1 If you find inaccuracies in this list, please send mail to
2 gdb-patches@sourceware.cygnus.com. If you would like to work on any
3 of these, you should consider sending mail to the same address, to
4 find out whether anyone else is working on it.
7 Known problems in GDB 5.0
8 =========================
10 Below is a list of problems identified during the GDB 5.0 release
11 cycle. People hope to have these problems fixed in a follow-on
16 The BFD directory requires bug-fixed AUTOMAKE et.al.
18 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
19 contained the full path to texinfo.tex when it should have only
20 contained the directory. The bug has been fixed in the current
21 AUTOMAKE sources. Automake snapshots can be found in:
22 ftp://sourceware.cygnus.com/pub/gdb/snapshots
23 and ftp://sourceware.cygnus.com/pub/binutils
27 RFD: infrun.c: No bpstat_stop_status call after proceed over break?
28 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00665.html
30 GDB misses watchpoint triggers after proceeding over a breakpoint on
35 x86 linux GDB and SIGALRM (???)
36 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00803.html
38 This problem has been fixed, but a regression test still needs to be
39 added to the testsuite:
40 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00309.html
46 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
47 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
49 (Broken) support for GDB's remote protocol across UDP is to be
50 included in the follow-on release.
54 Can't build IRIX -> arm GDB.
55 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
58 > Now I'm building for an embedded arm target. If there is a way of turning
59 > remote-rdi off, I couldn't find it. It looks like it gets built by default
60 > in gdb/configure.tgt(line 58) Anyway, the build dies in
61 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
62 > aren't one of the architectures supported.
66 Problem with weak functions
67 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
69 Dan Nicolaescu writes:
70 > It seems that gdb-4.95.1 does not display correctly the function when
71 > stoping in weak functions.
73 > It stops in a function that is defined as weak, not in the function
74 > that is actualy run...
78 GDB 5.0 doesn't work on Linux/SPARC
82 Code Cleanups: Next Release
83 ===========================
85 The following are small cleanups that will hopefully be completed by
90 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
92 Patches in the database.
98 Eliminate all uses of PARAMS in GDB's source code.
102 Fix copyright notices.
104 Turns out that ``1998-2000'' isn't considered valid :-(
106 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
110 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
111 determine the default isa/byte-order.
115 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependant
116 BIG_ENDIAN and LITTLE_ENDIAN.
120 Code Cleanups: General
121 ======================
123 The following are more general cleanups and fixes. They are not tied
124 to any specific release.
128 Eliminate more compiler warnings.
130 Of course there also needs to be the usual debate over which warnings
131 are valid and how to best go about this.
133 One method: choose a single option; get agreement that it is
134 reasonable; try it out to see if there isn't anything silly about it
135 (-Wunused-parameters is an example of that) then incrementally hack
138 The other method is to enable all warnings and eliminate them from one
143 Elimination of ``(catch_errors_ftype *) func''.
145 Like make_cleanup_func it isn't portable.
146 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
147 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
156 [PATCH/5] src/intl/Makefile.in:distclean additions
157 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
159 Do not forget to merge the patch back into the trunk.
163 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
165 At present defs.h includes <endian.h> (which is linux specific) yet
166 almost nothing depends on it. Suggest "gdb_endian.h" which can also
167 handle <machine/endian.h> and only include that where it is really
172 Replace asprintf() calls with xasprintf() calls.
174 As with things like strdup() most calls to asprintf() don't check the
179 Replace strsave() + mstrsave() with libiberty:xstrdup().
183 Replace savestring() with something from libiberty.
185 An xstrldup()? but that would have different semantics.
189 Rationalize use of floatformat_unknown in GDB sources.
191 Instead of defaulting to floatformat_unknown, should hosts/targets
192 specify the value explicitly?
194 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
198 Add a ``name'' member to include/floatformat.h:struct floatformat.
199 Print that name in gdbarch.c.
203 Sort out the harris mess in include/floatformat.h (it hardwires two
204 different floating point formats).
208 See of the GDB local floatformat_do_doublest() and libiberty's
209 floatformat_to_double (which was once GDB's ...) can be merged some
214 Eliminate mmalloc() from GDB.
216 Also eliminate it from defs.h.
220 Eliminate PTR. ISO-C allows ``void *''.
226 GDB should never abort. GDB should either throw ``error ()'' or
227 ``internal_error ()''. Better still GDB should naturally unwind with
232 GDB probably doesn't build on FreeBSD pre 2.2.x
233 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
235 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
242 Readline 4.? is out. A merge wouldn't hurt. Patches are in:
244 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00436.html
248 Deprecate "fg". Apparently ``fg'' is actually continue.
250 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
254 Deprecate current use of ``floatformat_unknown''.
256 Require all targets to explicitly provide their float format instead
257 of defaulting to floatformat unknown. Doing the latter leads to nasty
260 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
264 Rationalize floatformat_to_double() vs floatformat_to_doublest().
266 Looks like GDB migrated floatformat_to_double() to libiberty but then
267 turned around and created a ..._to_doublest() the latter containing
270 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
274 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
276 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
280 Follow through `make check' with --enable-shared.
282 When the srcware tree is configured with --enable-shared, the `expect'
283 program won't run properly. Jim Wilson found out gdb has a local hack
284 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
287 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
291 The ``maintenance deprecate set endian big'' command doesn't notice
292 that it is deprecating ``set endian'' and not ``set endian big'' (big
293 is implemented using an enum). Is anyone going to notice this?
297 When tab expanding something like ``set arch<tab>'' ignore the
298 deprecated ``set archdebug'' and expand to ``set architecture''.
302 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
303 like from arm-tdep.c.
307 Fix uses of ->function.cfunc = set_function().
309 The command.c code calls sfunc() when a set command. Rather than
310 change it suggest fixing the callback function so that it is more
313 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
315 See also ``Fix implementation of ``target xxx''.'' below.
319 IRIX 3.x support is probably broken.
323 New Features and Fixes
324 ======================
326 These are harder than cleanups but easier than work involving
327 fundamental architectural change.
331 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
332 so that you can see how the GDB was created.
334 Some of these (*m.h) would be added to the generated config.h. That
335 in turn would fix a long standing bug where by the build process many
336 not notice a changed tm.h file. Since everything depends on config.h,
337 a change to *m.h forces a change to config.h and, consequently forces
342 Add an "info bfd" command that displays supported object formats,
343 similarly to objdump -i.
345 Is there a command already?
349 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
351 This requires internationalization.
355 Convert GDB build process to AUTOMAKE.
357 See also sub-directory configure below.
359 The current convention is (kind of) to use $(<header>_h) in all
360 dependency lists. It isn't done in a consistent way.
364 Cleanup configury support for optional sub-directories.
366 Check how GCC handles multiple front ends for an example of how things
367 could work. A tentative first step is to rationalize things so that
368 all sub directories are handled in a fashion similar to gdb/mi.
370 See also automake above.
374 Restructure gdb directory tree so that it avoids any 8.3 and 14
379 Add a transcript mechanism to GDB.
381 Such a mechanism might log all gdb input and output to a file in a
382 form that would allow it to be replayed. It could involve ``gdb
383 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
387 Can the xdep files be replaced by autoconf?
391 Document trace machinery
395 Document ui-out and ui-file.
397 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
401 Update texinfo.tex to latest?
407 Incorporate agentexpr.texi into gdb.texinfo
409 agentexpr.texi mostly describes the details of the byte code used for
410 tracepoints, not the internals of the support for this in GDB. So it
411 looks like gdb.texinfo is a better place for this information.
413 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
417 Document overlay machinery.
421 ``(gdb) catch signal SIGNAL''
423 Overlaps with ``handle SIGNAL'' but the implied behavour is different.
424 You can attach commands to a catch but not a handle. A handle has a
425 limited number of hardwired actions.
429 Get the TUI working on all platforms.
433 Add support for ``gdb --- PROGRAM ARGS ...''.
434 Add support for ``gdb -cmd=...''
436 Along with many variations. Check:
438 ????? for a full discussion.
444 Implement ``(gdb) !ls''.
446 Which is very different from ``(gdb) ! ls''. Implementing the latter
449 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
453 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
458 When tab expanding something like ``set arch<tab>'' ignore the
459 deprecated ``set archdebug'' and expand to ``set architecture''.
463 Replace the code that uses the host FPU with an emulator of the target
473 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
474 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
476 The thread_db assisted debugging code doesn't handle exiting threads
477 properly, at least in combination with glibc 2.1.3 (the framework is
478 there, just not the actual code). There are at least two problems
479 that prevent this from working.
481 As an additional reference point, the pre thread_db code did not work
486 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
488 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
490 Christopher Blizzard writes:
492 So, I've done some more digging into this and it looks like Jim
493 Kingdon has reported this problem in the past:
495 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
497 I can reproduce this problem both with and without Tom's patch. Has
498 anyone seen this before? Maybe have a solution for it hanging around?
501 There's a test case for this documented at:
503 when debugging threaded applications you get extra SIGTRAPs
504 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
506 [There should be a GDB testcase - cagney]
510 GDB5 TOT on unixware 7
511 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
514 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
515 > practical matter, the current thread support is somewhat more annoying
516 > than when GDB was thread-unaware.
520 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
522 Add support for packet enable/disable commands with these thread
523 packets. General cleanup.
525 [PATCH] Document the ThreadInfo remote protocol queries
526 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
528 [PATCH] "info threads" queries for remote.c
529 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
536 New languages come onto the scene all the time.
540 Pascal (Pierre Muller, David Taylor)
542 Pierre Muller has contributed patches for adding Pascal Language
545 2 pascal language patches inserted in database
546 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
549 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
553 Java (Anthony Green, David Taylor)
555 Anthony Green has a number of Java patches that did not make it into
559 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
562 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
564 Patch: handle N_MAIN stab
565 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
571 Modify gdb to work correctly with Pascal.
575 Re: Various C++ things
577 value_headof/value_from_vtable_info are worthless, and should be
578 removed. The one place in printcmd.c that uses it should use the RTTI
581 RTTI for g++ should be using the typeinfo functions rather than the
582 vtables. The typeinfo functions are always at offset 4 from the
583 beginning of the vtable, and are always right. The vtables will have
584 weird names like E::VB sometimes. The typeinfo function will always
585 be "E type_info function", or somesuch.
587 value_virtual_fn_field needs to be fixed so there are no failures for
588 virtual functions for C++ using g++.
590 Testsuite cases are the major priority right now for C++ support,
591 since i have to make a lot of changes that could potentially break
596 Add support for Modula3
598 Get DEC/Compaq to contribute their Modula-3 support.
602 Remote Protocol Support
603 =======================
607 set/show remote X-packet ...
609 ``(gdb) help set remote X-packet'' doesn't list the applicable
610 responses. The help message needs to be expanded.
614 Remote protocol doco feedback.
616 Too much feedback to mention needs to be merged in (901660). Search
617 for the word ``remote''.
620 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
621 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
622 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
626 GDB doesn't recover gracefully from remote protocol errors.
628 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
629 ignored and a timeout is required before GDB retries. A pre-cursor to
630 fixing this this is making GDB's remote protocol packet more robust.
632 While downloading to a remote protocol target, gdb ignores packet
633 errors in so far as it will continue to edownload with chunk N+1 even
634 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
635 take a painfully long time to run. As a PS that test needs to be
636 fixed so that it builds on 16 bit machines.
640 Add the cycle step command.
642 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
646 Resolve how to scale things to support very large packets.
650 Resolve how to handle a target that changes things like its endianess
651 on the fly - should it be returned in the ``T'' packet?
653 Underlying problem is that the register file is target endian. If the
654 target endianess changes gdb doesn't know.
661 If / when GDB starts to support the debugging of multi-processor
662 (rather than multi-thread) applications the symtab code will need to
663 be updated a little so that several independant symbol tables are
664 active at a given time.
666 The other interesting change is a clarification of the exact meaning
667 of CORE_ADDR and that has had consequences for a few targets (that
668 were abusing that data type).
672 Investiagate ways of reducing memory.
676 Investigate ways of improving load time.
680 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
682 Consequence of recent symtab clarification. No marks for figuring out
683 who maintains the d10v.
687 Get the MIPS to correctly sign extend all address <-> pointer
690 Consequence of recent symtab clarification. No marks for figuring out
691 who maintains the MIPS.
698 There are never to many testcases.
702 Better thread testsuite.
706 Better C++ testsuite.
710 Look at adding a GDB specific testsuite directory so that white box
711 tests of key internals can be added (eg ui_file).
715 Separate out tests that involve the floating point (FP).
717 (Something for people brining up new targets). FP and non-fp tests
718 are combined. I think there should be set of basic tests that
719 exercise pure integer support and then a more expanded set that
720 exercise FP and FP/integer interactions.
722 As an example, the MIPS, for n32 as problems with passing FP's and
723 structs. Since most inferior call tests include FP it is difficult to
724 determine of the integer tests are ok.
728 Architectural Changes: General
729 ==============================
731 These are harder than simple cleanups / fixes and, consequently
732 involve more work. Typically an Architectural Change will be broken
733 down into a more digestible set of cleanups and fixes.
737 Cleanup software single step.
739 At present many targets implement software single step by directly
740 blatting memory (see rs6000-tdep.c). Those targets should register
741 the applicable breakpoints using the breakpoint framework. Perhaphs a
742 new internal breakpoint class ``step'' is needed.
746 Replace READ_FP() with FRAME_HANDLE().
748 READ_FP() is a hangover from the days of the vax when the ABI really
749 did have a frame pointer register. Modern architectures typically
750 construct a virtual frame-handle from the stack pointer and various
751 other bits of string.
753 Unfortunatly GDB still treats this synthetic FP register as though it
754 is real. That in turn really confuses users (arm and ``print $fp'' VS
755 ``info registers fp''). The synthetic FP should be separated out of
756 the true register set presented to the user.
760 Register Cache Cleanup (below from Andrew Cagney)
762 I would depict the current register architecture as something like:
769 register + REGISTER_BYTE(reg_nr)
772 -------------------------
773 | extern register[] |
774 -------------------------
776 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
777 really clear on what mechanisms they should be using to manipulate that
778 buffer. Further, much code assumes, dangerously, that registers are
779 contigious. Having got mips-tdep.c to support multiple ABIs, believe
780 me, that is a bad assumption. Finally, that register cache layout is
781 determined by the current remote/local target and _not_ the less
782 specific target ISA. In fact, in many cases it is determined by the
783 somewhat arbitrary layout of the [gG] packets!
786 How I would like the register file to work is more like:
814 The main objectives being:
816 o a clear separation between the low
817 level target and the high level GDB
819 o a mechanism that solves the general
820 problem of register aliases, overlaps
821 etc instead of treating them as optional
822 extras that can be wedged in as an after
823 thought (that is a reasonable description
824 of the current code).
826 Identify then solve the hard case and the
827 rest just falls out. GDB solved the easy
828 case and then tried to ignore the real
831 o a removal of the assumption that the
832 mapping between the register cache
833 and virtual registers is largely static.
834 If you flip the USR/SSR stack register
835 select bit in the status-register then
836 the corresponding stack registers should
839 o a mechanism that clearly separates the
840 gdb internal register cache from any
841 target (not architecture) dependant
842 specifics such as [gG] packets.
844 Of course, like anything, it sounds good in theory. In reality, it
845 would have to contend with many<->many relationships at both the
846 virt<->cache and cache<->target level. For instance:
849 Modifying an mmx register may involve
850 scattering values across both FP and
851 mmpx specific parts of a buffer
854 When writing back a SP it may need to
855 both be written to both SP and USP.
860 Rather than let this like the last time it was discussed, just slip, I'm
861 first going to add this e-mail (+ references) to TODO. I'd then like to
862 sketch out a broad strategy I think could get us there.
865 First thing I'd suggest is separating out the ``extern registers[]''
866 code so that we can at least identify what is using it. At present
867 things are scattered across many files. That way we can at least
868 pretend that there is a cache instead of a global array :-)
870 I'd then suggest someone putting up a proposal for the pseudo-reg /
871 high-level side interface so that code can be adopted to it. For old
872 code, initially a blanket rename of write_register_bytes() to
873 deprecated_write_register_bytes() would help.
875 Following that would, finaly be the corresponding changes to the target.
879 Check that GDB can handle all BFD architectures (Andrew Cagney)
881 There should be a test that checks that BFD/GDB are in sync with
882 regard to architecture changes. Something like a test that first
883 queries GDB for all supported architectures and then feeds each back
884 to GDB.. Anyone interested in learning how to write tests? :-)
888 Architectural Change: Multi-arch et al.
889 =======================================
891 The long term objective is to remove all assumptions that there is a
892 single target with a single address space with a single instruction
893 set architecture and single application binary interface.
895 This is an ongoing effort. The first milestone is to enable
896 ``multi-arch'' where by all architectural decisions are made at
899 It should be noted that ``gdbarch'' is really ``gdbabi'' and
900 ``gdbisa''. Once things are multi-arched breaking that down correctly
901 will become much easier.
905 GDBARCH cleanup (Andrew Cagney)
907 The non-generated parts of gdbarch.{sh,h,c} should be separated out
908 into arch-utils.[hc].
910 Document that gdbarch_init_ftype could easily fail because it didn't
911 identify an architecture.
915 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
917 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
922 Fix target_signal_from_host() etc.
924 The name is wrong for starters. ``target_signal'' should probably be
925 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
926 After that it needs to be multi-arched and made independant of any
927 host signal numbering.
931 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
934 This is a barrier to replacing mips_extra_func_info with something
935 that works with multi-arch.
939 Multi-arch mips_extra_func_info.
941 This first needs the alpha to be updated so that it uses ``struct
946 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
948 Surely one of them is redundant.
952 Convert ALL architectures to MULTI-ARCH.
956 Select the initial multi-arch ISA / ABI based on --target or similar.
958 At present the default is based on what ever is first in the BFD
959 archures table. It should be determined based on the ``--target=...''
964 Make MIPS pure multi-arch.
966 It is only at the multi-arch enabled stage.
972 Enable the code to recognize --enable-targets=.... like BINUTILS does.
974 Can the tm.h and nm.h files be eliminated by multi-arch.
978 Architectural Change: MI, LIBGDB and scripting languages
979 ========================================================
981 See also architectural changes related to the event loop. LIBGDB
982 can't be finished until there is a generic event loop being used by
985 The long term objective is it to be possible to integrate GDB into
990 Implement generic ``(gdb) commmand > file''
992 Once everything is going through ui_file it should be come fairly
995 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
999 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1001 gdb_stdtarg is easily confused with gdb_stdarg.
1005 Extra ui_file methods - dump.
1007 Very useful for whitebox testing.
1011 Eliminate error_begin().
1013 With ui_file, there is no need for the statefull error_begin ()
1018 Send normal output to gdb_stdout.
1019 Send error messages to gdb_stderror.
1020 Send debug and log output log gdb_stdlog.
1022 GDB still contains many cases where (f)printf or printf_filtered () is
1023 used when it should be sending the messages to gdb_stderror or
1024 gdb_stdlog. The thought of #defining printf to something has crossed
1029 Re-do GDB's output pager.
1031 GDB's output pager still relies on people correctly using *_filtered
1032 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1033 Hopefully, with all normal output going to gdb_stdout, the pager can
1034 just look at the ui_file that the output is on and then use that to
1035 decide what to do about paging. Sounds good in theory.
1039 Check/cleanup MI documentation.
1041 The list of commands specified in the documentation needs to be
1042 checked against the mi-cmds.c table in a mechanical way (so that they
1043 two can be kept up-to-date).
1047 Convert MI into libgdb
1049 MI provides a text interface into what should be many of the libgdb
1050 functions. The implementation of those functions should be separated
1051 into the MI interface and the functions proper. Those functions being
1052 moved to gdb/lib say.
1058 The first part can already be found in defs.h.
1062 MI's input does not use buffering.
1064 At present the MI interface reads raw characters of from an unbuffered
1065 FD. This is to avoid several nasty buffer/race conditions. That code
1066 should be changed so that it registers its self with the event loop
1067 (on the input FD) and then push commands up to MI as they arrive.
1069 The serial code already does this.
1073 Make MI interface accessable from existing CLI.
1077 Add a breakpoint-edit command to MI.
1079 It would be similar to MI's breakpoint create but would apply to an
1080 existing breakpoint. It saves the need to delete/create breakpoints
1081 when ever they are changed.
1085 Add directory path to MI breakpoint.
1087 That way the GUI's task of finding the file within which the
1088 breakpoint was set is simplified.
1092 Add a mechanism to reject certain expression classes to MI
1094 There are situtations where you don't want GDB's expression
1095 parser/evaluator to perform inferior function calls or variable
1096 assignments. A way of restricting the expression parser so that such
1097 operations are not accepted would be very helpful.
1101 Remove sideffects from libgdb breakpoint create function.
1103 The user can use the CLI to create a breakpoint with partial
1104 information - no file (gdb would use the file from the last
1107 The libgdb interface currently affects that environment which can lead
1108 to confusion when a user is setting breakpoints via both the MI and
1111 This is also a good example of how getting the CLI ``right'' will be
1116 Move gdb_lasterr to ui_out?
1118 The way GDB throws errors and records them needs a re-think. ui_out
1119 handles the correct output well. It doesn't resolve what to do with
1120 output / error-messages when things go wrong.
1124 do_setshow_command contains a 1024 byte buffer.
1126 The function assumes that there will never be any more than 1024 bytes
1127 of enum. It should use mem_file.
1131 Should struct cmd_list_element . completer take the command as an
1136 Should the bulk of top.c:line_completion_function() be moved to
1137 command.[hc]? complete_on_cmdlist() and complete_on_enums() could
1138 then be made private.
1142 top.c (execute_command): Should a command being valid when the target
1143 is running be made an attribute (predicate) to the command rather than
1144 an explicit set of tests.
1148 top.c (execute_command): Should the bulk of this function be moved
1149 into command.[hc] so that top.c doesn't grub around in the command
1154 Architectural Change: Async
1155 ===========================
1157 While GDB uses an event loop when prompting the user for input. That
1158 event loop is not exploited by targets when they allow the target
1159 program to continue. Typically targets still block in (target_wait())
1160 until the program again halts.
1162 The closest a target comes to supporting full asynchronous mode are
1163 the remote targets ``async'' and ``extended-async''.
1167 Asynchronous expression evaluator
1169 Inferior function calls hang GDB.
1173 Fix implementation of ``target xxx''.
1175 At present when the user specifies ``target xxxx'', the CLI maps that
1176 directly onto a target open method. It is then assumed that the
1177 target open method should do all sorts of complicated things as this
1178 is the only chance it has. Check how the various remote targets
1179 duplicate the target operations. Check also how the various targets
1180 behave differently for purely arbitrary reasons.
1182 What should happen is that ``target xxxx'' should call a generic
1183 ``target'' function and that should then co-ordinate the opening of
1184 ``xxxx''. This becomes especially important when you're trying to
1185 open an asynchronous target that may need to perform background tasks
1186 as part of the ``attach'' phase.
1188 Unfortunatly, due to limitations in the old/creaking command.h
1189 interface, that isn't possible. The function being called isn't told
1190 of the ``xxx'' or any other context information.
1192 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1193 CLI code so that it passes to the callback function (attatched to a
1194 command) useful information such as the actual command and a context
1195 for that command. Other changes such as making ``struct command''
1196 opaque may also help.
1199 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1203 Make "target xxx" command interruptible.
1205 As things become async this becomes possible. A target would start
1206 the connect and then return control to the event loop. A cntrl-c
1207 would notify the target that the operation is to be abandoned and the
1208 target code could respond.
1212 Add a "suspend" subcommand of the "continue" command to suspend gdb
1213 while continuing execution of the subprocess. Useful when you are
1214 debugging servers and you want to dodge out and initiate a connection
1215 to a server running under gdb.
1224 Frequently requested but not approved requests.
1228 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1230 The benefits on this one are thought to be marginal - GDBs design
1231 means that unused parameters are very common. GCC 3.0 will also
1232 include the option -Wno-unused-parameter which means that ``-Wall
1233 -Wno-unused-parameters -Werror'' can be specified.
1242 This list is not up to date, and opinions vary about the importance or
1243 even desirability of some of the items. If you do fix something, it
1244 always pays to check the below.
1248 @c This does not work (yet if ever). FIXME.
1249 @c @item --parse=@var{lang} @dots{}
1250 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1251 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1252 @c list of all supported languages, omit the argument. Without this
1253 @c option, @value{GDBN} is configured to parse all supported languages.
1257 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1258 is its default value. Clean this up.
1262 It should be possible to use symbols from shared libraries before we know
1263 exactly where the libraries will be loaded. E.g. "b perror" before running
1264 the program. This could maybe be done as an extension of the "breakpoint
1265 re-evaluation" after new symbols are loaded.
1269 Make single_step() insert and remove breakpoints in one operation.
1271 [If this is talking about having single_step() insert the breakpoints,
1272 run the target then pull the breakpoints then it is wrong. The
1273 function has to return as control has to eventually be passed back to
1274 the main event loop.]
1278 Speed up single stepping by avoiding extraneous ptrace calls.
1282 Speed up single stepping by not inserting and removing breakpoints
1283 each time the inferior starts and stops.
1285 Breakpoints should not be inserted and deleted all the time. Only the
1286 one(s) there should be removed when we have to step over one. Support
1287 breakpoints that don't have to be removed to step over them.
1289 [this has resulted in numerous debates. The issue isn't clear cut]
1293 Provide "voodoo" debugging of core files. This creates a zombie
1294 process as a child of the debugger, and loads it up with the data,
1295 stack, and regs of the core file. This allows you to call functions
1296 in the executable, to manipulate the data in the core file.
1302 GDB reopens the source file on every line, as you "next" through it.
1304 [still true? I've a memory of this being fixed]
1308 Perhaps "i source" should take an argument like that of "list".
1312 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1313 it matches the source line indicated.
1317 The prompt at end of screen should accept space as well as CR.
1321 Backtrace should point out what the currently selected frame is, in
1322 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1323 ...)" rather than "#3 foo (bar, ...)".
1327 "i program" should work for core files, and display more info, like what
1328 actually caused it to die.
1332 "x/10i" should shorten the long name, if any, on subsequent lines.
1336 "next" over a function that longjumps, never stops until next time you happen
1337 to get to that spot by accident. E.g. "n" over execute_command which has
1342 "set zeroprint off", don't bother printing members of structs which
1343 are entirely zero. Useful for those big structs with few useful
1348 GDB does four ioctl's for every command, probably switching terminal modes
1349 to/from inferior or for readline or something.
1353 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1354 if the state is the same, too.
1358 "i frame" shows wrong "arglist at" location, doesn't show where the args
1359 should be found, only their actual values.
1363 There should be a way for "set" commands to validate the new setting
1364 before it takes effect.
1368 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1372 i line VAR produces "Line number not known for symbol ``var''.". I
1373 thought we were stashing that info now!
1377 We should be able to write to random files at hex offsets like adb.
1381 [elena - delete this]
1383 Handle add_file with separate text, data, and bss addresses. Maybe
1384 handle separate addresses for each segment in the object file?
1388 [Jimb/Elena delete this one]
1390 Handle free_named_symtab to cope with multiply-loaded object files
1391 in a dynamic linking environment. Should remember the last copy loaded,
1392 but not get too snowed if it finds references to the older copy.
1396 [elena delete this also]
1398 Remove all references to:
1405 now that we have BFD. All remaining are in machine dependent files.
1409 Re-organize help categories into things that tend to fit on a screen
1414 Add in commands like ADB's for searching for patterns, etc. We should
1415 be able to examine and patch raw unsymboled binaries as well in gdb as
1416 we can in adb. (E.g. increase the timeout in /bin/login without source).
1418 [actually, add ADB interface :-]
1422 When doing "step" or "next", if a few lines of source are skipped between
1423 the previous line and the current one, print those lines, not just the
1424 last line of a multiline statement.
1428 Handling of "&" address-of operator needs some serious overhaul
1429 for ANSI C and consistency on arrays and functions.
1430 For "float point[15];":
1431 ptype &point[4] ==> Attempt to take address of non-lvalue.
1432 For "char *malloc();":
1433 ptype malloc ==> "char *()"; should be same as
1434 ptype &malloc ==> "char *(*)()"
1435 call printf ("%x\n", malloc) ==> weird value, should be same as
1436 call printf ("%x\n", &malloc) ==> correct value
1440 Fix dbxread.c symbol reading in the presence of interrupts. It
1441 currently leaves a cleanup to blow away the entire symbol table when a
1442 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1444 [I suspect that the grype was that, on a slow system, you might want
1445 to cntrl-c and get just half the symbols and then load the rest later
1446 - scary to be honest]
1450 Mipsread.c reads include files depth-first, because the dependencies
1451 in the psymtabs are way too inclusive (it seems to me). Figure out what
1452 really depends on what, to avoid recursing 20 or 30 times while reading
1457 value_add() should be subtracting the lower bound of arrays, if known,
1458 and possibly checking against the upper bound for error reporting.
1462 When listing source lines, check for a preceding \n, to verify that
1463 the file hasn't changed out from under us.
1465 [fixed by some other means I think. That hack wouldn't actually work
1466 reliably - the file might move such that another \n appears. ]
1470 Get all the remote systems (where the protocol allows it) to be able to
1471 stop the remote system when the GDB user types ^C (like remote.c
1472 does). For ebmon, use ^Ak.
1476 Possible feature: A version of the "disassemble" command which shows
1477 both source and assembly code ("set symbol-filename on" is a partial
1480 [has this been done? It was certainly done for MI and GDBtk]
1484 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1485 using a 0 address for bad purposes internally).
1489 Make "info path" and path_command work again (but independent of the
1490 environment either of gdb or that we'll pass to the inferior).
1494 Make GDB understand the GCC feature for putting octal constants in
1495 enums. Make it so overflow on an enum constant does not error_type
1496 the whole type. Allow arbitrarily large enums with type attributes.
1497 Put all this stuff in the testsuite.
1501 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1502 the value in hex; process type attributes). Add this to the
1503 testsuite. This way future compilers can add new types and old
1504 versions of GDB can do something halfway reasonable.
1508 Fix mdebugread.c:parse_type to do fundamental types right (see
1509 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1510 that the debug format fixes the sizes of these things and it shouldn't
1511 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1512 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1513 be aware of that). Also use a switch statement for clarity and speed.
1517 Investigate adding symbols in target_load--some targets do, some
1522 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1523 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1528 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1533 Make a watchpoint which contains a function call an error (it is
1534 broken now, making it work is probably not worth the effort).
1538 New test case based on weird.exp but in which type numbers are not
1539 renumbered (thus multiply defining a type). This currently causes an
1540 infinite loop on "p v_comb".
1544 [Hey! Hint Hint Delete Delete!!!]
1546 Fix 386 floating point so that floating point registers are real
1547 registers (but code can deal at run-time if they are missing, like
1548 mips and 68k). This would clean up "info float" and related stuff.
1552 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1553 about not being able to access memory location 0.
1555 -------------------- enummask.c
1576 If try to modify value in file with "set write off" should give
1577 appropriate error not "cannot access memory at address 0x65e0".
1581 Allow core file without exec file on RS/6000.
1585 Make sure "shell" with no arguments works right on DOS.
1589 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1590 the same directory can be NFS-mounted on unix or DOS, and work the
1595 [Is this another delete???]
1597 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
1598 get RS/6000 to work right, might not be immediately relevant).
1602 Work out some kind of way to allow running the inferior to be done as
1603 a sub-execution of, eg. breakpoint command lists. Currently running
1604 the inferior interupts any command list execution. This would require
1605 some rewriting of wait_for_inferior & friends, and hence should
1606 probably be done in concert with the above.
1610 Add function arguments to gdb user defined functions.
1614 Add convenience variables that refer to exec file, symbol file,
1615 selected frame source file, selected frame function, selected frame
1620 Modify the handling of symbols grouped through BINCL/EINCL stabs to
1621 allocate a partial symtab for each BINCL/EINCL grouping. This will
1622 seriously decrease the size of inter-psymtab dependencies and hence
1623 lessen the amount that needs to be read in when a new source file is
1628 Add a command for searching memory, a la adb. It specifies size,
1629 mask, value, start address. ADB searches until it finds it or hits
1630 an error (or is interrupted).
1634 Remove the range and type checking code and documentation, if not