrm,}
12214 \centerline{with headings in {\bf\fontname\tenbf}}
12215 \centerline{and examples in {\tt\fontname\tentt}.}
12216 \centerline{{\it\fontname\tenit\/},}
12217 \centerline{{\bf\fontname\tenbf}, and}
12218 \centerline{{\sl\fontname\tensl\/}}
12219 \centerline{are used for emphasis.}\vfill}
12221 % Blame: doc@cygnus.com, 1991.