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.
10 Below is a list of problems identified during the GDB 5.0 release
11 cycle. People hope to have these problems fixed in 5.1.
15 Update GDB's coding standard documentation. Known topics:
17 o alloca/malloc et.al.
29 Wow, three bug reports for the same problem in one day! We should
30 probably make fixing this a real priority :-).
32 Anyway, thanks for reporting.
34 The following patch will fix the problems with setting breakpoints in
35 dynamically loaded objects:
37 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00230.html
39 This patch isn't checked in yet (ping Michael/JimB), but I hope this
40 will be in the next GDB release.
42 There should really be a test in the testsuite for this problem, since
43 it keeps coming up :-(. Any volunteers?
49 x86 linux GDB and SIGALRM (???)
50 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00803.html
52 This problem has been fixed, but a regression test still needs to be
53 added to the testsuite:
54 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00309.html
58 [The test has been submitted for approval - cagney]
62 RFD: infrun.c: No bpstat_stop_status call after proceed over break?
63 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00665.html
65 GDB misses watchpoint triggers after proceeding over a breakpoint on
70 GDB 5.0 doesn't work on Linux/SPARC
72 There are two parts to this.
74 o GDB 5.0 doesn't work on GNU/Linux/SPARC32
76 o GDB 5.0 doesn't work on the new target
79 GDB does build on both these targets.
81 The first problem is the one that should be fixed.
85 GDB 5.1 - New features
86 ======================
88 The following new features should be included in 5.1.
92 Enable MI by default. Old code can be deleted after 5.1 is out.
96 o syntax change where a list would
98 [ foo=a, foo=b, foo=c ]
100 { foo=a, foo=b, foo=c }
102 o kill off the idea of a reverse
111 Pascal (Pierre Muller, David Taylor)
113 Pierre Muller has contributed patches for adding Pascal Language
116 2 pascal language patches inserted in database
117 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
120 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
122 [I think this has been merged, need to confirm - cagney]
126 Java (Anthony Green, David Taylor)
128 Anthony Green has a number of Java patches that did not make it into
129 the 5.0 release. The first two are in cvs now, but the third needs
130 some fixing up before it can go in.
133 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
136 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
138 Patch: handle N_MAIN stab
139 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
145 A predicate to this is the multi-arching of SOFTWARE_SINGLE_STEP(). A
146 patch has been submitted.
153 The following code cleanups will hopefully be applied to GDB 5.1.
157 Resolve the build status of all broken targets as identified by the
160 o arm-* vs NetBSD's lack of ``unix''
161 o arm-* vs IRIX (see below)
167 Obsolete some targets.
169 Possible selection criteria are:
171 o uses a deprecated feature
175 o doesn't have a maintainer
179 o post proposals to gdb@ (DONE)
181 o post announcement to gdb-announce@
182 crossed with gdb@ reply-to to gdb@
199 configure.{in,host,tgt}
201 *-tdep.c *-nat.c *-xdep.c
208 Change documentation to GFDL license.
210 ``It is time to make an effort to start using the GFDL more
211 thoroughly. Would all GNU maintainers please change the license to
212 the GFDL, for all manuals and other major documentation files?
214 The GFDL and some instructions for using it can be found in
215 http://www.gnu.org/copyleft/''
221 Fix copyright notices.
223 Turns out that ``1998-2000'' isn't considered valid :-(
225 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
229 GDB 5.1 - Known Problems
230 ========================
236 The z8k has suffered bit rot and is known to not build. The problem
237 was occuring in the opcodes directory.
241 Solaris 8 x86 CURSES_H problem
242 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
244 The original problem was worked around with:
246 2000-06-06 Michael Snyder <msnyder@cygnus.com>
248 * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
249 * configure: Regenerate.
251 When building both GDB and SID using the same source tree the problem
252 will still occure. sid/component/configure.in mis-configures
253 <curses.h> and leaves wrong information in the config cache.
262 Thread support. Right now, as soon as a thread finishes and exits,
263 you're hosed. This problem is reported once a week or so.
267 GDB 5.2 - New features
268 ======================
272 GCC 3.0 ABI support (but hopefully sooner...).
276 Objective C/C++ support (but hopefully sooner...).
283 The following cleanups have been identified as part of GDB 5.2.
287 Remove old code that does not use ui_out functions and all the related
288 "ifdef"s. This also allows the elimination of -DUI_OUT from
289 Makefile.in and configure.in.
295 Eliminate all warnings for at least one host/target for the flags:
296 -Wimplicit -Wreturn-type -Wcomment -Wtrigraphs -Wformat -Wparentheses
297 -Wpointer-arith -Wuninitialized
301 Restructure gdb directory tree so that it avoids any 8.3 and 14
306 Convert GDB build process to AUTOMAKE.
308 See also sub-directory configure below.
310 The current convention is (kind of) to use $(<header>_h) in all
311 dependency lists. It isn't done in a consistent way.
315 GDB 5.2 - Known Problems
316 ========================
320 Code Cleanups: General
321 ======================
323 The following are more general cleanups and fixes. They are not tied
324 to any specific release.
328 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
332 Can't build IRIX -> arm GDB.
333 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
336 > Now I'm building for an embedded arm target. If there is a way of turning
337 > remote-rdi off, I couldn't find it. It looks like it gets built by default
338 > in gdb/configure.tgt(line 58) Anyway, the build dies in
339 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
340 > aren't one of the architectures supported.
344 Problem with weak functions
345 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
347 Dan Nicolaescu writes:
348 > It seems that gdb-4.95.1 does not display correctly the function when
349 > stoping in weak functions.
351 > It stops in a function that is defined as weak, not in the function
352 > that is actually run...
356 Follow through `make check' with --enable-shared.
358 When the srcware tree is configured with --enable-shared, the `expect'
359 program won't run properly. Jim Wilson found out gdb has a local hack
360 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
363 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
367 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
369 Patches in the database.
373 printcmd.c (print_address_numeric):
375 NOTE: This assumes that the significant address information is kept in
376 the least significant bits of ADDR - the upper bits were either zero
377 or sign extended. Should ADDRESS_TO_POINTER() or some
378 ADDRESS_TO_PRINTABLE() be used to do the conversion?
382 The BFD directory requires bug-fixed AUTOMAKE et.al.
384 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
385 contained the full path to texinfo.tex when it should have only
386 contained the directory. The bug has been fixed in the current
387 AUTOMAKE sources. Automake snapshots can be found in:
388 ftp://sourceware.cygnus.com/pub/gdb/snapshots
389 and ftp://sourceware.cygnus.com/pub/binutils
393 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
394 determine the default isa/byte-order.
398 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
399 BIG_ENDIAN and LITTLE_ENDIAN.
403 Eliminate more compiler warnings.
405 Of course there also needs to be the usual debate over which warnings
406 are valid and how to best go about this.
408 One method: choose a single option; get agreement that it is
409 reasonable; try it out to see if there isn't anything silly about it
410 (-Wunused-parameters is an example of that) then incrementally hack
413 The other method is to enable all warnings and eliminate them from one
418 Elimination of ``(catch_errors_ftype *) func''.
420 Like make_cleanup_func it isn't portable.
421 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
422 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
426 Nuke #define CONST_PTR.
434 [PATCH/5] src/intl/Makefile.in:distclean additions
435 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
437 Do not forget to merge the patch back into the trunk.
441 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
443 At present defs.h includes <endian.h> (which is linux specific) yet
444 almost nothing depends on it. Suggest "gdb_endian.h" which can also
445 handle <machine/endian.h> and only include that where it is really
450 Replace savestring() with something from libiberty.
452 An xstrldup()? but that would have different semantics.
456 Rationalize use of floatformat_unknown in GDB sources.
458 Instead of defaulting to floatformat_unknown, should hosts/targets
459 specify the value explicitly?
461 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
465 Add a ``name'' member to include/floatformat.h:struct floatformat.
466 Print that name in gdbarch.c.
470 Sort out the harris mess in include/floatformat.h (it hardwires two
471 different floating point formats).
475 See of the GDB local floatformat_do_doublest() and libiberty's
476 floatformat_to_double (which was once GDB's ...) can be merged some
481 Eliminate mmalloc(), mstrsave() et.al. from GDB.
483 Also eliminate it from defs.h.
487 Eliminate PTR. ISO-C allows ``void *''.
493 GDB should never abort. GDB should either throw ``error ()'' or
494 ``internal_error ()''. Better still GDB should naturally unwind with
499 GDB probably doesn't build on FreeBSD pre 2.2.x
500 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
502 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
507 Deprecate "fg". Apparently ``fg'' is actually continue.
509 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
513 Deprecate current use of ``floatformat_unknown''.
515 Require all targets to explicitly provide their float format instead
516 of defaulting to floatformat unknown. Doing the latter leads to nasty
519 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
523 Rationalize floatformat_to_double() vs floatformat_to_doublest().
525 Looks like GDB migrated floatformat_to_double() to libiberty but then
526 turned around and created a ..._to_doublest() the latter containing
529 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
533 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
535 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
539 The ``maintenance deprecate set endian big'' command doesn't notice
540 that it is deprecating ``set endian'' and not ``set endian big'' (big
541 is implemented using an enum). Is anyone going to notice this?
545 When tab expanding something like ``set arch<tab>'' ignore the
546 deprecated ``set archdebug'' and expand to ``set architecture''.
550 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
551 like from arm-tdep.c.
555 Fix uses of ->function.cfunc = set_function().
557 The command.c code calls sfunc() when a set command. Rather than
558 change it suggest fixing the callback function so that it is more
561 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
563 See also ``Fix implementation of ``target xxx''.'' below.
567 IRIX 3.x support is probably broken.
571 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
572 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
574 Apart from the d30v, are there any sim/common simulators that make use
577 A brief summary of what happened is that sim/common/sim-break.c was
578 created as a good idea. It turned out a better idea was to use
579 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
583 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
584 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
588 Eliminate ``extern'' from C files.
592 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
594 Extreme care is recommeded - perhaps only modify tests that are
595 exercised by the testsuite (as determined using some type of code
600 Replace the file gdb/CONTRIBUTE with a file that is generated from the
601 gdb/doc/*.texinfo directory.
605 New Features and Fixes
606 ======================
608 These are harder than cleanups but easier than work involving
609 fundamental architectural change.
613 Hardware watchpoint problems on x86 OSes, including Linux:
615 1. Delete/disable hardware watchpoints should free hardware debug
617 2. Watch for different values on a viariable with one hardware debug
620 According to Eli Zaretskii <eliz@delorie.com>:
622 These are not GDB/ia32 issues per se: the above features are all
623 implemented in the DJGPP port of GDB and work in v5.0. Every
624 x86-based target should be able to lift the relevant parts of
625 go32-nat.c and use them almost verbatim. You get debug register
626 sharing through reference counts, and the ability to watch large
627 regions (up to 16 bytes) using multiple registers. (The required
628 infrastructure in high-level GDB application code, mostly in
629 breakpoint.c, is also working since v5.0.)
633 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
634 so that you can see how the GDB was created.
638 Add an "info bfd" command that displays supported object formats,
639 similarly to objdump -i.
641 Is there a command already?
645 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
647 This requires internationalization.
653 (gdb) p fwprintf(stdout,L"%S\n", f)
654 No symbol "L" in current context.
658 Cleanup configury support for optional sub-directories.
660 Check how GCC handles multiple front ends for an example of how things
661 could work. A tentative first step is to rationalize things so that
662 all sub directories are handled in a fashion similar to gdb/mi.
664 See also automake above.
668 Add a transcript mechanism to GDB.
670 Such a mechanism might log all gdb input and output to a file in a
671 form that would allow it to be replayed. It could involve ``gdb
672 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
676 Can the xdep files be replaced by autoconf?
680 Document trace machinery
684 Document ui-out and ui-file.
686 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
690 Update texinfo.tex to latest?
694 Incorporate agentexpr.texi into gdb.texinfo
696 agentexpr.texi mostly describes the details of the byte code used for
697 tracepoints, not the internals of the support for this in GDB. So it
698 looks like gdb.texinfo is a better place for this information.
700 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
704 Document overlay machinery.
708 ``(gdb) catch signal SIGNAL''
710 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
711 You can attach commands to a catch but not a handle. A handle has a
712 limited number of hardwired actions.
718 o readline/*.h bitrot
720 The TUI isn't up-to-date with
721 respect to the readline currently
722 bundled with GDB. Importing a
723 new readline is on the 5.1 wish
724 list so this can only get worse.
726 Grep for things like term_cursor_move.
728 (To be honest, I don't see anyone
729 importing a new readline before 5.1 is
732 o tui.c:va_catch_errors() bitrot
734 This nasty piece of work used knowledge
735 of the internals of GDBs error functions :-(
736 Ever since those internals were cleaned
737 up this code has been broken. :-(
739 o tuiWin.c:c_makeVisibleWithNewHeight() broken
740 tuiLayout.c:_extractDisplayStartAddr() broken
742 Both these function call find_line_pc()
743 incorrectly (wrong args, wrong return value).
745 I suspect this bug has always been there!
746 It had been hidden because those files
747 didn't include the necessary header files
750 o tuiRegs() host dependant
752 Not suprisingly, this isn't a very portable
753 section of code. However, I'm sure people
754 could live with no regs in the short to
757 o defs.h: #include "tui.h" et.al.
759 I'm not sure where this came from.
760 It was a really bad idea.
762 To get things to compile I did a nasty
763 hack (Just declare what was needed and
764 replace any expressions like xx->y.z()
765 in GDB proper with function calls). I
766 could commit it slightly cleaned up if
769 Medium Term. the #ifdef TUI and TuiDo()
770 should be changed to hooks (like GDBTK).
771 The gdb-events.[hc] is there for that
774 o tui.c:_tuiReset() host dependant
776 tui.c contains a lump of termio[s]
777 I suspect an equivalent block of
778 code can be lifted from readline.
779 An equivalent readline function may
782 o curses.h vs ncurses.h.
784 Simple portability problem.
788 This function is a mystery - where is it?
790 o tui-file.[hc] cleanup
792 This can be significantly simplified.
794 o The code should be pacified. (-Werror -W...)
796 There are plenty of #includes,
797 duplicate #includes, missing function decls
800 Some of the problems I found were through
801 fixing a few of the warnings.
803 o The code should be GNUtified.
805 It would be very nice to have this code
806 look like the rest of GDB. That way people
807 would be more accepting of it as a true
810 Until it is GNUtified it is going to stick
811 out like a sore thumb to the programmer.
813 o The code should be clearly copyrighted
815 (FSF, with due credit to HP)
819 Add support for ``gdb --- PROGRAM ARGS ...''.
820 Add support for ``gdb -cmd=...''
822 Along with many variations. Check:
824 ????? for a full discussion.
830 Implement ``(gdb) !ls''.
832 Which is very different from ``(gdb) ! ls''. Implementing the latter
835 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
839 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
844 When tab expanding something like ``set arch<tab>'' ignore the
845 deprecated ``set archdebug'' and expand to ``set architecture''.
849 Replace the code that uses the host FPU with an emulator of the target
854 The "ocd reset" command needs to flush the dcache, which requires breaking
855 the abstraction layer between the target independent and target code. One
856 way to address this is provide a generic "reset" command and target vector.
858 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
867 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
868 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
870 The thread_db assisted debugging code doesn't handle exiting threads
871 properly, at least in combination with glibc 2.1.3 (the framework is
872 there, just not the actual code). There are at least two problems
873 that prevent this from working.
875 As an additional reference point, the pre thread_db code did not work
880 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
882 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
884 Christopher Blizzard writes:
886 So, I've done some more digging into this and it looks like Jim
887 Kingdon has reported this problem in the past:
889 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
891 I can reproduce this problem both with and without Tom's patch. Has
892 anyone seen this before? Maybe have a solution for it hanging around?
895 There's a test case for this documented at:
897 when debugging threaded applications you get extra SIGTRAPs
898 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
900 [There should be a GDB testcase - cagney]
904 GDB5 TOT on unixware 7
905 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
908 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
909 > practical matter, the current thread support is somewhat more annoying
910 > than when GDB was thread-unaware.
917 New languages come onto the scene all the time.
921 Re: Various C++ things
923 value_headof/value_from_vtable_info are worthless, and should be
924 removed. The one place in printcmd.c that uses it should use the RTTI
927 RTTI for g++ should be using the typeinfo functions rather than the
928 vtables. The typeinfo functions are always at offset 4 from the
929 beginning of the vtable, and are always right. The vtables will have
930 weird names like E::VB sometimes. The typeinfo function will always
931 be "E type_info function", or somesuch.
933 value_virtual_fn_field needs to be fixed so there are no failures for
934 virtual functions for C++ using g++.
936 Testsuite cases are the major priority right now for C++ support,
937 since i have to make a lot of changes that could potentially break
942 Add support for Modula3
944 Get DEC/Compaq to contribute their Modula-3 support.
948 Remote Protocol Support
949 =======================
953 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
954 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
956 (Broken) support for GDB's remote protocol across UDP is to be
957 included in the follow-on release.
959 It should be noted that UDP can only work when the [Gg] packet fits in
962 There is also much debate over the merit of this.
966 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
968 Add support for packet enable/disable commands with these thread
969 packets. General cleanup.
971 [PATCH] Document the ThreadInfo remote protocol queries
972 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
974 [PATCH] "info threads" queries for remote.c
975 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
979 Remote protocol doco feedback.
981 Too much feedback to mention needs to be merged in (901660). Search
982 for the word ``remote''.
985 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
986 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
987 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
991 GDB doesn't recover gracefully from remote protocol errors.
993 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
994 ignored and a timeout is required before GDB retries. A pre-cursor to
995 fixing this this is making GDB's remote protocol packet more robust.
997 While downloading to a remote protocol target, gdb ignores packet
998 errors in so far as it will continue to download with chunk N+1 even
999 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
1000 take a painfully long time to run. As a PS that test needs to be
1001 fixed so that it builds on 16 bit machines.
1005 Fix the ``!'' packet.
1007 JT reported that the existing targets do, in fact return ``OK'' so it
1008 is possible to merge remote and extended-remote targets.
1012 Drop ``<address>'' from the [SsCc] packets.
1014 I don't think that GDB generates them so having it in the protocol is
1019 Fix doco on the ``q'' packet.
1021 It has evolved into a generic RPC. The notes should reflect this and,
1022 perhaps, the ``Q'' packet can be deprecated.
1024 The doco should mention that ``OK'' is a valid packet response.
1026 The doco should explain why ``OK'' needs to be a valid packet
1031 Add the cycle step command.
1033 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1037 Resolve how to scale things to support very large packets.
1041 Resolve how to handle a target that changes things like its endianess
1042 on the fly - should it be returned in the ``T'' packet?
1044 Underlying problem is that the register file is target endian. If the
1045 target endianess changes gdb doesn't know.
1052 If / when GDB starts to support the debugging of multi-processor
1053 (rather than multi-thread) applications the symtab code will need to
1054 be updated a little so that several independent symbol tables are
1055 active at a given time.
1057 The other interesting change is a clarification of the exact meaning
1058 of CORE_ADDR and that has had consequences for a few targets (that
1059 were abusing that data type).
1063 Investiagate ways of reducing memory.
1067 Investigate ways of improving load time.
1071 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1073 Consequence of recent symtab clarification. No marks for figuring out
1074 who maintains the d10v.
1078 Get the MIPS to correctly sign extend all address <-> pointer
1081 Consequence of recent symtab clarification. No marks for figuring out
1082 who maintains the MIPS.
1086 GDB truncates 64 bit enums.
1088 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1095 There are never to many testcases.
1099 Better thread testsuite.
1103 Better C++ testsuite.
1107 Look at adding a GDB specific testsuite directory so that white box
1108 tests of key internals can be added (eg ui_file).
1112 Separate out tests that involve the floating point (FP).
1114 (Something for people brining up new targets). FP and non-fp tests
1115 are combined. I think there should be set of basic tests that
1116 exercise pure integer support and then a more expanded set that
1117 exercise FP and FP/integer interactions.
1119 As an example, the MIPS, for n32 as problems with passing FP's and
1120 structs. Since most inferior call tests include FP it is difficult to
1121 determine of the integer tests are ok.
1125 Architectural Changes: General
1126 ==============================
1128 These are harder than simple cleanups / fixes and, consequently
1129 involve more work. Typically an Architectural Change will be broken
1130 down into a more digestible set of cleanups and fixes.
1134 Cleanup software single step.
1136 At present many targets implement software single step by directly
1137 blatting memory (see rs6000-tdep.c). Those targets should register
1138 the applicable breakpoints using the breakpoint framework. Perhaphs a
1139 new internal breakpoint class ``step'' is needed.
1143 Replace READ_FP() with FRAME_HANDLE().
1145 READ_FP() is a hangover from the days of the vax when the ABI really
1146 did have a frame pointer register. Modern architectures typically
1147 construct a virtual frame-handle from the stack pointer and various
1148 other bits of string.
1150 Unfortunately GDB still treats this synthetic FP register as though it
1151 is real. That in turn really confuses users (arm and ``print $fp'' VS
1152 ``info registers fp''). The synthetic FP should be separated out of
1153 the true register set presented to the user.
1157 Register Cache Cleanup (below from Andrew Cagney)
1159 I would depict the current register architecture as something like:
1161 High GDB --> Low GDB
1166 register + REGISTER_BYTE(reg_nr)
1169 -------------------------
1170 | extern register[] |
1171 -------------------------
1173 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1174 really clear on what mechanisms they should be using to manipulate that
1175 buffer. Further, much code assumes, dangerously, that registers are
1176 contigious. Having got mips-tdep.c to support multiple ABIs, believe
1177 me, that is a bad assumption. Finally, that register cache layout is
1178 determined by the current remote/local target and _not_ the less
1179 specific target ISA. In fact, in many cases it is determined by the
1180 somewhat arbitrary layout of the [gG] packets!
1183 How I would like the register file to work is more like:
1211 The main objectives being:
1213 o a clear separation between the low
1214 level target and the high level GDB
1216 o a mechanism that solves the general
1217 problem of register aliases, overlaps
1218 etc instead of treating them as optional
1219 extras that can be wedged in as an after
1220 thought (that is a reasonable description
1221 of the current code).
1223 Identify then solve the hard case and the
1224 rest just falls out. GDB solved the easy
1225 case and then tried to ignore the real
1228 o a removal of the assumption that the
1229 mapping between the register cache
1230 and virtual registers is largely static.
1231 If you flip the USR/SSR stack register
1232 select bit in the status-register then
1233 the corresponding stack registers should
1236 o a mechanism that clearly separates the
1237 gdb internal register cache from any
1238 target (not architecture) dependent
1239 specifics such as [gG] packets.
1241 Of course, like anything, it sounds good in theory. In reality, it
1242 would have to contend with many<->many relationships at both the
1243 virt<->cache and cache<->target level. For instance:
1246 Modifying an mmx register may involve
1247 scattering values across both FP and
1248 mmpx specific parts of a buffer
1251 When writing back a SP it may need to
1252 both be written to both SP and USP.
1257 Rather than let this like the last time it was discussed, just slip, I'm
1258 first going to add this e-mail (+ references) to TODO. I'd then like to
1259 sketch out a broad strategy I think could get us there.
1262 First thing I'd suggest is separating out the ``extern registers[]''
1263 code so that we can at least identify what is using it. At present
1264 things are scattered across many files. That way we can at least
1265 pretend that there is a cache instead of a global array :-)
1267 I'd then suggest someone putting up a proposal for the pseudo-reg /
1268 high-level side interface so that code can be adopted to it. For old
1269 code, initially a blanket rename of write_register_bytes() to
1270 deprecated_write_register_bytes() would help.
1272 Following that would, finaly be the corresponding changes to the target.
1276 Check that GDB can handle all BFD architectures (Andrew Cagney)
1278 There should be a test that checks that BFD/GDB are in sync with
1279 regard to architecture changes. Something like a test that first
1280 queries GDB for all supported architectures and then feeds each back
1281 to GDB.. Anyone interested in learning how to write tests? :-)
1285 Architectural Change: Multi-arch et al.
1286 =======================================
1288 The long term objective is to remove all assumptions that there is a
1289 single target with a single address space with a single instruction
1290 set architecture and single application binary interface.
1292 This is an ongoing effort. The first milestone is to enable
1293 ``multi-arch'' where by all architectural decisions are made at
1296 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1297 ``gdbisa''. Once things are multi-arched breaking that down correctly
1298 will become much easier.
1302 GDBARCH cleanup (Andrew Cagney)
1304 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1305 into arch-utils.[hc].
1307 Document that gdbarch_init_ftype could easily fail because it didn't
1308 identify an architecture.
1312 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
1314 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1319 Fix target_signal_from_host() etc.
1321 The name is wrong for starters. ``target_signal'' should probably be
1322 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
1323 After that it needs to be multi-arched and made independent of any
1324 host signal numbering.
1328 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1331 This is a barrier to replacing mips_extra_func_info with something
1332 that works with multi-arch.
1336 Multi-arch mips_extra_func_info.
1338 This first needs the alpha to be updated so that it uses ``struct
1343 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1345 Surely one of them is redundant.
1349 Convert ALL architectures to MULTI-ARCH.
1353 Select the initial multi-arch ISA / ABI based on --target or similar.
1355 At present the default is based on what ever is first in the BFD
1356 archures table. It should be determined based on the ``--target=...''
1361 Make MIPS pure multi-arch.
1363 It is only at the multi-arch enabled stage.
1369 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1371 Can the tm.h and nm.h files be eliminated by multi-arch.
1375 Architectural Change: MI, LIBGDB and scripting languages
1376 ========================================================
1378 See also architectural changes related to the event loop. LIBGDB
1379 can't be finished until there is a generic event loop being used by
1382 The long term objective is it to be possible to integrate GDB into
1383 scripting languages.
1387 Implement generic ``(gdb) commmand > file''
1389 Once everything is going through ui_file it should be come fairly
1392 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1396 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1398 gdb_stdtarg is easily confused with gdb_stdarg.
1402 Extra ui_file methods - dump.
1404 Very useful for whitebox testing.
1408 Eliminate error_begin().
1410 With ui_file, there is no need for the statefull error_begin ()
1415 Send normal output to gdb_stdout.
1416 Send error messages to gdb_stderror.
1417 Send debug and log output log gdb_stdlog.
1419 GDB still contains many cases where (f)printf or printf_filtered () is
1420 used when it should be sending the messages to gdb_stderror or
1421 gdb_stdlog. The thought of #defining printf to something has crossed
1426 Re-do GDB's output pager.
1428 GDB's output pager still relies on people correctly using *_filtered
1429 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1430 Hopefully, with all normal output going to gdb_stdout, the pager can
1431 just look at the ui_file that the output is on and then use that to
1432 decide what to do about paging. Sounds good in theory.
1436 Check/cleanup MI documentation.
1438 The list of commands specified in the documentation needs to be
1439 checked against the mi-cmds.c table in a mechanical way (so that they
1440 two can be kept up-to-date).
1444 Convert MI into libgdb
1446 MI provides a text interface into what should be many of the libgdb
1447 functions. The implementation of those functions should be separated
1448 into the MI interface and the functions proper. Those functions being
1449 moved to gdb/lib say.
1455 The first part can already be found in defs.h.
1459 MI's input does not use buffering.
1461 At present the MI interface reads raw characters of from an unbuffered
1462 FD. This is to avoid several nasty buffer/race conditions. That code
1463 should be changed so that it registers its self with the event loop
1464 (on the input FD) and then push commands up to MI as they arrive.
1466 The serial code already does this.
1470 Make MI interface accessible from existing CLI.
1474 Add a breakpoint-edit command to MI.
1476 It would be similar to MI's breakpoint create but would apply to an
1477 existing breakpoint. It saves the need to delete/create breakpoints
1478 when ever they are changed.
1482 Add directory path to MI breakpoint.
1484 That way the GUI's task of finding the file within which the
1485 breakpoint was set is simplified.
1489 Add a mechanism to reject certain expression classes to MI
1491 There are situtations where you don't want GDB's expression
1492 parser/evaluator to perform inferior function calls or variable
1493 assignments. A way of restricting the expression parser so that such
1494 operations are not accepted would be very helpful.
1498 Remove sideffects from libgdb breakpoint create function.
1500 The user can use the CLI to create a breakpoint with partial
1501 information - no file (gdb would use the file from the last
1504 The libgdb interface currently affects that environment which can lead
1505 to confusion when a user is setting breakpoints via both the MI and
1508 This is also a good example of how getting the CLI ``right'' will be
1513 Move gdb_lasterr to ui_out?
1515 The way GDB throws errors and records them needs a re-think. ui_out
1516 handles the correct output well. It doesn't resolve what to do with
1517 output / error-messages when things go wrong.
1521 do_setshow_command contains a 1024 byte buffer.
1523 The function assumes that there will never be any more than 1024 bytes
1524 of enum. It should use mem_file.
1528 Should struct cmd_list_element . completer take the command as an
1533 Should the bulk of top.c:line_completion_function() be moved to
1534 command.[hc]? complete_on_cmdlist() and complete_on_enums() could
1535 then be made private.
1539 top.c (execute_command): Should a command being valid when the target
1540 is running be made an attribute (predicate) to the command rather than
1541 an explicit set of tests.
1545 top.c (execute_command): Should the bulk of this function be moved
1546 into command.[hc] so that top.c doesn't grub around in the command
1551 Architectural Change: Async
1552 ===========================
1554 While GDB uses an event loop when prompting the user for input. That
1555 event loop is not exploited by targets when they allow the target
1556 program to continue. Typically targets still block in (target_wait())
1557 until the program again halts.
1559 The closest a target comes to supporting full asynchronous mode are
1560 the remote targets ``async'' and ``extended-async''.
1564 Asynchronous expression evaluator
1566 Inferior function calls hang GDB.
1570 Fix implementation of ``target xxx''.
1572 At present when the user specifies ``target xxxx'', the CLI maps that
1573 directly onto a target open method. It is then assumed that the
1574 target open method should do all sorts of complicated things as this
1575 is the only chance it has. Check how the various remote targets
1576 duplicate the target operations. Check also how the various targets
1577 behave differently for purely arbitrary reasons.
1579 What should happen is that ``target xxxx'' should call a generic
1580 ``target'' function and that should then co-ordinate the opening of
1581 ``xxxx''. This becomes especially important when you're trying to
1582 open an asynchronous target that may need to perform background tasks
1583 as part of the ``attach'' phase.
1585 Unfortunately, due to limitations in the old/creaking command.h
1586 interface, that isn't possible. The function being called isn't told
1587 of the ``xxx'' or any other context information.
1589 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1590 CLI code so that it passes to the callback function (attatched to a
1591 command) useful information such as the actual command and a context
1592 for that command. Other changes such as making ``struct command''
1593 opaque may also help.
1596 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1600 Make "target xxx" command interruptible.
1602 As things become async this becomes possible. A target would start
1603 the connect and then return control to the event loop. A cntrl-c
1604 would notify the target that the operation is to be abandoned and the
1605 target code could respond.
1609 Add a "suspend" subcommand of the "continue" command to suspend gdb
1610 while continuing execution of the subprocess. Useful when you are
1611 debugging servers and you want to dodge out and initiate a connection
1612 to a server running under gdb.
1621 Frequently requested but not approved requests.
1625 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1627 The benefits on this one are thought to be marginal - GDBs design
1628 means that unused parameters are very common. GCC 3.0 will also
1629 include the option -Wno-unused-parameter which means that ``-Wall
1630 -Wno-unused-parameters -Werror'' can be specified.
1639 This list is not up to date, and opinions vary about the importance or
1640 even desirability of some of the items. If you do fix something, it
1641 always pays to check the below.
1645 @c This does not work (yet if ever). FIXME.
1646 @c @item --parse=@var{lang} @dots{}
1647 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1648 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1649 @c list of all supported languages, omit the argument. Without this
1650 @c option, @value{GDBN} is configured to parse all supported languages.
1654 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1655 is its default value. Clean this up.
1659 It should be possible to use symbols from shared libraries before we know
1660 exactly where the libraries will be loaded. E.g. "b perror" before running
1661 the program. This could maybe be done as an extension of the "breakpoint
1662 re-evaluation" after new symbols are loaded.
1666 Make single_step() insert and remove breakpoints in one operation.
1668 [If this is talking about having single_step() insert the breakpoints,
1669 run the target then pull the breakpoints then it is wrong. The
1670 function has to return as control has to eventually be passed back to
1671 the main event loop.]
1675 Speed up single stepping by avoiding extraneous ptrace calls.
1679 Speed up single stepping by not inserting and removing breakpoints
1680 each time the inferior starts and stops.
1682 Breakpoints should not be inserted and deleted all the time. Only the
1683 one(s) there should be removed when we have to step over one. Support
1684 breakpoints that don't have to be removed to step over them.
1686 [this has resulted in numerous debates. The issue isn't clear cut]
1690 Provide "voodoo" debugging of core files. This creates a zombie
1691 process as a child of the debugger, and loads it up with the data,
1692 stack, and regs of the core file. This allows you to call functions
1693 in the executable, to manipulate the data in the core file.
1699 GDB reopens the source file on every line, as you "next" through it.
1701 [still true? I've a memory of this being fixed]
1705 Perhaps "i source" should take an argument like that of "list".
1709 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1710 it matches the source line indicated.
1714 The prompt at end of screen should accept space as well as CR.
1718 Backtrace should point out what the currently selected frame is, in
1719 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1720 ...)" rather than "#3 foo (bar, ...)".
1724 "i program" should work for core files, and display more info, like what
1725 actually caused it to die.
1729 "x/10i" should shorten the long name, if any, on subsequent lines.
1733 "next" over a function that longjumps, never stops until next time you happen
1734 to get to that spot by accident. E.g. "n" over execute_command which has
1739 "set zeroprint off", don't bother printing members of structs which
1740 are entirely zero. Useful for those big structs with few useful
1745 GDB does four ioctl's for every command, probably switching terminal modes
1746 to/from inferior or for readline or something.
1750 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1751 if the state is the same, too.
1755 "i frame" shows wrong "arglist at" location, doesn't show where the args
1756 should be found, only their actual values.
1760 There should be a way for "set" commands to validate the new setting
1761 before it takes effect.
1765 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1769 i line VAR produces "Line number not known for symbol ``var''.". I
1770 thought we were stashing that info now!
1774 We should be able to write to random files at hex offsets like adb.
1778 [elena - delete this]
1780 Handle add_file with separate text, data, and bss addresses. Maybe
1781 handle separate addresses for each segment in the object file?
1785 [Jimb/Elena delete this one]
1787 Handle free_named_symtab to cope with multiply-loaded object files
1788 in a dynamic linking environment. Should remember the last copy loaded,
1789 but not get too snowed if it finds references to the older copy.
1793 [elena delete this also]
1795 Remove all references to:
1802 now that we have BFD. All remaining are in machine dependent files.
1806 Re-organize help categories into things that tend to fit on a screen
1811 Add in commands like ADB's for searching for patterns, etc. We should
1812 be able to examine and patch raw unsymboled binaries as well in gdb as
1813 we can in adb. (E.g. increase the timeout in /bin/login without source).
1815 [actually, add ADB interface :-]
1819 When doing "step" or "next", if a few lines of source are skipped between
1820 the previous line and the current one, print those lines, not just the
1821 last line of a multiline statement.
1825 Handling of "&" address-of operator needs some serious overhaul
1826 for ANSI C and consistency on arrays and functions.
1827 For "float point[15];":
1828 ptype &point[4] ==> Attempt to take address of non-lvalue.
1829 For "char *malloc();":
1830 ptype malloc ==> "char *()"; should be same as
1831 ptype &malloc ==> "char *(*)()"
1832 call printf ("%x\n", malloc) ==> weird value, should be same as
1833 call printf ("%x\n", &malloc) ==> correct value
1837 Fix dbxread.c symbol reading in the presence of interrupts. It
1838 currently leaves a cleanup to blow away the entire symbol table when a
1839 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1841 [I suspect that the grype was that, on a slow system, you might want
1842 to cntrl-c and get just half the symbols and then load the rest later
1843 - scary to be honest]
1847 Mipsread.c reads include files depth-first, because the dependencies
1848 in the psymtabs are way too inclusive (it seems to me). Figure out what
1849 really depends on what, to avoid recursing 20 or 30 times while reading
1854 value_add() should be subtracting the lower bound of arrays, if known,
1855 and possibly checking against the upper bound for error reporting.
1859 When listing source lines, check for a preceding \n, to verify that
1860 the file hasn't changed out from under us.
1862 [fixed by some other means I think. That hack wouldn't actually work
1863 reliably - the file might move such that another \n appears. ]
1867 Get all the remote systems (where the protocol allows it) to be able to
1868 stop the remote system when the GDB user types ^C (like remote.c
1869 does). For ebmon, use ^Ak.
1873 Possible feature: A version of the "disassemble" command which shows
1874 both source and assembly code ("set symbol-filename on" is a partial
1877 [has this been done? It was certainly done for MI and GDBtk]
1881 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1882 using a 0 address for bad purposes internally).
1886 Make "info path" and path_command work again (but independent of the
1887 environment either of gdb or that we'll pass to the inferior).
1891 Make GDB understand the GCC feature for putting octal constants in
1892 enums. Make it so overflow on an enum constant does not error_type
1893 the whole type. Allow arbitrarily large enums with type attributes.
1894 Put all this stuff in the testsuite.
1898 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1899 the value in hex; process type attributes). Add this to the
1900 testsuite. This way future compilers can add new types and old
1901 versions of GDB can do something halfway reasonable.
1905 Fix mdebugread.c:parse_type to do fundamental types right (see
1906 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1907 that the debug format fixes the sizes of these things and it shouldn't
1908 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1909 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1910 be aware of that). Also use a switch statement for clarity and speed.
1914 Investigate adding symbols in target_load--some targets do, some
1919 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1920 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1925 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1930 Make a watchpoint which contains a function call an error (it is
1931 broken now, making it work is probably not worth the effort).
1935 New test case based on weird.exp but in which type numbers are not
1936 renumbered (thus multiply defining a type). This currently causes an
1937 infinite loop on "p v_comb".
1941 [Hey! Hint Hint Delete Delete!!!]
1943 Fix 386 floating point so that floating point registers are real
1944 registers (but code can deal at run-time if they are missing, like
1945 mips and 68k). This would clean up "info float" and related stuff.
1949 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1950 about not being able to access memory location 0.
1952 -------------------- enummask.c
1973 If try to modify value in file with "set write off" should give
1974 appropriate error not "cannot access memory at address 0x65e0".
1978 Allow core file without exec file on RS/6000.
1982 Make sure "shell" with no arguments works right on DOS.
1986 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1987 the same directory can be NFS-mounted on unix or DOS, and work the
1992 [Is this another delete???]
1994 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
1995 get RS/6000 to work right, might not be immediately relevant).
1999 Work out some kind of way to allow running the inferior to be done as
2000 a sub-execution of, eg. breakpoint command lists. Currently running
2001 the inferior interupts any command list execution. This would require
2002 some rewriting of wait_for_inferior & friends, and hence should
2003 probably be done in concert with the above.
2007 Add function arguments to gdb user defined functions.
2011 Add convenience variables that refer to exec file, symbol file,
2012 selected frame source file, selected frame function, selected frame
2017 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2018 allocate a partial symtab for each BINCL/EINCL grouping. This will
2019 seriously decrease the size of inter-psymtab dependencies and hence
2020 lessen the amount that needs to be read in when a new source file is
2025 Add a command for searching memory, a la adb. It specifies size,
2026 mask, value, start address. ADB searches until it finds it or hits
2027 an error (or is interrupted).
2031 Remove the range and type checking code and documentation, if not