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 Code Cleanups: General
113 ======================
115 The following are more general cleanups and fixes. They are not tied
116 to any specific release.
120 Eliminate more compiler warnings.
122 Of course there also needs to be the usual debate over which warnings
123 are valid and how to best go about this.
125 One method: choose a single option; get agreement that it is
126 reasonable; try it out to see if there isn't anything silly about it
127 (-Wunused-parameters is an example of that) then incrementally hack
130 The other method is to enable all warnings and eliminate them from one
135 Elimination of ``(catch_errors_ftype *) func''.
137 Like make_cleanup_func it isn't portable.
138 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
139 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
148 [PATCH/5] src/intl/Makefile.in:distclean additions
149 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
151 Do not forget to merge the patch back into the trunk.
155 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
157 At present defs.h includes <endian.h> (which is linux specific) yet
158 almost nothing depends on it. Suggest "gdb_endian.h" which can also
159 handle <machine/endian.h> and only include that where it is really
164 Replace asprintf() calls with xasprintf() calls.
166 As with things like strdup() most calls to asprintf() don't check the
171 Replace strsave() + mstrsave() with libiberty:xstrdup().
175 Replace savestring() with something from libiberty.
177 An xstrldup()? but that would have different semantics.
181 Rationalize use of floatformat_unknown in GDB sources.
183 Instead of defaulting to floatformat_unknown, should hosts/targets
184 specify the value explicitly?
186 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
190 Add a ``name'' member to include/floatformat.h:struct floatformat.
191 Print that name in gdbarch.c.
195 Sort out the harris mess in include/floatformat.h (it hardwires two
196 different floating point formats).
200 See of the GDB local floatformat_do_doublest() and libiberty's
201 floatformat_to_double (which was once GDB's ...) can be merged some
206 Eliminate mmalloc() from GDB.
208 Also eliminate it from defs.h.
212 Eliminate PTR. ISO-C allows ``void *''.
218 GDB should never abort. GDB should either throw ``error ()'' or
219 ``internal_error ()''. Better still GDB should naturally unwind with
224 GDB probably doesn't build on FreeBSD pre 2.2.x
225 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
227 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
234 Readline 4.? is out. A merge wouldn't hurt. Patches are in:
236 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00436.html
240 Deprecate "fg". Apparently ``fg'' is actually continue.
242 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
246 Deprecate current use of ``floatformat_unknown''.
248 Require all targets to explicitly provide their float format instead
249 of defaulting to floatformat unknown. Doing the latter leads to nasty
252 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
256 Rationalize floatformat_to_double() vs floatformat_to_doublest().
258 Looks like GDB migrated floatformat_to_double() to libiberty but then
259 turned around and created a ..._to_doublest() the latter containing
262 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
266 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
268 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
272 Always build ser-tcp.c.
274 The patch as submitted was just going to add ser-tcp.c to the Alpha's
275 makefile. A better patch is to instead add ser-tcp.c to SER_HARDWARE
276 and make it a standard part of all debuggers.
278 If problems occure then configure.in can sort them out.
280 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00544.html
284 Follow through `make check' with --enable-shared.
286 When the srcware tree is configured with --enable-shared, the `expect'
287 program won't run properly. Jim Wilson found out gdb has a local hack
288 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
291 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
295 Change the parameter ``char *list[]'' (etc) to ``const char (*)[]'' so
296 that dynamic lists from things like gdbarch_printable_names() can be
301 The ``maintenance deprecate set endian big'' command doesn't notice
302 that it is deprecating ``set endian'' and not ``set endian big'' (big
303 is implemented using an enum). Is anyone going to notice this?
307 When tab expanding something like ``set arch<tab>'' ignore the
308 deprecated ``set archdebug'' and expand to ``set architecture''.
313 New Features and Fixes
314 ======================
316 These are harder than cleanups but easier than work involving
317 fundamental architectural change.
321 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
322 so that you can see how the GDB was created.
324 Some of these (*m.h) would be added to the generated config.h. That
325 in turn would fix a long standing bug where by the build process many
326 not notice a changed tm.h file. Since everything depends on config.h,
327 a change to *m.h forces a change to config.h and, consequently forces
332 Add an "info bfd" command that displays supported object formats,
333 similarly to objdump -i.
335 Is there a command already?
339 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
341 This requires internationalization.
345 Convert GDB build process to AUTOMAKE.
347 See also sub-directory configure below.
349 The current convention is (kind of) to use $(<header>_h) in all
350 dependency lists. It isn't done in a consistent way.
354 Cleanup configury support for optional sub-directories.
356 Check how GCC handles multiple front ends for an example of how things
357 could work. A tentative first step is to rationalize things so that
358 all sub directories are handled in a fashion similar to gdb/mi.
360 See also automake above.
364 Restructure gdb directory tree so that it avoids any 8.3 and 14
369 Add a transcript mechanism to GDB.
371 Such a mechanism might log all gdb input and output to a file in a
372 form that would allow it to be replayed. It could involve ``gdb
373 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
377 Can the xdep files be replaced by autoconf?
381 Document trace machinery
385 Document ui-out and ui-file.
387 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
391 Update texinfo.tex to latest?
397 Incorporate agentexpr.texi into gdb.texinfo
399 agentexpr.texi mostly describes the details of the byte code used for
400 tracepoints, not the internals of the support for this in GDB. So it
401 looks like gdb.texinfo is a better place for this information.
403 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
407 Document overlay machinery.
411 ``(gdb) catch signal SIGNAL''
413 Overlaps with ``handle SIGNAL'' but the implied behavour is different.
414 You can attach commands to a catch but not a handle. A handle has a
415 limited number of hardwired actions.
419 Get the TUI working on all platforms.
423 Add support for ``gdb --- PROGRAM ARGS ...''.
424 Add support for ``gdb -cmd=...''
426 Along with many variations. Check:
428 ????? for a full discussion.
434 Implement ``(gdb) !ls''.
436 Which is very different from ``(gdb) ! ls''. Implementing the latter
439 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
443 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
448 When tab expanding something like ``set arch<tab>'' ignore the
449 deprecated ``set archdebug'' and expand to ``set architecture''.
453 Replace the code that uses the host FPU with an emulator of the target
463 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
464 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
466 The thread_db assisted debugging code doesn't handle exiting threads
467 properly, at least in combination with glibc 2.1.3 (the framework is
468 there, just not the actual code). There are at least two problems
469 that prevent this from working.
471 As an additional reference point, the pre thread_db code did not work
476 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
478 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
480 Christopher Blizzard writes:
482 So, I've done some more digging into this and it looks like Jim
483 Kingdon has reported this problem in the past:
485 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
487 I can reproduce this problem both with and without Tom's patch. Has
488 anyone seen this before? Maybe have a solution for it hanging around?
491 There's a test case for this documented at:
493 when debugging threaded applications you get extra SIGTRAPs
494 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
496 [There should be a GDB testcase - cagney]
500 GDB5 TOT on unixware 7
501 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
504 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
505 > practical matter, the current thread support is somewhat more annoying
506 > than when GDB was thread-unaware.
510 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
512 Add support for packet enable/disable commands with these thread
513 packets. General cleanup.
515 [PATCH] Document the ThreadInfo remote protocol queries
516 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
518 [PATCH] "info threads" queries for remote.c
519 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
526 New languages come onto the scene all the time.
530 Pascal (Pierre Muller, David Taylor)
532 Pierre Muller has contributed patches for adding Pascal Language
535 2 pascal language patches inserted in database
536 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
539 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
543 Java (Anthony Green, David Taylor)
545 Anthony Green has a number of Java patches that did not make it into
549 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
552 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
554 Patch: handle N_MAIN stab
555 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
561 Modify gdb to work correctly with Pascal.
565 Re: Various C++ things
567 value_headof/value_from_vtable_info are worthless, and should be
568 removed. The one place in printcmd.c that uses it should use the RTTI
571 RTTI for g++ should be using the typeinfo functions rather than the
572 vtables. The typeinfo functions are always at offset 4 from the
573 beginning of the vtable, and are always right. The vtables will have
574 weird names like E::VB sometimes. The typeinfo function will always
575 be "E type_info function", or somesuch.
577 value_virtual_fn_field needs to be fixed so there are no failures for
578 virtual functions for C++ using g++.
580 Testsuite cases are the major priority right now for C++ support,
581 since i have to make a lot of changes that could potentially break
586 Add support for Modula3
588 Get DEC/Compaq to contribute their Modula-3 support.
592 Remote Protocol Support
593 =======================
597 set/show remote X-packet ...
599 ``(gdb) help set remote X-packet'' doesn't list the applicable
600 responses. The help message needs to be expanded.
604 Remote protocol doco feedback.
606 Too much feedback to mention needs to be merged in (901660). Search
607 for the word ``remote''.
610 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
611 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
612 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
616 GDB doesn't recover gracefully from remote protocol errors.
618 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
619 ignored and a timeout is required before GDB retries. A pre-cursor to
620 fixing this this is making GDB's remote protocol packet more robust.
622 While downloading to a remote protocol target, gdb ignores packet
623 errors in so far as it will continue to edownload with chunk N+1 even
624 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
625 take a painfully long time to run. As a PS that test needs to be
626 fixed so that it builds on 16 bit machines.
630 Add the cycle step command.
632 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
636 Resolve how to scale things to support very large packets.
640 Resolve how to handle a target that changes things like its endianess
641 on the fly - should it be returned in the ``T'' packet?
643 Underlying problem is that the register file is target endian. If the
644 target endianess changes gdb doesn't know.
651 If / when GDB starts to support the debugging of multi-processor
652 (rather than multi-thread) applications the symtab code will need to
653 be updated a little so that several independant symbol tables are
654 active at a given time.
656 The other interesting change is a clarification of the exact meaning
657 of CORE_ADDR and that has had consequences for a few targets (that
658 were abusing that data type).
662 Investiagate ways of reducing memory.
666 Investigate ways of improving load time.
670 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
672 Consequence of recent symtab clarification. No marks for figuring out
673 who maintains the d10v.
677 Get the MIPS to correctly sign extend all address <-> pointer
680 Consequence of recent symtab clarification. No marks for figuring out
681 who maintains the MIPS.
688 There are never to many testcases.
692 Better thread testsuite.
696 Better C++ testsuite.
700 Look at adding a GDB specific testsuite directory so that white box
701 tests of key internals can be added (eg ui_file).
705 Separate out tests that involve the floating point (FP).
707 (Something for people brining up new targets). FP and non-fp tests
708 are combined. I think there should be set of basic tests that
709 exercise pure integer support and then a more expanded set that
710 exercise FP and FP/integer interactions.
712 As an example, the MIPS, for n32 as problems with passing FP's and
713 structs. Since most inferior call tests include FP it is difficult to
714 determine of the integer tests are ok.
718 Architectural Changes: General
719 ==============================
721 These are harder than simple cleanups / fixes and, consequently
722 involve more work. Typically an Architectural Change will be broken
723 down into a more digestible set of cleanups and fixes.
727 Cleanup software single step.
729 At present many targets implement software single step by directly
730 blatting memory (see rs6000-tdep.c). Those targets should register
731 the applicable breakpoints using the breakpoint framework. Perhaphs a
732 new internal breakpoint class ``step'' is needed.
736 Replace READ_FP() with FRAME_HANDLE().
738 READ_FP() is a hangover from the days of the vax when the ABI really
739 did have a frame pointer register. Modern architectures typically
740 construct a virtual frame-handle from the stack pointer and various
741 other bits of string.
743 Unfortunatly GDB still treats this synthetic FP register as though it
744 is real. That in turn really confuses users (arm and ``print $fp'' VS
745 ``info registers fp''). The synthetic FP should be separated out of
746 the true register set presented to the user.
750 Register Cache Cleanup (below from Andrew Cagney)
752 I would depict the current register architecture as something like:
759 register + REGISTER_BYTE(reg_nr)
762 -------------------------
763 | extern register[] |
764 -------------------------
766 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
767 really clear on what mechanisms they should be using to manipulate that
768 buffer. Further, much code assumes, dangerously, that registers are
769 contigious. Having got mips-tdep.c to support multiple ABIs, believe
770 me, that is a bad assumption. Finally, that register cache layout is
771 determined by the current remote/local target and _not_ the less
772 specific target ISA. In fact, in many cases it is determined by the
773 somewhat arbitrary layout of the [gG] packets!
776 How I would like the register file to work is more like:
804 The main objectives being:
806 o a clear separation between the low
807 level target and the high level GDB
809 o a mechanism that solves the general
810 problem of register aliases, overlaps
811 etc instead of treating them as optional
812 extras that can be wedged in as an after
813 thought (that is a reasonable description
814 of the current code).
816 Identify then solve the hard case and the
817 rest just falls out. GDB solved the easy
818 case and then tried to ignore the real
821 o a removal of the assumption that the
822 mapping between the register cache
823 and virtual registers is largely static.
824 If you flip the USR/SSR stack register
825 select bit in the status-register then
826 the corresponding stack registers should
829 o a mechanism that clearly separates the
830 gdb internal register cache from any
831 target (not architecture) dependant
832 specifics such as [gG] packets.
834 Of course, like anything, it sounds good in theory. In reality, it
835 would have to contend with many<->many relationships at both the
836 virt<->cache and cache<->target level. For instance:
839 Modifying an mmx register may involve
840 scattering values across both FP and
841 mmpx specific parts of a buffer
844 When writing back a SP it may need to
845 both be written to both SP and USP.
850 Rather than let this like the last time it was discussed, just slip, I'm
851 first going to add this e-mail (+ references) to TODO. I'd then like to
852 sketch out a broad strategy I think could get us there.
855 First thing I'd suggest is separating out the ``extern registers[]''
856 code so that we can at least identify what is using it. At present
857 things are scattered across many files. That way we can at least
858 pretend that there is a cache instead of a global array :-)
860 I'd then suggest someone putting up a proposal for the pseudo-reg /
861 high-level side interface so that code can be adopted to it. For old
862 code, initially a blanket rename of write_register_bytes() to
863 deprecated_write_register_bytes() would help.
865 Following that would, finaly be the corresponding changes to the target.
869 Check that GDB can handle all BFD architectures (Andrew Cagney)
871 There should be a test that checks that BFD/GDB are in sync with
872 regard to architecture changes. Something like a test that first
873 queries GDB for all supported architectures and then feeds each back
874 to GDB.. Anyone interested in learning how to write tests? :-)
878 Architectural Change: Multi-arch et al.
879 =======================================
881 The long term objective is to remove all assumptions that there is a
882 single target with a single address space with a single instruction
883 set architecture and single application binary interface.
885 This is an ongoing effort. The first milestone is to enable
886 ``multi-arch'' where by all architectural decisions are made at
889 It should be noted that ``gdbarch'' is really ``gdbabi'' and
890 ``gdbisa''. Once things are multi-arched breaking that down correctly
891 will become much easier.
895 GDBARCH cleanup (Andrew Cagney)
897 The non-generated parts of gdbarch.{sh,h,c} should be separated out
898 into arch-utils.[hc].
900 Document that gdbarch_init_ftype could easily fail because it didn't
901 identify an architecture.
905 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
907 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
912 Fix target_signal_from_host() etc.
914 The name is wrong for starters. ``target_signal'' should probably be
915 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
916 After that it needs to be multi-arched and made independant of any
917 host signal numbering.
921 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
924 This is a barrier to replacing mips_extra_func_info with something
925 that works with multi-arch.
929 Multi-arch mips_extra_func_info.
931 This first needs the alpha to be updated so that it uses ``struct
936 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
938 Surely one of them is redundant.
942 Convert ALL architectures to MULTI-ARCH.
946 Select the initial multi-arch ISA / ABI based on --target or similar.
948 At present the default is based on what ever is first in the BFD
949 archures table. It should be determined based on the ``--target=...''
956 Enable the code to recognize --enable-targets=.... like BINUTILS does.
958 Can the tm.h and nm.h files be eliminated by multi-arch.
962 Architectural Change: MI, LIBGDB and scripting languages
963 ========================================================
965 See also architectural changes related to the event loop. LIBGDB
966 can't be finished until there is a generic event loop being used by
969 The long term objective is it to be possible to integrate GDB into
974 Implement generic ``(gdb) commmand > file''
976 Once everything is going through ui_file it should be come fairly
979 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
983 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
985 gdb_stdtarg is easily confused with gdb_stdarg.
989 Extra ui_file methods - dump.
991 Very useful for whitebox testing.
995 Eliminate error_begin().
997 With ui_file, there is no need for the statefull error_begin ()
1002 Send normal output to gdb_stdout.
1003 Send error messages to gdb_stderror.
1004 Send debug and log output log gdb_stdlog.
1006 GDB still contains many cases where (f)printf or printf_filtered () is
1007 used when it should be sending the messages to gdb_stderror or
1008 gdb_stdlog. The thought of #defining printf to something has crossed
1013 Re-do GDB's output pager.
1015 GDB's output pager still relies on people correctly using *_filtered
1016 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1017 Hopefully, with all normal output going to gdb_stdout, the pager can
1018 just look at the ui_file that the output is on and then use that to
1019 decide what to do about paging. Sounds good in theory.
1023 Check/cleanup MI documentation.
1025 The list of commands specified in the documentation needs to be
1026 checked against the mi-cmds.c table in a mechanical way (so that they
1027 two can be kept up-to-date).
1031 Convert MI into libgdb
1033 MI provides a text interface into what should be many of the libgdb
1034 functions. The implementation of those functions should be separated
1035 into the MI interface and the functions proper. Those functions being
1036 moved to gdb/lib say.
1042 The first part can already be found in defs.h.
1046 MI's input does not use buffering.
1048 At present the MI interface reads raw characters of from an unbuffered
1049 FD. This is to avoid several nasty buffer/race conditions. That code
1050 should be changed so that it registers its self with the event loop
1051 (on the input FD) and then push commands up to MI as they arrive.
1053 The serial code already does this.
1057 Make MI interface accessable from existing CLI.
1061 Add a breakpoint-edit command to MI.
1063 It would be similar to MI's breakpoint create but would apply to an
1064 existing breakpoint. It saves the need to delete/create breakpoints
1065 when ever they are changed.
1069 Add directory path to MI breakpoint.
1071 That way the GUI's task of finding the file within which the
1072 breakpoint was set is simplified.
1076 Add a mechanism to reject certain expression classes to MI
1078 There are situtations where you don't want GDB's expression
1079 parser/evaluator to perform inferior function calls or variable
1080 assignments. A way of restricting the expression parser so that such
1081 operations are not accepted would be very helpful.
1085 Remove sideffects from libgdb breakpoint create function.
1087 The user can use the CLI to create a breakpoint with partial
1088 information - no file (gdb would use the file from the last
1091 The libgdb interface currently affects that environment which can lead
1092 to confusion when a user is setting breakpoints via both the MI and
1095 This is also a good example of how getting the CLI ``right'' will be
1100 Move gdb_lasterr to ui_out?
1102 The way GDB throws errors and records them needs a re-think. ui_out
1103 handles the correct output well. It doesn't resolve what to do with
1104 output / error-messages when things go wrong.
1108 Architectural Change: Async
1109 ===========================
1111 While GDB uses an event loop when prompting the user for input. That
1112 event loop is not exploited by targets when they allow the target
1113 program to continue. Typically targets still block in (target_wait())
1114 until the program again halts.
1116 The closest a target comes to supporting full asynchronous mode are
1117 the remote targets ``async'' and ``extended-async''.
1121 Asynchronous expression evaluator
1123 Inferior function calls hang GDB.
1127 Fix implementation of ``target xxx''.
1129 At present when the user specifies ``target xxxx'', the CLI maps that
1130 directly onto a target open method. It is then assumed that the
1131 target open method should do all sorts of complicated things as this
1132 is the only chance it has. Check how the various remote targets
1133 duplicate the target operations. Check also how the various targets
1134 behave differently for purely arbitrary reasons.
1136 What should happen is that ``target xxxx'' should call a generic
1137 ``target'' function and that should then co-ordinate the opening of
1138 ``xxxx''. This becomes especially important when you're trying to
1139 open an asynchronous target that may need to perform background tasks
1140 as part of the ``attach'' phase.
1142 Unfortunatly, due to limitations in the old/creaking command.h
1143 interface, that isn't possible. The function being called isn't told
1144 of the ``xxx'' or any other context information.
1146 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1147 CLI code so that it passes to the callback function (attatched to a
1148 command) useful information such as the actual command and a context
1149 for that command. Other changes such as making ``struct command''
1150 opaque may also help.
1154 Make "target xxx" command interruptible.
1156 As things become async this becomes possible. A target would start
1157 the connect and then return control to the event loop. A cntrl-c
1158 would notify the target that the operation is to be abandoned and the
1159 target code could respond.
1163 Add a "suspend" subcommand of the "continue" command to suspend gdb
1164 while continuing execution of the subprocess. Useful when you are
1165 debugging servers and you want to dodge out and initiate a connection
1166 to a server running under gdb.
1175 Frequently requested but not approved requests.
1179 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1181 The benefits on this one are thought to be marginal - GDBs design
1182 means that unused parameters are very common. GCC 3.0 will also
1183 include the option -Wno-unused-parameter which means that ``-Wall
1184 -Wno-unused-parameters -Werror'' can be specified.
1193 This list is not up to date, and opinions vary about the importance or
1194 even desirability of some of the items. If you do fix something, it
1195 always pays to check the below.
1199 @c This does not work (yet if ever). FIXME.
1200 @c @item --parse=@var{lang} @dots{}
1201 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1202 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1203 @c list of all supported languages, omit the argument. Without this
1204 @c option, @value{GDBN} is configured to parse all supported languages.
1208 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1209 is its default value. Clean this up.
1213 It should be possible to use symbols from shared libraries before we know
1214 exactly where the libraries will be loaded. E.g. "b perror" before running
1215 the program. This could maybe be done as an extension of the "breakpoint
1216 re-evaluation" after new symbols are loaded.
1220 Make single_step() insert and remove breakpoints in one operation.
1222 [If this is talking about having single_step() insert the breakpoints,
1223 run the target then pull the breakpoints then it is wrong. The
1224 function has to return as control has to eventually be passed back to
1225 the main event loop.]
1229 Speed up single stepping by avoiding extraneous ptrace calls.
1233 Speed up single stepping by not inserting and removing breakpoints
1234 each time the inferior starts and stops.
1236 Breakpoints should not be inserted and deleted all the time. Only the
1237 one(s) there should be removed when we have to step over one. Support
1238 breakpoints that don't have to be removed to step over them.
1240 [this has resulted in numerous debates. The issue isn't clear cut]
1244 Provide "voodoo" debugging of core files. This creates a zombie
1245 process as a child of the debugger, and loads it up with the data,
1246 stack, and regs of the core file. This allows you to call functions
1247 in the executable, to manipulate the data in the core file.
1253 GDB reopens the source file on every line, as you "next" through it.
1255 [still true? I've a memory of this being fixed]
1259 Perhaps "i source" should take an argument like that of "list".
1263 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1264 it matches the source line indicated.
1268 The prompt at end of screen should accept space as well as CR.
1272 Backtrace should point out what the currently selected frame is, in
1273 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1274 ...)" rather than "#3 foo (bar, ...)".
1278 "i program" should work for core files, and display more info, like what
1279 actually caused it to die.
1283 "x/10i" should shorten the long name, if any, on subsequent lines.
1287 "next" over a function that longjumps, never stops until next time you happen
1288 to get to that spot by accident. E.g. "n" over execute_command which has
1293 "set zeroprint off", don't bother printing members of structs which
1294 are entirely zero. Useful for those big structs with few useful
1299 GDB does four ioctl's for every command, probably switching terminal modes
1300 to/from inferior or for readline or something.
1304 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1305 if the state is the same, too.
1309 "i frame" shows wrong "arglist at" location, doesn't show where the args
1310 should be found, only their actual values.
1314 There should be a way for "set" commands to validate the new setting
1315 before it takes effect.
1319 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1323 i line VAR produces "Line number not known for symbol ``var''.". I
1324 thought we were stashing that info now!
1328 We should be able to write to random files at hex offsets like adb.
1332 [elena - delete this]
1334 Handle add_file with separate text, data, and bss addresses. Maybe
1335 handle separate addresses for each segment in the object file?
1339 [Jimb/Elena delete this one]
1341 Handle free_named_symtab to cope with multiply-loaded object files
1342 in a dynamic linking environment. Should remember the last copy loaded,
1343 but not get too snowed if it finds references to the older copy.
1347 [elena delete this also]
1349 Remove all references to:
1356 now that we have BFD. All remaining are in machine dependent files.
1360 Re-organize help categories into things that tend to fit on a screen
1365 Add in commands like ADB's for searching for patterns, etc. We should
1366 be able to examine and patch raw unsymboled binaries as well in gdb as
1367 we can in adb. (E.g. increase the timeout in /bin/login without source).
1369 [actually, add ADB interface :-]
1373 When doing "step" or "next", if a few lines of source are skipped between
1374 the previous line and the current one, print those lines, not just the
1375 last line of a multiline statement.
1379 Handling of "&" address-of operator needs some serious overhaul
1380 for ANSI C and consistency on arrays and functions.
1381 For "float point[15];":
1382 ptype &point[4] ==> Attempt to take address of non-lvalue.
1383 For "char *malloc();":
1384 ptype malloc ==> "char *()"; should be same as
1385 ptype &malloc ==> "char *(*)()"
1386 call printf ("%x\n", malloc) ==> weird value, should be same as
1387 call printf ("%x\n", &malloc) ==> correct value
1391 Fix dbxread.c symbol reading in the presence of interrupts. It
1392 currently leaves a cleanup to blow away the entire symbol table when a
1393 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1395 [I suspect that the grype was that, on a slow system, you might want
1396 to cntrl-c and get just half the symbols and then load the rest later
1397 - scary to be honest]
1401 Mipsread.c reads include files depth-first, because the dependencies
1402 in the psymtabs are way too inclusive (it seems to me). Figure out what
1403 really depends on what, to avoid recursing 20 or 30 times while reading
1408 value_add() should be subtracting the lower bound of arrays, if known,
1409 and possibly checking against the upper bound for error reporting.
1413 When listing source lines, check for a preceding \n, to verify that
1414 the file hasn't changed out from under us.
1416 [fixed by some other means I think. That hack wouldn't actually work
1417 reliably - the file might move such that another \n appears. ]
1421 Get all the remote systems (where the protocol allows it) to be able to
1422 stop the remote system when the GDB user types ^C (like remote.c
1423 does). For ebmon, use ^Ak.
1427 Possible feature: A version of the "disassemble" command which shows
1428 both source and assembly code ("set symbol-filename on" is a partial
1431 [has this been done? It was certainly done for MI and GDBtk]
1435 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1436 using a 0 address for bad purposes internally).
1440 Make "info path" and path_command work again (but independent of the
1441 environment either of gdb or that we'll pass to the inferior).
1445 Make GDB understand the GCC feature for putting octal constants in
1446 enums. Make it so overflow on an enum constant does not error_type
1447 the whole type. Allow arbitrarily large enums with type attributes.
1448 Put all this stuff in the testsuite.
1452 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1453 the value in hex; process type attributes). Add this to the
1454 testsuite. This way future compilers can add new types and old
1455 versions of GDB can do something halfway reasonable.
1459 Fix mdebugread.c:parse_type to do fundamental types right (see
1460 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1461 that the debug format fixes the sizes of these things and it shouldn't
1462 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1463 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1464 be aware of that). Also use a switch statement for clarity and speed.
1468 Investigate adding symbols in target_load--some targets do, some
1473 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1474 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1479 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1484 Make a watchpoint which contains a function call an error (it is
1485 broken now, making it work is probably not worth the effort).
1489 New test case based on weird.exp but in which type numbers are not
1490 renumbered (thus multiply defining a type). This currently causes an
1491 infinite loop on "p v_comb".
1495 [Hey! Hint Hint Delete Delete!!!]
1497 Fix 386 floating point so that floating point registers are real
1498 registers (but code can deal at run-time if they are missing, like
1499 mips and 68k). This would clean up "info float" and related stuff.
1503 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1504 about not being able to access memory location 0.
1506 -------------------- enummask.c
1527 If try to modify value in file with "set write off" should give
1528 appropriate error not "cannot access memory at address 0x65e0".
1532 Allow core file without exec file on RS/6000.
1536 Make sure "shell" with no arguments works right on DOS.
1540 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1541 the same directory can be NFS-mounted on unix or DOS, and work the
1546 [Is this another delete???]
1548 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
1549 get RS/6000 to work right, might not be immediately relevant).
1553 Work out some kind of way to allow running the inferior to be done as
1554 a sub-execution of, eg. breakpoint command lists. Currently running
1555 the inferior interupts any command list execution. This would require
1556 some rewriting of wait_for_inferior & friends, and hence should
1557 probably be done in concert with the above.
1561 Add function arguments to gdb user defined functions.
1565 Add convenience variables that refer to exec file, symbol file,
1566 selected frame source file, selected frame function, selected frame
1571 Modify the handling of symbols grouped through BINCL/EINCL stabs to
1572 allocate a partial symtab for each BINCL/EINCL grouping. This will
1573 seriously decrease the size of inter-psymtab dependencies and hence
1574 lessen the amount that needs to be read in when a new source file is
1579 Add a command for searching memory, a la adb. It specifies size,
1580 mask, value, start address. ADB searches until it finds it or hits
1581 an error (or is interrupted).
1585 Remove the range and type checking code and documentation, if not