1 README for gdb-4.4 release
2 John Gilmore & Stu Grossman 31 Jan 1992
4 This is GDB, the GNU source-level debugger, presently running under un*x.
5 A summary of new features is in the file `WHATS.NEW'.
8 Unpacking and Installation -- quick overview
9 ==========================
11 In this release, the GDB debugger sources, the generic GNU include
12 files, the BFD ("binary file description") library, the readline library,
13 and a miscellaneous library all have directories of their own underneath
14 the gdb-4.4 directory. The idea is that a variety of GNU tools can
15 share a common copy of these things. Configuration scripts and
16 makefiles exist to cruise up and down this directory tree and
17 automatically build all the pieces in the right order.
19 When you unpack the gdb-4.4.tar.Z file, you'll get a directory called
20 `gdb-4.4', which contains:
22 DOC.configure bfd/ configure* glob/ readline/
23 Makefile.in config/ configure.in include/ texinfo/
24 README config.sub* gdb/ libiberty/
26 To build GDB, you can just do:
29 ./configure HOSTTYPE (e.g. sun4, decstation)
31 cp gdb/gdb /usr/local/bin/gdb (or wherever you want)
33 This will configure and build all the libraries as well as GDB.
34 If you get compiler warnings during this stage, see the `Reporting Bugs'
35 section below; there are a few known problems.
37 GDB can be used as a cross-debugger, running on a machine of one type
38 while debugging a program running on a machine of another type. See below.
44 The GDB 4.4 release includes an already-formatted reference card,
45 ready for printing on a PostScript printer, as
46 `gdb-4.4/gdb/refcard.ps'. It uses the most common PostScript fonts:
47 the Times family, Courier, and Symbol. If you have a PostScript
48 printer, you can print the reference card by just sending `refcard.ps'
51 The release also includes the online Info version of this manual
52 already formatted: the main Info file is `gdb-4.4/gdb/gdb.info', and it
53 refers to subordinate files matching `gdb.info*' in the same directory.
55 If you want to make these Info files yourself from the GDB manual's
56 source, you need the GNU `makeinfo' program. Once you have it, you
62 to make the Info file.
64 If you want to format and print copies of the manual, you need
67 * TeX, the public domain typesetting program written by Donald
68 Knuth, must be installed on your system and available through
71 * `gdb-4.4/texinfo': TeX macros defining the GNU Documentation
74 * *A DVI output program.* TeX does not actually make marks on
75 paper; it produces output files called DVI files. If your system
76 has TeX installed, chances are it has a program for printing out
77 these files; one popular example is `dvips', which can print DVI
78 files on PostScript printers.
80 Once you have these things, you can type
85 to format the text of this manual, and print it with the usual output
86 method for TeX DVI files at your site.
88 If you want to print the reference card, but do not have a
89 PostScript printer, or you want to use Computer Modern fonts instead,
90 you can still print it if you have TeX. Format the reference card by
96 The GDB reference card is designed to print in landscape mode on US
97 "letter" size paper; that is, on a sheet 11 inches wide by 8.5 inches
98 high. You will need to specify this form of printing as an option to
99 your DVI output program.
105 GDB comes with a `configure' script that automates the process of
106 preparing GDB for installation; you can then use `make' to build the
109 The gdb distribution includes all the source code you need for gdb
110 in a single directory `gdb-4.4'. That directory in turn contains:
112 `gdb-4.4/configure (and supporting files)'
113 script for configuring GDB and all its supporting libraries.
116 the source specific to GDB itself
119 source for the Binary File Descriptor Library
125 source for the `-liberty' free software library
128 source for the GNU command-line interface
130 It is most convenient to run `configure' from the `gdb-4.4' directory.
131 The simplest way to configure and build GDB is the following:
137 where HOST is something like `sun4' or `decstation', that identifies
138 the platform where GDB will run. This builds the three libraries
139 `bfd', `readline', and `libiberty', then `gdb' itself. The configured
140 source files, and the binaries, are left in the corresponding source
143 `configure' is a Bourne-shell (`/bin/sh') script; if your system
144 does not recognize this automatically when you run a different shell,
145 you may need to run `sh' on it explicitly: `sh configure HOST'.
147 You can *run* the `configure' script from any of the subordinate
148 directories in the GDB distribution (if you only want to configure
149 that subdirectory); but be sure to specify a path to it. For example,
150 to configure only the `bfd' subdirectory,
155 You can install `gdb' anywhere; it has no hardwired paths. Simply
156 copy `gdb/gdb' to the desired directory.
158 However, you should make sure that the shell on your path (named by
159 the `SHELL' environment variable) is publicly readable; some systems
160 refuse to let GDB debug child processes whose programs are not
161 readable, and GDB uses the shell to start your program.
164 Configuration Subdirectories
165 ============================
167 If you want to run GDB versions for several host or target machines,
168 you'll need a different gdb compiled for each combination of host and
169 target. `configure' is designed to make this easy by allowing you to
170 generate each configuration in a separate subdirectory. If your
171 `make' program handles the `VPATH' feature (GNU `make' does), running
172 `make' in each of these directories then builds the gdb program
175 `configure' creates these subdirectories for you when you
176 simultaneously specify several configurations; but it is a good habit
177 even for a single configuration. You can specify the use of
178 subdirectories using the `+subdirs' option (abbreviated `+sub'). For
179 example, you can build GDB this way on a Sun 4 as follows:
182 ./configure +sub sun4
186 When `configure' uses subdirectories to build programs or
187 libraries, it creates nested directories `H-HOST/T-TARGET'.
188 `configure' uses these two directory levels because GDB can be
189 configured for cross-compiling: GDB can run on one machine (the host)
190 while debugging programs that run on another machine (the target).
191 You specify cross-debugging targets by giving the `+target=TARGET'
192 option to `configure'. Specifying only hosts still gives you two
193 levels of subdirectory for each host, with the same configuration
194 suffix on both; that is, if you give any number of hosts but no
195 targets, GDB will be configured for native debugging on each host. On
196 the other hand, whenever you specify both hosts and targets on the
197 same command line, `configure' creates all combinations of the hosts
198 and targets you list.
200 If you run `configure' from a directory (notably, `gdb-4.4') that
201 contains source directories for multiple libraries or programs,
202 `configure' creates the `H-HOST/T-TARGET' subdirectories in each
203 library or program's source directory. For example, typing:
206 configure sun4 +target=vxworks960
208 creates the following directories:
210 gdb-4.4/H-sun4/T-vxworks960
211 gdb-4.4/bfd/H-sun4/T-vxworks960
212 gdb-4.4/gdb/H-sun4/T-vxworks960
213 gdb-4.4/libiberty/H-sun4/T-vxworks960
214 gdb-4.4/readline/H-sun4/T-vxworks960
216 When you run `make' to build a program or library, you must run it
217 in a configured directory. If you made a single configuration,
218 without subdirectories, run `make' in the source directory. If you
219 have `H-HOST/T-TARGET' subdirectories, run `make' in those
222 The `Makefile' generated by `configure' for each source directory
223 runs recursively, so that typing `make' in `gdb-4.4' (or in a
224 `gdb-4.4/H-HOST/T-TARGET' subdirectory) builds all the required
227 When you have multiple hosts or targets configured, you can run
228 `make' on them in parallel (for example, if they are NFS-mounted on
229 each of the hosts); they will not interfere with each other.
231 You can also use the `+objdir=ALTROOT' option to have the
232 configured files placed in a parallel directory structure rather than
233 alongside the source files; *note configure Options::..
236 Specifying Names for Hosts and Targets
237 ======================================
239 The specifications used for hosts and targets in the `configure'
240 script are based on a three-part naming scheme, but some short
241 predefined aliases are also supported. The full naming scheme encodes
242 three pieces of information in the following pattern:
244 ARCHITECTURE-VENDOR-OS
246 For example, you can use the alias `sun4' as a HOST argument or in
247 a `+target=TARGET' option, but the equivalent full name is
250 The following table shows all the architectures, hosts, and OS
251 prefixes that `configure' recognizes in GDB 4.4. Entries in the "OS
252 prefix" column ending in a `*' may be followed by a release number.
255 ARCHITECTURE VENDOR OS prefix
256 ------------+--------------------------+---------------------------
258 580 | altos hp | aix* msdos*
259 a29k | amd ibm | amigados newsos*
260 alliant | amdahl intel | aout nindy*
261 arm | aout isi | bout osf*
262 c1 | apollo little | bsd* sco*
263 c2 | att mips | coff sunos*
264 cray2 | bcs motorola | ctix* svr4
265 h8300 | bout ncr | dgux* sym*
266 i386 | bull next | dynix* sysv*
267 i860 | cbm nyu | ebmon ultrix*
268 i960 | coff sco | esix* unicos*
269 m68000 | convergent sequent | hds unos*
270 m68k | convex sgi | hpux* uts
271 m88k | cray sony | irix* v88r*
272 mips | dec sun | isc* vms*
273 ns32k | encore unicom | kern vxworks*
274 pyramid | gould utek | mach*
284 *Warning:* `configure' can represent a very large number of
285 combinations of architecture, vendor, and OS. There is by no
286 means support available for all possible combinations!
288 The `configure' script accompanying GDB 4.4 does not provide any
289 query facility to list all supported host and target names or aliases.
290 `configure' calls the Bourne shell script `config.sub' to map
291 abbreviations to full names; you can read the script, if you wish, or
292 you can use it to test your guesses on abbreviations--for example:
298 % sh config.sub decstation
300 % sh config.sub hp300bsd
302 % sh config.sub i386v
304 % sh config.sub i486v
305 *** Configuration "i486v" not recognized
307 `config.sub' is also distributed in the directory `gdb-4.4'.
313 Here is a summary of all the `configure' options and arguments that
314 you might use for building GDB:
316 configure [+destdir=DIR] [+subdirs]
317 [+objdir=ALTROOT] [+norecursion] [+rm]
318 [+target=TARGET...] HOST...
320 You may introduce options with the character `-' rather than `+' if
321 you prefer; but you may abbreviate option names if you use `+'.
324 DIR is an installation directory *path prefix*. After you
325 configure with this option, `make install' will install GDB as
326 `DIR/bin/gdb', and the libraries in `DIR/lib'. If you specify
327 `+destdir=/usr/local', for example, `make install' creates
328 `/usr/local/bin/gdb'.
331 Write configuration specific files in subdirectories of the form
335 (and configure the `Makefile' to generate object code in
336 subdirectories of this form as well). Without this option, if you
337 specify only one configuration for GDB, `configure' will use the
338 same directory for source, configured files, and binaries. This
339 option is used automatically if you specify more than one HOST or
340 more than one `+target=TARGET' option on the `configure' command
344 Configure only the directory where `configure' is executed; do not
345 propagate configuration to subdirectories.
348 ALTROOT is an alternative directory used as the root for
349 configured files. `configure' will create directories under
350 ALTROOT in parallel to the source directories. If you use
351 `+objdir=ALTROOT' with `+subdirs', `configure' also builds the
352 `H-HOST/T-TARGET' subdirectories in the directory tree rooted in
356 Remove the configuration that the other arguments specify.
359 Configure GDB for cross-debugging programs running on each
360 specified TARGET. You may specify as many `+target' options as
361 you wish. Without this option, GDB is configured to debug
362 programs that run on the same machine (HOST) as GDB itself.
364 There is no convenient way to generate a list of all available
368 Configure GDB to run on each specified HOST. You may specify as
369 many host names as you wish.
371 There is no convenient way to generate a list of all available
374 `configure' accepts other options, for compatibility with configuring
375 other GNU tools recursively; but these are the only options that
376 affect GDB or its supporting libraries.
379 Languages other than C
381 C++ support has been integrated into gdb. Partial Modula-2 support is
382 now in GDB. GDB should work with FORTRAN programs. (If you have
383 problems, please send a bug report; you may have to refer to some
384 FORTRAN variables with a trailing underscore). I am not aware of
385 anyone who is working on getting gdb to use the syntax of any other
386 language. Pascal programs which use sets, subranges, file variables,
387 or nested functions will not currently work.
392 I have't done this myself so I can't really offer any advice.
393 Remote debugging over serial lines works fine, but the kernel debugging
394 code in here has not been tested in years. Van Jacobson claims to have
395 better kernel debugging.
400 The files m68k-stub.c and i386-stub.c contain two examples of remote
401 stubs to be used with remote.c. They are designeded to run standalone
402 on a 68k or 386 cpu and communicate properly with the remote.c stub
405 The file rem-multi.shar contains a general stub that can probably
406 run on various different flavors of unix to allow debugging over a
407 serial line from one machine to another.
409 Some working remote interfaces for talking to existing ROM monitors
411 remote-eb.c AMD 29000 "EBMON"
412 remote-nindy.c Intel 960 "Nindy"
413 remote-adapt.c AMD 29000 "Adapt"
414 remote-mm.c AMD 29000 "minimon"
416 Remote-vx.c and the vx-share subdirectory contain a remote interface for the
417 VxWorks realtime kernel, which communicates over TCP using the Sun
418 RPC library. This would be a useful starting point for other remote-
419 via-ethernet back ends.
424 The correct address for reporting bugs found in gdb is
425 "bug-gdb@prep.ai.mit.edu". Please email all bugs to that address.
426 Please include the GDB version number (e.g. gdb-4.4), and how
427 you configured it (e.g. "sun4" or "mach386 host, i586-intel-synopsys
432 * If you run with a watchpoint enabled, breakpoints will become
433 erratic and might not stop the program. Disabling or deleting the
434 watchpoint will fix the problem.
436 GDB can produce warnings about symbols that it does not understand. By
437 default, these warnings are disabled. You can enable them by executing
438 `set complaint 10' (which you can put in your ~/.gdbinit if you like).
439 I recommend doing this if you are working on a compiler, assembler,
440 linker, or gdb, since it will point out problems that you may be able
441 to fix. Warnings produced during symbol reading indicate some mismatch
442 between the object file and GDB's symbol reading code. In many cases,
443 it's a mismatch between the specs for the object file format, and what
444 the compiler actually outputs or the debugger actually understands.
446 If you port gdb to a new machine, please send the required changes to
447 bug-gdb@prep.ai.mit.edu. There's lots of information about doing your
448 own port in the file gdb-4.4/gdb/doc/gdbint.texinfo, which you can
449 print out, or read with `info' (see the Makefile.in there). If your
450 changes are more than a few lines, obtain and send in a copyright
451 assignment from gnu@prep.ai.mit.edu, as described in the section
452 `Writing Code for GDB'.
457 xgdb is obsolete. We are not doing any development or support of it.
459 There is an "xxgdb", which shows more promise, which was posted to
462 For those intersted in auto display of source and the availability of
463 an editor while debugging I suggest trying gdb-mode in gnu-emacs
464 (Try typing M-x gdb RETURN). Comments on this mode are welcome.
469 We appreciate having users contribute code that is of general use, but
470 for it to be included in future GDB releases it must be cleanly
471 written. We do not want to include changes that will needlessly make
472 future maintainance difficult. It is not much harder to do things
473 right, and in the long term it is worth it to the GNU project, and
474 probably to you individually as well.
476 If you make substantial changes, you'll have to file a copyright
477 assignment with the Free Software Foundation before we can produce a
478 release that includes your changes. Send mail requesting the copyright
479 assignment to gnu@prep.ai.mit.edu. Do this early, like before the
480 changes actually work, or even before you start them, because a manager
481 or lawyer on your end will probably make this a slow process.
483 Please code according to the GNU coding standards. If you do not have
484 a copy, you can request one by sending mail to gnu@prep.ai.mit.edu.
486 Please try to avoid making machine-specific changes to
487 machine-independent files. If this is unavoidable, put a hook in the
488 machine-independent file which calls a (possibly) machine-dependent
489 macro (for example, the IGNORE_SYMBOL macro can be used for any
490 symbols which need to be ignored on a specific machine. Calling
491 IGNORE_SYMBOL in dbxread.c is a lot cleaner than a maze of #if
492 defined's). The machine-independent code should do whatever "most"
493 machines want if the macro is not defined in param.h. Using #if
494 defined can sometimes be OK (e.g. SET_STACK_LIMIT_HUGE) but should be
495 conditionalized on a specific feature of an operating system (set in
496 tm.h or xm.h) rather than something like #if defined(vax) or #if
497 defined(SYSV). If you use an #ifdef on some symbol that is defined
498 in a header file (e.g. #ifdef TIOCSETP), *please* make sure that you
499 have #include'd the relevant header file in that module!
501 It is better to replace entire routines which may be system-specific,
502 rather than put in a whole bunch of hooks which are probably not going
503 to be helpful for any purpose other than your changes. For example,
504 if you want to modify dbxread.c to deal with DBX debugging symbols
505 which are in COFF files rather than BSD a.out files, do something
506 along the lines of a macro GET_NEXT_SYMBOL, which could have
507 different definitions for COFF and a.out, rather than trying to put
508 the necessary changes throughout all the code in dbxread.c that
509 currently assumes BSD format.
511 When generalizing GDB along a particular interface, please use an
512 attribute-struct rather than inserting tests or switch statements
513 everywhere. For example, GDB has been generalized to handle multiple
514 kinds of remote interfaces -- not by #ifdef's everywhere, but by
515 defining the "target_ops" structure and having a current target (as
516 well as a stack of targets below it, for memory references). Whenever
517 something needs to be done that depends on which remote interface we
518 are using, a flag in the current target_ops structure is tested (e.g.
519 `target_has_stack'), or a function is called through a pointer in the
520 current target_ops structure. In this way, when a new remote interface
521 is added, only one module needs to be touched -- the one that actually
522 implements the new remote interface. Other examples of
523 attribute-structs are BFD access to multiple kinds of object file
524 formats, or GDB's access to multiple source languages.
526 Please avoid duplicating code. For example, in GDB 3.x all the stuff
527 in infptrace.c was duplicated in *-dep.c, and so changing something
528 was very painful. In GDB 4.x, these have all been consolidated
529 into infptrace.c. infptrace.c can deal with variations between
530 systems the same way any system-independent file would (hooks, #if
531 defined, etc.), and machines which are radically different don't need
532 to use infptrace.c at all. The same was true of core_file_command
533 and exec_file_command.
536 Debugging gdb with itself
538 If gdb is limping on your machine, this is the preferred way to get it
539 fully functional. Be warned that in some ancient Unix systems, like
540 Ultrix 4.0, a program can't be running in one process while it is being
541 debugged in another. Rather than doing "./gdb ./gdb", which works on
542 Suns and such, you can copy gdb to gdb2 and then do "./gdb ./gdb2".
544 When you run gdb in the gdb source directory, it will read a ".gdbinit"
545 file that sets up some simple things to make debugging gdb easier. The
546 "info" command, when executed without a subcommand in a gdb being
547 debugged by gdb, will pop you back up to the top level gdb. See
548 .gdbinit for details.
550 I strongly recommend printing out the reference card and using it.
551 Send reference-card suggestions to bug-gdb@prep.ai.mit.edu, just like bugs.
553 If you use emacs, you will probably want to do a "make TAGS" after you
554 configure your distribution; this will put the machine dependent
555 routines for your local machine where they will be accessed first by a
558 Also, make sure that you've either compiled gdb with your local cc, or
559 have run `fixincludes' if you are compiling with gcc.
561 (this is for editing this file with GNU emacs)