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 I know there are problems with single stepping through signal
39 handlers. These problems were present in 4.18. They were just masked
40 because 4.18 failed to recognize signal handlers. Fixing it is not
41 easy, and will require changes to handle_inferior_event(), that I
42 prefer not to make before the 5.0 release.
48 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
49 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
51 (Broken) support for GDB's remote protocol across UDP is to be
52 included in the follow-on release.
56 Can't build IRIX -> arm GDB.
57 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
60 > Now I'm building for an embedded arm target. If there is a way of turning
61 > remote-rdi off, I couldn't find it. It looks like it gets built by default
62 > in gdb/configure.tgt(line 58) Anyway, the build dies in
63 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
64 > aren't one of the architectures supported.
68 Problem with weak functions
69 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
71 Dan Nicolaescu writes:
72 > It seems that gdb-4.95.1 does not display correctly the function when
73 > stoping in weak functions.
75 > It stops in a function that is defined as weak, not in the function
76 > that is actualy run...
80 GDB 5.0 doesn't work on Linux/SPARC
84 Code Cleanups: Next Release
85 ===========================
87 The following are small cleanups that will hopefully be completed by
92 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
94 Patches in the database.
100 Eliminate all uses of PARAMS in GDB's source code.
104 Fix copyright notices.
106 Turns out that ``1998-2000'' isn't considered valid :-(
108 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
112 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
113 determine the default isa/byte-order.
117 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependant
118 BIG_ENDIAN and LITTLE_ENDIAN.
122 Code Cleanups: General
123 ======================
125 The following are more general cleanups and fixes. They are not tied
126 to any specific release.
130 Eliminate more compiler warnings.
132 Of course there also needs to be the usual debate over which warnings
133 are valid and how to best go about this.
135 One method: choose a single option; get agreement that it is
136 reasonable; try it out to see if there isn't anything silly about it
137 (-Wunused-parameters is an example of that) then incrementally hack
140 The other method is to enable all warnings and eliminate them from one
145 Elimination of ``(catch_errors_ftype *) func''.
147 Like make_cleanup_func it isn't portable.
148 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
149 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
158 [PATCH/5] src/intl/Makefile.in:distclean additions
159 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
161 Do not forget to merge the patch back into the trunk.
165 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
167 At present defs.h includes <endian.h> (which is linux specific) yet
168 almost nothing depends on it. Suggest "gdb_endian.h" which can also
169 handle <machine/endian.h> and only include that where it is really
174 Replace asprintf() calls with xasprintf() calls.
176 As with things like strdup() most calls to asprintf() don't check the
181 Replace strsave() + mstrsave() with libiberty:xstrdup().
185 Replace savestring() with something from libiberty.
187 An xstrldup()? but that would have different semantics.
191 Rationalize use of floatformat_unknown in GDB sources.
193 Instead of defaulting to floatformat_unknown, should hosts/targets
194 specify the value explicitly?
196 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
200 Add a ``name'' member to include/floatformat.h:struct floatformat.
201 Print that name in gdbarch.c.
205 Sort out the harris mess in include/floatformat.h (it hardwires two
206 different floating point formats).
210 See of the GDB local floatformat_do_doublest() and libiberty's
211 floatformat_to_double (which was once GDB's ...) can be merged some
216 Eliminate mmalloc() from GDB.
218 Also eliminate it from defs.h.
222 Eliminate PTR. ISO-C allows ``void *''.
228 GDB should never abort. GDB should either throw ``error ()'' or
229 ``internal_error ()''. Better still GDB should naturally unwind with
234 GDB probably doesn't build on FreeBSD pre 2.2.x
235 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
237 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
244 Readline 4.? is out. A merge wouldn't hurt. Patches are in:
246 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00436.html
250 Deprecate "fg". Apparently ``fg'' is actually continue.
252 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
256 Deprecate current use of ``floatformat_unknown''.
258 Require all targets to explicitly provide their float format instead
259 of defaulting to floatformat unknown. Doing the latter leads to nasty
262 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
266 Rationalize floatformat_to_double() vs floatformat_to_doublest().
268 Looks like GDB migrated floatformat_to_double() to libiberty but then
269 turned around and created a ..._to_doublest() the latter containing
272 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
276 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
278 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
282 Follow through `make check' with --enable-shared.
284 When the srcware tree is configured with --enable-shared, the `expect'
285 program won't run properly. Jim Wilson found out gdb has a local hack
286 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
289 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
293 The ``maintenance deprecate set endian big'' command doesn't notice
294 that it is deprecating ``set endian'' and not ``set endian big'' (big
295 is implemented using an enum). Is anyone going to notice this?
299 When tab expanding something like ``set arch<tab>'' ignore the
300 deprecated ``set archdebug'' and expand to ``set architecture''.
304 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
305 like from arm-tdep.c.
309 Fix uses of ->function.cfunc = set_function().
311 The command.c code calls sfunc() when a set command. Rather than
312 change it suggest fixing the callback function so that it is more
315 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
317 See also ``Fix implementation of ``target xxx''.'' below.
321 IRIX 3.x support is probably broken.
325 New Features and Fixes
326 ======================
328 These are harder than cleanups but easier than work involving
329 fundamental architectural change.
333 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
334 so that you can see how the GDB was created.
336 Some of these (*m.h) would be added to the generated config.h. That
337 in turn would fix a long standing bug where by the build process many
338 not notice a changed tm.h file. Since everything depends on config.h,
339 a change to *m.h forces a change to config.h and, consequently forces
344 Add an "info bfd" command that displays supported object formats,
345 similarly to objdump -i.
347 Is there a command already?
351 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
353 This requires internationalization.
357 Convert GDB build process to AUTOMAKE.
359 See also sub-directory configure below.
361 The current convention is (kind of) to use $(<header>_h) in all
362 dependency lists. It isn't done in a consistent way.
366 Cleanup configury support for optional sub-directories.
368 Check how GCC handles multiple front ends for an example of how things
369 could work. A tentative first step is to rationalize things so that
370 all sub directories are handled in a fashion similar to gdb/mi.
372 See also automake above.
376 Restructure gdb directory tree so that it avoids any 8.3 and 14
381 Add a transcript mechanism to GDB.
383 Such a mechanism might log all gdb input and output to a file in a
384 form that would allow it to be replayed. It could involve ``gdb
385 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
389 Can the xdep files be replaced by autoconf?
393 Document trace machinery
397 Document ui-out and ui-file.
399 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
403 Update texinfo.tex to latest?
409 Incorporate agentexpr.texi into gdb.texinfo
411 agentexpr.texi mostly describes the details of the byte code used for
412 tracepoints, not the internals of the support for this in GDB. So it
413 looks like gdb.texinfo is a better place for this information.
415 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
419 Document overlay machinery.
423 ``(gdb) catch signal SIGNAL''
425 Overlaps with ``handle SIGNAL'' but the implied behavour is different.
426 You can attach commands to a catch but not a handle. A handle has a
427 limited number of hardwired actions.
431 Get the TUI working on all platforms.
435 Add support for ``gdb --- PROGRAM ARGS ...''.
436 Add support for ``gdb -cmd=...''
438 Along with many variations. Check:
440 ????? for a full discussion.
446 Implement ``(gdb) !ls''.
448 Which is very different from ``(gdb) ! ls''. Implementing the latter
451 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
455 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
460 When tab expanding something like ``set arch<tab>'' ignore the
461 deprecated ``set archdebug'' and expand to ``set architecture''.
465 Replace the code that uses the host FPU with an emulator of the target
475 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
476 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
478 The thread_db assisted debugging code doesn't handle exiting threads
479 properly, at least in combination with glibc 2.1.3 (the framework is
480 there, just not the actual code). There are at least two problems
481 that prevent this from working.
483 As an additional reference point, the pre thread_db code did not work
488 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
490 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
492 Christopher Blizzard writes:
494 So, I've done some more digging into this and it looks like Jim
495 Kingdon has reported this problem in the past:
497 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
499 I can reproduce this problem both with and without Tom's patch. Has
500 anyone seen this before? Maybe have a solution for it hanging around?
503 There's a test case for this documented at:
505 when debugging threaded applications you get extra SIGTRAPs
506 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
508 [There should be a GDB testcase - cagney]
512 GDB5 TOT on unixware 7
513 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
516 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
517 > practical matter, the current thread support is somewhat more annoying
518 > than when GDB was thread-unaware.
522 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
524 Add support for packet enable/disable commands with these thread
525 packets. General cleanup.
527 [PATCH] Document the ThreadInfo remote protocol queries
528 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
530 [PATCH] "info threads" queries for remote.c
531 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
538 New languages come onto the scene all the time.
542 Pascal (Pierre Muller, David Taylor)
544 Pierre Muller has contributed patches for adding Pascal Language
547 2 pascal language patches inserted in database
548 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
551 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
555 Java (Anthony Green, David Taylor)
557 Anthony Green has a number of Java patches that did not make it into
561 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
564 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
566 Patch: handle N_MAIN stab
567 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
573 Modify gdb to work correctly with Pascal.
577 Re: Various C++ things
579 value_headof/value_from_vtable_info are worthless, and should be
580 removed. The one place in printcmd.c that uses it should use the RTTI
583 RTTI for g++ should be using the typeinfo functions rather than the
584 vtables. The typeinfo functions are always at offset 4 from the
585 beginning of the vtable, and are always right. The vtables will have
586 weird names like E::VB sometimes. The typeinfo function will always
587 be "E type_info function", or somesuch.
589 value_virtual_fn_field needs to be fixed so there are no failures for
590 virtual functions for C++ using g++.
592 Testsuite cases are the major priority right now for C++ support,
593 since i have to make a lot of changes that could potentially break
598 Add support for Modula3
600 Get DEC/Compaq to contribute their Modula-3 support.
604 Remote Protocol Support
605 =======================
609 set/show remote X-packet ...
611 ``(gdb) help set remote X-packet'' doesn't list the applicable
612 responses. The help message needs to be expanded.
616 Remote protocol doco feedback.
618 Too much feedback to mention needs to be merged in (901660). Search
619 for the word ``remote''.
622 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
623 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
624 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
628 GDB doesn't recover gracefully from remote protocol errors.
630 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
631 ignored and a timeout is required before GDB retries. A pre-cursor to
632 fixing this this is making GDB's remote protocol packet more robust.
634 While downloading to a remote protocol target, gdb ignores packet
635 errors in so far as it will continue to edownload with chunk N+1 even
636 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
637 take a painfully long time to run. As a PS that test needs to be
638 fixed so that it builds on 16 bit machines.
642 Add the cycle step command.
644 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
648 Resolve how to scale things to support very large packets.
652 Resolve how to handle a target that changes things like its endianess
653 on the fly - should it be returned in the ``T'' packet?
655 Underlying problem is that the register file is target endian. If the
656 target endianess changes gdb doesn't know.
663 If / when GDB starts to support the debugging of multi-processor
664 (rather than multi-thread) applications the symtab code will need to
665 be updated a little so that several independant symbol tables are
666 active at a given time.
668 The other interesting change is a clarification of the exact meaning
669 of CORE_ADDR and that has had consequences for a few targets (that
670 were abusing that data type).
674 Investiagate ways of reducing memory.
678 Investigate ways of improving load time.
682 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
684 Consequence of recent symtab clarification. No marks for figuring out
685 who maintains the d10v.
689 Get the MIPS to correctly sign extend all address <-> pointer
692 Consequence of recent symtab clarification. No marks for figuring out
693 who maintains the MIPS.
700 There are never to many testcases.
704 Better thread testsuite.
708 Better C++ testsuite.
712 Look at adding a GDB specific testsuite directory so that white box
713 tests of key internals can be added (eg ui_file).
717 Separate out tests that involve the floating point (FP).
719 (Something for people brining up new targets). FP and non-fp tests
720 are combined. I think there should be set of basic tests that
721 exercise pure integer support and then a more expanded set that
722 exercise FP and FP/integer interactions.
724 As an example, the MIPS, for n32 as problems with passing FP's and
725 structs. Since most inferior call tests include FP it is difficult to
726 determine of the integer tests are ok.
730 Architectural Changes: General
731 ==============================
733 These are harder than simple cleanups / fixes and, consequently
734 involve more work. Typically an Architectural Change will be broken
735 down into a more digestible set of cleanups and fixes.
739 Cleanup software single step.
741 At present many targets implement software single step by directly
742 blatting memory (see rs6000-tdep.c). Those targets should register
743 the applicable breakpoints using the breakpoint framework. Perhaphs a
744 new internal breakpoint class ``step'' is needed.
748 Replace READ_FP() with FRAME_HANDLE().
750 READ_FP() is a hangover from the days of the vax when the ABI really
751 did have a frame pointer register. Modern architectures typically
752 construct a virtual frame-handle from the stack pointer and various
753 other bits of string.
755 Unfortunatly GDB still treats this synthetic FP register as though it
756 is real. That in turn really confuses users (arm and ``print $fp'' VS
757 ``info registers fp''). The synthetic FP should be separated out of
758 the true register set presented to the user.
762 Register Cache Cleanup (below from Andrew Cagney)
764 I would depict the current register architecture as something like:
771 register + REGISTER_BYTE(reg_nr)
774 -------------------------
775 | extern register[] |
776 -------------------------
778 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
779 really clear on what mechanisms they should be using to manipulate that
780 buffer. Further, much code assumes, dangerously, that registers are
781 contigious. Having got mips-tdep.c to support multiple ABIs, believe
782 me, that is a bad assumption. Finally, that register cache layout is
783 determined by the current remote/local target and _not_ the less
784 specific target ISA. In fact, in many cases it is determined by the
785 somewhat arbitrary layout of the [gG] packets!
788 How I would like the register file to work is more like:
816 The main objectives being:
818 o a clear separation between the low
819 level target and the high level GDB
821 o a mechanism that solves the general
822 problem of register aliases, overlaps
823 etc instead of treating them as optional
824 extras that can be wedged in as an after
825 thought (that is a reasonable description
826 of the current code).
828 Identify then solve the hard case and the
829 rest just falls out. GDB solved the easy
830 case and then tried to ignore the real
833 o a removal of the assumption that the
834 mapping between the register cache
835 and virtual registers is largely static.
836 If you flip the USR/SSR stack register
837 select bit in the status-register then
838 the corresponding stack registers should
841 o a mechanism that clearly separates the
842 gdb internal register cache from any
843 target (not architecture) dependant
844 specifics such as [gG] packets.
846 Of course, like anything, it sounds good in theory. In reality, it
847 would have to contend with many<->many relationships at both the
848 virt<->cache and cache<->target level. For instance:
851 Modifying an mmx register may involve
852 scattering values across both FP and
853 mmpx specific parts of a buffer
856 When writing back a SP it may need to
857 both be written to both SP and USP.
862 Rather than let this like the last time it was discussed, just slip, I'm
863 first going to add this e-mail (+ references) to TODO. I'd then like to
864 sketch out a broad strategy I think could get us there.
867 First thing I'd suggest is separating out the ``extern registers[]''
868 code so that we can at least identify what is using it. At present
869 things are scattered across many files. That way we can at least
870 pretend that there is a cache instead of a global array :-)
872 I'd then suggest someone putting up a proposal for the pseudo-reg /
873 high-level side interface so that code can be adopted to it. For old
874 code, initially a blanket rename of write_register_bytes() to
875 deprecated_write_register_bytes() would help.
877 Following that would, finaly be the corresponding changes to the target.
881 Check that GDB can handle all BFD architectures (Andrew Cagney)
883 There should be a test that checks that BFD/GDB are in sync with
884 regard to architecture changes. Something like a test that first
885 queries GDB for all supported architectures and then feeds each back
886 to GDB.. Anyone interested in learning how to write tests? :-)
890 Architectural Change: Multi-arch et al.
891 =======================================
893 The long term objective is to remove all assumptions that there is a
894 single target with a single address space with a single instruction
895 set architecture and single application binary interface.
897 This is an ongoing effort. The first milestone is to enable
898 ``multi-arch'' where by all architectural decisions are made at
901 It should be noted that ``gdbarch'' is really ``gdbabi'' and
902 ``gdbisa''. Once things are multi-arched breaking that down correctly
903 will become much easier.
907 GDBARCH cleanup (Andrew Cagney)
909 The non-generated parts of gdbarch.{sh,h,c} should be separated out
910 into arch-utils.[hc].
912 Document that gdbarch_init_ftype could easily fail because it didn't
913 identify an architecture.
917 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
919 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
924 Fix target_signal_from_host() etc.
926 The name is wrong for starters. ``target_signal'' should probably be
927 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
928 After that it needs to be multi-arched and made independant of any
929 host signal numbering.
933 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
936 This is a barrier to replacing mips_extra_func_info with something
937 that works with multi-arch.
941 Multi-arch mips_extra_func_info.
943 This first needs the alpha to be updated so that it uses ``struct
948 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
950 Surely one of them is redundant.
954 Convert ALL architectures to MULTI-ARCH.
958 Select the initial multi-arch ISA / ABI based on --target or similar.
960 At present the default is based on what ever is first in the BFD
961 archures table. It should be determined based on the ``--target=...''
966 Make MIPS pure multi-arch.
968 It is only at the multi-arch enabled stage.
974 Enable the code to recognize --enable-targets=.... like BINUTILS does.
976 Can the tm.h and nm.h files be eliminated by multi-arch.
980 Architectural Change: MI, LIBGDB and scripting languages
981 ========================================================
983 See also architectural changes related to the event loop. LIBGDB
984 can't be finished until there is a generic event loop being used by
987 The long term objective is it to be possible to integrate GDB into
992 Implement generic ``(gdb) commmand > file''
994 Once everything is going through ui_file it should be come fairly
997 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1001 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1003 gdb_stdtarg is easily confused with gdb_stdarg.
1007 Extra ui_file methods - dump.
1009 Very useful for whitebox testing.
1013 Eliminate error_begin().
1015 With ui_file, there is no need for the statefull error_begin ()
1020 Send normal output to gdb_stdout.
1021 Send error messages to gdb_stderror.
1022 Send debug and log output log gdb_stdlog.
1024 GDB still contains many cases where (f)printf or printf_filtered () is
1025 used when it should be sending the messages to gdb_stderror or
1026 gdb_stdlog. The thought of #defining printf to something has crossed
1031 Re-do GDB's output pager.
1033 GDB's output pager still relies on people correctly using *_filtered
1034 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1035 Hopefully, with all normal output going to gdb_stdout, the pager can
1036 just look at the ui_file that the output is on and then use that to
1037 decide what to do about paging. Sounds good in theory.
1041 Check/cleanup MI documentation.
1043 The list of commands specified in the documentation needs to be
1044 checked against the mi-cmds.c table in a mechanical way (so that they
1045 two can be kept up-to-date).
1049 Convert MI into libgdb
1051 MI provides a text interface into what should be many of the libgdb
1052 functions. The implementation of those functions should be separated
1053 into the MI interface and the functions proper. Those functions being
1054 moved to gdb/lib say.
1060 The first part can already be found in defs.h.
1064 MI's input does not use buffering.
1066 At present the MI interface reads raw characters of from an unbuffered
1067 FD. This is to avoid several nasty buffer/race conditions. That code
1068 should be changed so that it registers its self with the event loop
1069 (on the input FD) and then push commands up to MI as they arrive.
1071 The serial code already does this.
1075 Make MI interface accessable from existing CLI.
1079 Add a breakpoint-edit command to MI.
1081 It would be similar to MI's breakpoint create but would apply to an
1082 existing breakpoint. It saves the need to delete/create breakpoints
1083 when ever they are changed.
1087 Add directory path to MI breakpoint.
1089 That way the GUI's task of finding the file within which the
1090 breakpoint was set is simplified.
1094 Add a mechanism to reject certain expression classes to MI
1096 There are situtations where you don't want GDB's expression
1097 parser/evaluator to perform inferior function calls or variable
1098 assignments. A way of restricting the expression parser so that such
1099 operations are not accepted would be very helpful.
1103 Remove sideffects from libgdb breakpoint create function.
1105 The user can use the CLI to create a breakpoint with partial
1106 information - no file (gdb would use the file from the last
1109 The libgdb interface currently affects that environment which can lead
1110 to confusion when a user is setting breakpoints via both the MI and
1113 This is also a good example of how getting the CLI ``right'' will be
1118 Move gdb_lasterr to ui_out?
1120 The way GDB throws errors and records them needs a re-think. ui_out
1121 handles the correct output well. It doesn't resolve what to do with
1122 output / error-messages when things go wrong.
1126 Architectural Change: Async
1127 ===========================
1129 While GDB uses an event loop when prompting the user for input. That
1130 event loop is not exploited by targets when they allow the target
1131 program to continue. Typically targets still block in (target_wait())
1132 until the program again halts.
1134 The closest a target comes to supporting full asynchronous mode are
1135 the remote targets ``async'' and ``extended-async''.
1139 Asynchronous expression evaluator
1141 Inferior function calls hang GDB.
1145 Fix implementation of ``target xxx''.
1147 At present when the user specifies ``target xxxx'', the CLI maps that
1148 directly onto a target open method. It is then assumed that the
1149 target open method should do all sorts of complicated things as this
1150 is the only chance it has. Check how the various remote targets
1151 duplicate the target operations. Check also how the various targets
1152 behave differently for purely arbitrary reasons.
1154 What should happen is that ``target xxxx'' should call a generic
1155 ``target'' function and that should then co-ordinate the opening of
1156 ``xxxx''. This becomes especially important when you're trying to
1157 open an asynchronous target that may need to perform background tasks
1158 as part of the ``attach'' phase.
1160 Unfortunatly, due to limitations in the old/creaking command.h
1161 interface, that isn't possible. The function being called isn't told
1162 of the ``xxx'' or any other context information.
1164 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1165 CLI code so that it passes to the callback function (attatched to a
1166 command) useful information such as the actual command and a context
1167 for that command. Other changes such as making ``struct command''
1168 opaque may also help.
1171 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1175 Make "target xxx" command interruptible.
1177 As things become async this becomes possible. A target would start
1178 the connect and then return control to the event loop. A cntrl-c
1179 would notify the target that the operation is to be abandoned and the
1180 target code could respond.
1184 Add a "suspend" subcommand of the "continue" command to suspend gdb
1185 while continuing execution of the subprocess. Useful when you are
1186 debugging servers and you want to dodge out and initiate a connection
1187 to a server running under gdb.
1196 Frequently requested but not approved requests.
1200 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1202 The benefits on this one are thought to be marginal - GDBs design
1203 means that unused parameters are very common. GCC 3.0 will also
1204 include the option -Wno-unused-parameter which means that ``-Wall
1205 -Wno-unused-parameters -Werror'' can be specified.
1214 This list is not up to date, and opinions vary about the importance or
1215 even desirability of some of the items. If you do fix something, it
1216 always pays to check the below.
1220 @c This does not work (yet if ever). FIXME.
1221 @c @item --parse=@var{lang} @dots{}
1222 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1223 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1224 @c list of all supported languages, omit the argument. Without this
1225 @c option, @value{GDBN} is configured to parse all supported languages.
1229 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1230 is its default value. Clean this up.
1234 It should be possible to use symbols from shared libraries before we know
1235 exactly where the libraries will be loaded. E.g. "b perror" before running
1236 the program. This could maybe be done as an extension of the "breakpoint
1237 re-evaluation" after new symbols are loaded.
1241 Make single_step() insert and remove breakpoints in one operation.
1243 [If this is talking about having single_step() insert the breakpoints,
1244 run the target then pull the breakpoints then it is wrong. The
1245 function has to return as control has to eventually be passed back to
1246 the main event loop.]
1250 Speed up single stepping by avoiding extraneous ptrace calls.
1254 Speed up single stepping by not inserting and removing breakpoints
1255 each time the inferior starts and stops.
1257 Breakpoints should not be inserted and deleted all the time. Only the
1258 one(s) there should be removed when we have to step over one. Support
1259 breakpoints that don't have to be removed to step over them.
1261 [this has resulted in numerous debates. The issue isn't clear cut]
1265 Provide "voodoo" debugging of core files. This creates a zombie
1266 process as a child of the debugger, and loads it up with the data,
1267 stack, and regs of the core file. This allows you to call functions
1268 in the executable, to manipulate the data in the core file.
1274 GDB reopens the source file on every line, as you "next" through it.
1276 [still true? I've a memory of this being fixed]
1280 Perhaps "i source" should take an argument like that of "list".
1284 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1285 it matches the source line indicated.
1289 The prompt at end of screen should accept space as well as CR.
1293 Backtrace should point out what the currently selected frame is, in
1294 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1295 ...)" rather than "#3 foo (bar, ...)".
1299 "i program" should work for core files, and display more info, like what
1300 actually caused it to die.
1304 "x/10i" should shorten the long name, if any, on subsequent lines.
1308 "next" over a function that longjumps, never stops until next time you happen
1309 to get to that spot by accident. E.g. "n" over execute_command which has
1314 "set zeroprint off", don't bother printing members of structs which
1315 are entirely zero. Useful for those big structs with few useful
1320 GDB does four ioctl's for every command, probably switching terminal modes
1321 to/from inferior or for readline or something.
1325 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1326 if the state is the same, too.
1330 "i frame" shows wrong "arglist at" location, doesn't show where the args
1331 should be found, only their actual values.
1335 There should be a way for "set" commands to validate the new setting
1336 before it takes effect.
1340 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1344 i line VAR produces "Line number not known for symbol ``var''.". I
1345 thought we were stashing that info now!
1349 We should be able to write to random files at hex offsets like adb.
1353 [elena - delete this]
1355 Handle add_file with separate text, data, and bss addresses. Maybe
1356 handle separate addresses for each segment in the object file?
1360 [Jimb/Elena delete this one]
1362 Handle free_named_symtab to cope with multiply-loaded object files
1363 in a dynamic linking environment. Should remember the last copy loaded,
1364 but not get too snowed if it finds references to the older copy.
1368 [elena delete this also]
1370 Remove all references to:
1377 now that we have BFD. All remaining are in machine dependent files.
1381 Re-organize help categories into things that tend to fit on a screen
1386 Add in commands like ADB's for searching for patterns, etc. We should
1387 be able to examine and patch raw unsymboled binaries as well in gdb as
1388 we can in adb. (E.g. increase the timeout in /bin/login without source).
1390 [actually, add ADB interface :-]
1394 When doing "step" or "next", if a few lines of source are skipped between
1395 the previous line and the current one, print those lines, not just the
1396 last line of a multiline statement.
1400 Handling of "&" address-of operator needs some serious overhaul
1401 for ANSI C and consistency on arrays and functions.
1402 For "float point[15];":
1403 ptype &point[4] ==> Attempt to take address of non-lvalue.
1404 For "char *malloc();":
1405 ptype malloc ==> "char *()"; should be same as
1406 ptype &malloc ==> "char *(*)()"
1407 call printf ("%x\n", malloc) ==> weird value, should be same as
1408 call printf ("%x\n", &malloc) ==> correct value
1412 Fix dbxread.c symbol reading in the presence of interrupts. It
1413 currently leaves a cleanup to blow away the entire symbol table when a
1414 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1416 [I suspect that the grype was that, on a slow system, you might want
1417 to cntrl-c and get just half the symbols and then load the rest later
1418 - scary to be honest]
1422 Mipsread.c reads include files depth-first, because the dependencies
1423 in the psymtabs are way too inclusive (it seems to me). Figure out what
1424 really depends on what, to avoid recursing 20 or 30 times while reading
1429 value_add() should be subtracting the lower bound of arrays, if known,
1430 and possibly checking against the upper bound for error reporting.
1434 When listing source lines, check for a preceding \n, to verify that
1435 the file hasn't changed out from under us.
1437 [fixed by some other means I think. That hack wouldn't actually work
1438 reliably - the file might move such that another \n appears. ]
1442 Get all the remote systems (where the protocol allows it) to be able to
1443 stop the remote system when the GDB user types ^C (like remote.c
1444 does). For ebmon, use ^Ak.
1448 Possible feature: A version of the "disassemble" command which shows
1449 both source and assembly code ("set symbol-filename on" is a partial
1452 [has this been done? It was certainly done for MI and GDBtk]
1456 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1457 using a 0 address for bad purposes internally).
1461 Make "info path" and path_command work again (but independent of the
1462 environment either of gdb or that we'll pass to the inferior).
1466 Make GDB understand the GCC feature for putting octal constants in
1467 enums. Make it so overflow on an enum constant does not error_type
1468 the whole type. Allow arbitrarily large enums with type attributes.
1469 Put all this stuff in the testsuite.
1473 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1474 the value in hex; process type attributes). Add this to the
1475 testsuite. This way future compilers can add new types and old
1476 versions of GDB can do something halfway reasonable.
1480 Fix mdebugread.c:parse_type to do fundamental types right (see
1481 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1482 that the debug format fixes the sizes of these things and it shouldn't
1483 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1484 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1485 be aware of that). Also use a switch statement for clarity and speed.
1489 Investigate adding symbols in target_load--some targets do, some
1494 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1495 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1500 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1505 Make a watchpoint which contains a function call an error (it is
1506 broken now, making it work is probably not worth the effort).
1510 New test case based on weird.exp but in which type numbers are not
1511 renumbered (thus multiply defining a type). This currently causes an
1512 infinite loop on "p v_comb".
1516 [Hey! Hint Hint Delete Delete!!!]
1518 Fix 386 floating point so that floating point registers are real
1519 registers (but code can deal at run-time if they are missing, like
1520 mips and 68k). This would clean up "info float" and related stuff.
1524 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1525 about not being able to access memory location 0.
1527 -------------------- enummask.c
1548 If try to modify value in file with "set write off" should give
1549 appropriate error not "cannot access memory at address 0x65e0".
1553 Allow core file without exec file on RS/6000.
1557 Make sure "shell" with no arguments works right on DOS.
1561 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1562 the same directory can be NFS-mounted on unix or DOS, and work the
1567 [Is this another delete???]
1569 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
1570 get RS/6000 to work right, might not be immediately relevant).
1574 Work out some kind of way to allow running the inferior to be done as
1575 a sub-execution of, eg. breakpoint command lists. Currently running
1576 the inferior interupts any command list execution. This would require
1577 some rewriting of wait_for_inferior & friends, and hence should
1578 probably be done in concert with the above.
1582 Add function arguments to gdb user defined functions.
1586 Add convenience variables that refer to exec file, symbol file,
1587 selected frame source file, selected frame function, selected frame
1592 Modify the handling of symbols grouped through BINCL/EINCL stabs to
1593 allocate a partial symtab for each BINCL/EINCL grouping. This will
1594 seriously decrease the size of inter-psymtab dependencies and hence
1595 lessen the amount that needs to be read in when a new source file is
1600 Add a command for searching memory, a la adb. It specifies size,
1601 mask, value, start address. ADB searches until it finds it or hits
1602 an error (or is interrupted).
1606 Remove the range and type checking code and documentation, if not