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)
169 Obsolete some targets.
171 Possible selection criteria are:
173 o uses a deprecated feature
177 o doesn't have a maintainer
181 o post proposals to gdb@ (DONE)
183 o post announcement to gdb-announce@
184 crossed with gdb@ reply-to to gdb@
206 configure.{in,host,tgt}
208 *-tdep.c *-nat.c *-xdep.c
215 Change documentation to GFDL license.
217 ``It is time to make an effort to start using the GFDL more
218 thoroughly. Would all GNU maintainers please change the license to
219 the GFDL, for all manuals and other major documentation files?
221 The GFDL and some instructions for using it can be found in
222 http://www.gnu.org/copyleft/''
228 Fix copyright notices.
230 Turns out that ``1998-2000'' isn't considered valid :-(
232 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
236 GDB 5.1 - Known Problems
237 ========================
243 The z8k has suffered bit rot and is known to not build. The problem
244 was occuring in the opcodes directory.
248 Solaris 8 x86 CURSES_H problem
249 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
251 The original problem was worked around with:
253 2000-06-06 Michael Snyder <msnyder@cygnus.com>
255 * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
256 * configure: Regenerate.
258 When building both GDB and SID using the same source tree the problem
259 will still occure. sid/component/configure.in mis-configures
260 <curses.h> and leaves wrong information in the config cache.
269 Thread support. Right now, as soon as a thread finishes and exits,
270 you're hosed. This problem is reported once a week or so.
274 GDB 5.2 - New features
275 ======================
279 GCC 3.0 ABI support (but hopefully sooner...).
283 Objective C/C++ support (but hopefully sooner...).
290 The following cleanups have been identified as part of GDB 5.2.
294 Remove old code that does not use ui_out functions and all the related
295 "ifdef"s. This also allows the elimination of -DUI_OUT from
296 Makefile.in and configure.in.
302 Eliminate all warnings for at least one host/target for the flags:
303 -Wimplicit -Wreturn-type -Wcomment -Wtrigraphs -Wformat -Wparentheses
304 -Wpointer-arith -Wuninitialized
308 Restructure gdb directory tree so that it avoids any 8.3 and 14
313 Convert GDB build process to AUTOMAKE.
315 See also sub-directory configure below.
317 The current convention is (kind of) to use $(<header>_h) in all
318 dependency lists. It isn't done in a consistent way.
322 GDB 5.2 - Known Problems
323 ========================
327 Code Cleanups: General
328 ======================
330 The following are more general cleanups and fixes. They are not tied
331 to any specific release.
335 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
339 Can't build IRIX -> arm GDB.
340 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
343 > Now I'm building for an embedded arm target. If there is a way of turning
344 > remote-rdi off, I couldn't find it. It looks like it gets built by default
345 > in gdb/configure.tgt(line 58) Anyway, the build dies in
346 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
347 > aren't one of the architectures supported.
351 Problem with weak functions
352 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
354 Dan Nicolaescu writes:
355 > It seems that gdb-4.95.1 does not display correctly the function when
356 > stoping in weak functions.
358 > It stops in a function that is defined as weak, not in the function
359 > that is actually run...
363 Follow through `make check' with --enable-shared.
365 When the srcware tree is configured with --enable-shared, the `expect'
366 program won't run properly. Jim Wilson found out gdb has a local hack
367 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
370 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
374 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
376 Patches in the database.
380 printcmd.c (print_address_numeric):
382 NOTE: This assumes that the significant address information is kept in
383 the least significant bits of ADDR - the upper bits were either zero
384 or sign extended. Should ADDRESS_TO_POINTER() or some
385 ADDRESS_TO_PRINTABLE() be used to do the conversion?
389 The BFD directory requires bug-fixed AUTOMAKE et.al.
391 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
392 contained the full path to texinfo.tex when it should have only
393 contained the directory. The bug has been fixed in the current
394 AUTOMAKE sources. Automake snapshots can be found in:
395 ftp://sourceware.cygnus.com/pub/gdb/snapshots
396 and ftp://sourceware.cygnus.com/pub/binutils
400 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
401 determine the default isa/byte-order.
405 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
406 BIG_ENDIAN and LITTLE_ENDIAN.
410 Eliminate more compiler warnings.
412 Of course there also needs to be the usual debate over which warnings
413 are valid and how to best go about this.
415 One method: choose a single option; get agreement that it is
416 reasonable; try it out to see if there isn't anything silly about it
417 (-Wunused-parameters is an example of that) then incrementally hack
420 The other method is to enable all warnings and eliminate them from one
425 Elimination of ``(catch_errors_ftype *) func''.
427 Like make_cleanup_func it isn't portable.
428 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
429 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
433 Nuke #define CONST_PTR.
441 [PATCH/5] src/intl/Makefile.in:distclean additions
442 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
444 Do not forget to merge the patch back into the trunk.
448 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
450 At present defs.h includes <endian.h> (which is linux specific) yet
451 almost nothing depends on it. Suggest "gdb_endian.h" which can also
452 handle <machine/endian.h> and only include that where it is really
457 Replace savestring() with something from libiberty.
459 An xstrldup()? but that would have different semantics.
463 Rationalize use of floatformat_unknown in GDB sources.
465 Instead of defaulting to floatformat_unknown, should hosts/targets
466 specify the value explicitly?
468 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
472 Add a ``name'' member to include/floatformat.h:struct floatformat.
473 Print that name in gdbarch.c.
477 Sort out the harris mess in include/floatformat.h (it hardwires two
478 different floating point formats).
482 See of the GDB local floatformat_do_doublest() and libiberty's
483 floatformat_to_double (which was once GDB's ...) can be merged some
488 Eliminate mmalloc(), mstrsave() et.al. from GDB.
490 Also eliminate it from defs.h.
494 Eliminate PTR. ISO-C allows ``void *''.
500 GDB should never abort. GDB should either throw ``error ()'' or
501 ``internal_error ()''. Better still GDB should naturally unwind with
506 GDB probably doesn't build on FreeBSD pre 2.2.x
507 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
509 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
514 Deprecate "fg". Apparently ``fg'' is actually continue.
516 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
520 Deprecate current use of ``floatformat_unknown''.
522 Require all targets to explicitly provide their float format instead
523 of defaulting to floatformat unknown. Doing the latter leads to nasty
526 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
530 Rationalize floatformat_to_double() vs floatformat_to_doublest().
532 Looks like GDB migrated floatformat_to_double() to libiberty but then
533 turned around and created a ..._to_doublest() the latter containing
536 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
540 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
542 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
546 The ``maintenance deprecate set endian big'' command doesn't notice
547 that it is deprecating ``set endian'' and not ``set endian big'' (big
548 is implemented using an enum). Is anyone going to notice this?
552 When tab expanding something like ``set arch<tab>'' ignore the
553 deprecated ``set archdebug'' and expand to ``set architecture''.
557 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
558 like from arm-tdep.c.
562 Fix uses of ->function.cfunc = set_function().
564 The command.c code calls sfunc() when a set command. Rather than
565 change it suggest fixing the callback function so that it is more
568 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
570 See also ``Fix implementation of ``target xxx''.'' below.
574 IRIX 3.x support is probably broken.
578 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
579 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
581 Apart from the d30v, are there any sim/common simulators that make use
584 A brief summary of what happened is that sim/common/sim-break.c was
585 created as a good idea. It turned out a better idea was to use
586 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
590 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
591 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
595 Eliminate ``extern'' from C files.
599 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
601 Extreme care is recommeded - perhaps only modify tests that are
602 exercised by the testsuite (as determined using some type of code
607 Replace the file gdb/CONTRIBUTE with a file that is generated from the
608 gdb/doc/*.texinfo directory.
612 New Features and Fixes
613 ======================
615 These are harder than cleanups but easier than work involving
616 fundamental architectural change.
620 Hardware watchpoint problems on x86 OSes, including Linux:
622 1. Delete/disable hardware watchpoints should free hardware debug
624 2. Watch for different values on a viariable with one hardware debug
627 According to Eli Zaretskii <eliz@delorie.com>:
629 These are not GDB/ia32 issues per se: the above features are all
630 implemented in the DJGPP port of GDB and work in v5.0. Every
631 x86-based target should be able to lift the relevant parts of
632 go32-nat.c and use them almost verbatim. You get debug register
633 sharing through reference counts, and the ability to watch large
634 regions (up to 16 bytes) using multiple registers. (The required
635 infrastructure in high-level GDB application code, mostly in
636 breakpoint.c, is also working since v5.0.)
640 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
641 so that you can see how the GDB was created.
645 Add an "info bfd" command that displays supported object formats,
646 similarly to objdump -i.
648 Is there a command already?
652 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
654 This requires internationalization.
660 (gdb) p fwprintf(stdout,L"%S\n", f)
661 No symbol "L" in current context.
665 Cleanup configury support for optional sub-directories.
667 Check how GCC handles multiple front ends for an example of how things
668 could work. A tentative first step is to rationalize things so that
669 all sub directories are handled in a fashion similar to gdb/mi.
671 See also automake above.
675 Add a transcript mechanism to GDB.
677 Such a mechanism might log all gdb input and output to a file in a
678 form that would allow it to be replayed. It could involve ``gdb
679 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
683 Can the xdep files be replaced by autoconf?
687 Document trace machinery
691 Document ui-out and ui-file.
693 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
697 Update texinfo.tex to latest?
701 Incorporate agentexpr.texi into gdb.texinfo
703 agentexpr.texi mostly describes the details of the byte code used for
704 tracepoints, not the internals of the support for this in GDB. So it
705 looks like gdb.texinfo is a better place for this information.
707 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
711 Document overlay machinery.
715 ``(gdb) catch signal SIGNAL''
717 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
718 You can attach commands to a catch but not a handle. A handle has a
719 limited number of hardwired actions.
725 o readline/*.h bitrot
727 The TUI isn't up-to-date with
728 respect to the readline currently
729 bundled with GDB. Importing a
730 new readline is on the 5.1 wish
731 list so this can only get worse.
733 Grep for things like term_cursor_move.
735 (To be honest, I don't see anyone
736 importing a new readline before 5.1 is
739 o tui.c:va_catch_errors() bitrot
741 This nasty piece of work used knowledge
742 of the internals of GDBs error functions :-(
743 Ever since those internals were cleaned
744 up this code has been broken. :-(
746 o tuiWin.c:c_makeVisibleWithNewHeight() broken
747 tuiLayout.c:_extractDisplayStartAddr() broken
749 Both these function call find_line_pc()
750 incorrectly (wrong args, wrong return value).
752 I suspect this bug has always been there!
753 It had been hidden because those files
754 didn't include the necessary header files
757 o tuiRegs() host dependant
759 Not suprisingly, this isn't a very portable
760 section of code. However, I'm sure people
761 could live with no regs in the short to
764 o defs.h: #include "tui.h" et.al.
766 I'm not sure where this came from.
767 It was a really bad idea.
769 To get things to compile I did a nasty
770 hack (Just declare what was needed and
771 replace any expressions like xx->y.z()
772 in GDB proper with function calls). I
773 could commit it slightly cleaned up if
776 Medium Term. the #ifdef TUI and TuiDo()
777 should be changed to hooks (like GDBTK).
778 The gdb-events.[hc] is there for that
781 o tui.c:_tuiReset() host dependant
783 tui.c contains a lump of termio[s]
784 I suspect an equivalent block of
785 code can be lifted from readline.
786 An equivalent readline function may
789 o curses.h vs ncurses.h.
791 Simple portability problem.
795 This function is a mystery - where is it?
797 o tui-file.[hc] cleanup
799 This can be significantly simplified.
801 o The code should be pacified. (-Werror -W...)
803 There are plenty of #includes,
804 duplicate #includes, missing function decls
807 Some of the problems I found were through
808 fixing a few of the warnings.
810 o The code should be GNUtified.
812 It would be very nice to have this code
813 look like the rest of GDB. That way people
814 would be more accepting of it as a true
817 Until it is GNUtified it is going to stick
818 out like a sore thumb to the programmer.
820 o The code should be clearly copyrighted
822 (FSF, with due credit to HP)
826 Add support for ``gdb --- PROGRAM ARGS ...''.
827 Add support for ``gdb -cmd=...''
829 Along with many variations. Check:
831 ????? for a full discussion.
837 Implement ``(gdb) !ls''.
839 Which is very different from ``(gdb) ! ls''. Implementing the latter
842 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
846 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
851 When tab expanding something like ``set arch<tab>'' ignore the
852 deprecated ``set archdebug'' and expand to ``set architecture''.
856 Replace the code that uses the host FPU with an emulator of the target
861 The "ocd reset" command needs to flush the dcache, which requires breaking
862 the abstraction layer between the target independent and target code. One
863 way to address this is provide a generic "reset" command and target vector.
865 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
874 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
875 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
877 The thread_db assisted debugging code doesn't handle exiting threads
878 properly, at least in combination with glibc 2.1.3 (the framework is
879 there, just not the actual code). There are at least two problems
880 that prevent this from working.
882 As an additional reference point, the pre thread_db code did not work
887 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
889 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
891 Christopher Blizzard writes:
893 So, I've done some more digging into this and it looks like Jim
894 Kingdon has reported this problem in the past:
896 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
898 I can reproduce this problem both with and without Tom's patch. Has
899 anyone seen this before? Maybe have a solution for it hanging around?
902 There's a test case for this documented at:
904 when debugging threaded applications you get extra SIGTRAPs
905 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
907 [There should be a GDB testcase - cagney]
911 GDB5 TOT on unixware 7
912 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
915 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
916 > practical matter, the current thread support is somewhat more annoying
917 > than when GDB was thread-unaware.
924 New languages come onto the scene all the time.
928 Re: Various C++ things
930 value_headof/value_from_vtable_info are worthless, and should be
931 removed. The one place in printcmd.c that uses it should use the RTTI
934 RTTI for g++ should be using the typeinfo functions rather than the
935 vtables. The typeinfo functions are always at offset 4 from the
936 beginning of the vtable, and are always right. The vtables will have
937 weird names like E::VB sometimes. The typeinfo function will always
938 be "E type_info function", or somesuch.
940 value_virtual_fn_field needs to be fixed so there are no failures for
941 virtual functions for C++ using g++.
943 Testsuite cases are the major priority right now for C++ support,
944 since i have to make a lot of changes that could potentially break
949 Add support for Modula3
951 Get DEC/Compaq to contribute their Modula-3 support.
955 Remote Protocol Support
956 =======================
960 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
961 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
963 (Broken) support for GDB's remote protocol across UDP is to be
964 included in the follow-on release.
966 It should be noted that UDP can only work when the [Gg] packet fits in
969 There is also much debate over the merit of this.
973 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
975 Add support for packet enable/disable commands with these thread
976 packets. General cleanup.
978 [PATCH] Document the ThreadInfo remote protocol queries
979 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
981 [PATCH] "info threads" queries for remote.c
982 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
986 Remote protocol doco feedback.
988 Too much feedback to mention needs to be merged in (901660). Search
989 for the word ``remote''.
992 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
993 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
994 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
998 GDB doesn't recover gracefully from remote protocol errors.
1000 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
1001 ignored and a timeout is required before GDB retries. A pre-cursor to
1002 fixing this this is making GDB's remote protocol packet more robust.
1004 While downloading to a remote protocol target, gdb ignores packet
1005 errors in so far as it will continue to download with chunk N+1 even
1006 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
1007 take a painfully long time to run. As a PS that test needs to be
1008 fixed so that it builds on 16 bit machines.
1012 Fix the ``!'' packet.
1014 JT reported that the existing targets do, in fact return ``OK'' so it
1015 is possible to merge remote and extended-remote targets.
1019 Drop ``<address>'' from the [SsCc] packets.
1021 I don't think that GDB generates them so having it in the protocol is
1026 Fix doco on the ``q'' packet.
1028 It has evolved into a generic RPC. The notes should reflect this and,
1029 perhaps, the ``Q'' packet can be deprecated.
1031 The doco should mention that ``OK'' is a valid packet response.
1033 The doco should explain why ``OK'' needs to be a valid packet
1038 Add the cycle step command.
1040 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1044 Resolve how to scale things to support very large packets.
1048 Resolve how to handle a target that changes things like its endianess
1049 on the fly - should it be returned in the ``T'' packet?
1051 Underlying problem is that the register file is target endian. If the
1052 target endianess changes gdb doesn't know.
1059 If / when GDB starts to support the debugging of multi-processor
1060 (rather than multi-thread) applications the symtab code will need to
1061 be updated a little so that several independent symbol tables are
1062 active at a given time.
1064 The other interesting change is a clarification of the exact meaning
1065 of CORE_ADDR and that has had consequences for a few targets (that
1066 were abusing that data type).
1070 Investiagate ways of reducing memory.
1074 Investigate ways of improving load time.
1078 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1080 Consequence of recent symtab clarification. No marks for figuring out
1081 who maintains the d10v.
1085 Get the MIPS to correctly sign extend all address <-> pointer
1088 Consequence of recent symtab clarification. No marks for figuring out
1089 who maintains the MIPS.
1093 GDB truncates 64 bit enums.
1095 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1102 There are never to many testcases.
1106 Better thread testsuite.
1110 Better C++ testsuite.
1114 Look at adding a GDB specific testsuite directory so that white box
1115 tests of key internals can be added (eg ui_file).
1119 Separate out tests that involve the floating point (FP).
1121 (Something for people brining up new targets). FP and non-fp tests
1122 are combined. I think there should be set of basic tests that
1123 exercise pure integer support and then a more expanded set that
1124 exercise FP and FP/integer interactions.
1126 As an example, the MIPS, for n32 as problems with passing FP's and
1127 structs. Since most inferior call tests include FP it is difficult to
1128 determine of the integer tests are ok.
1132 Architectural Changes: General
1133 ==============================
1135 These are harder than simple cleanups / fixes and, consequently
1136 involve more work. Typically an Architectural Change will be broken
1137 down into a more digestible set of cleanups and fixes.
1141 Cleanup software single step.
1143 At present many targets implement software single step by directly
1144 blatting memory (see rs6000-tdep.c). Those targets should register
1145 the applicable breakpoints using the breakpoint framework. Perhaphs a
1146 new internal breakpoint class ``step'' is needed.
1150 Replace READ_FP() with FRAME_HANDLE().
1152 READ_FP() is a hangover from the days of the vax when the ABI really
1153 did have a frame pointer register. Modern architectures typically
1154 construct a virtual frame-handle from the stack pointer and various
1155 other bits of string.
1157 Unfortunately GDB still treats this synthetic FP register as though it
1158 is real. That in turn really confuses users (arm and ``print $fp'' VS
1159 ``info registers fp''). The synthetic FP should be separated out of
1160 the true register set presented to the user.
1164 Register Cache Cleanup (below from Andrew Cagney)
1166 I would depict the current register architecture as something like:
1168 High GDB --> Low GDB
1173 register + REGISTER_BYTE(reg_nr)
1176 -------------------------
1177 | extern register[] |
1178 -------------------------
1180 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1181 really clear on what mechanisms they should be using to manipulate that
1182 buffer. Further, much code assumes, dangerously, that registers are
1183 contigious. Having got mips-tdep.c to support multiple ABIs, believe
1184 me, that is a bad assumption. Finally, that register cache layout is
1185 determined by the current remote/local target and _not_ the less
1186 specific target ISA. In fact, in many cases it is determined by the
1187 somewhat arbitrary layout of the [gG] packets!
1190 How I would like the register file to work is more like:
1218 The main objectives being:
1220 o a clear separation between the low
1221 level target and the high level GDB
1223 o a mechanism that solves the general
1224 problem of register aliases, overlaps
1225 etc instead of treating them as optional
1226 extras that can be wedged in as an after
1227 thought (that is a reasonable description
1228 of the current code).
1230 Identify then solve the hard case and the
1231 rest just falls out. GDB solved the easy
1232 case and then tried to ignore the real
1235 o a removal of the assumption that the
1236 mapping between the register cache
1237 and virtual registers is largely static.
1238 If you flip the USR/SSR stack register
1239 select bit in the status-register then
1240 the corresponding stack registers should
1243 o a mechanism that clearly separates the
1244 gdb internal register cache from any
1245 target (not architecture) dependent
1246 specifics such as [gG] packets.
1248 Of course, like anything, it sounds good in theory. In reality, it
1249 would have to contend with many<->many relationships at both the
1250 virt<->cache and cache<->target level. For instance:
1253 Modifying an mmx register may involve
1254 scattering values across both FP and
1255 mmpx specific parts of a buffer
1258 When writing back a SP it may need to
1259 both be written to both SP and USP.
1264 Rather than let this like the last time it was discussed, just slip, I'm
1265 first going to add this e-mail (+ references) to TODO. I'd then like to
1266 sketch out a broad strategy I think could get us there.
1269 First thing I'd suggest is separating out the ``extern registers[]''
1270 code so that we can at least identify what is using it. At present
1271 things are scattered across many files. That way we can at least
1272 pretend that there is a cache instead of a global array :-)
1274 I'd then suggest someone putting up a proposal for the pseudo-reg /
1275 high-level side interface so that code can be adopted to it. For old
1276 code, initially a blanket rename of write_register_bytes() to
1277 deprecated_write_register_bytes() would help.
1279 Following that would, finaly be the corresponding changes to the target.
1283 Check that GDB can handle all BFD architectures (Andrew Cagney)
1285 There should be a test that checks that BFD/GDB are in sync with
1286 regard to architecture changes. Something like a test that first
1287 queries GDB for all supported architectures and then feeds each back
1288 to GDB.. Anyone interested in learning how to write tests? :-)
1292 Architectural Change: Multi-arch et al.
1293 =======================================
1295 The long term objective is to remove all assumptions that there is a
1296 single target with a single address space with a single instruction
1297 set architecture and single application binary interface.
1299 This is an ongoing effort. The first milestone is to enable
1300 ``multi-arch'' where by all architectural decisions are made at
1303 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1304 ``gdbisa''. Once things are multi-arched breaking that down correctly
1305 will become much easier.
1309 GDBARCH cleanup (Andrew Cagney)
1311 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1312 into arch-utils.[hc].
1314 Document that gdbarch_init_ftype could easily fail because it didn't
1315 identify an architecture.
1319 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
1321 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1326 Fix target_signal_from_host() etc.
1328 The name is wrong for starters. ``target_signal'' should probably be
1329 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
1330 After that it needs to be multi-arched and made independent of any
1331 host signal numbering.
1335 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1338 This is a barrier to replacing mips_extra_func_info with something
1339 that works with multi-arch.
1343 Multi-arch mips_extra_func_info.
1345 This first needs the alpha to be updated so that it uses ``struct
1350 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1352 Surely one of them is redundant.
1356 Convert ALL architectures to MULTI-ARCH.
1360 Select the initial multi-arch ISA / ABI based on --target or similar.
1362 At present the default is based on what ever is first in the BFD
1363 archures table. It should be determined based on the ``--target=...''
1368 Make MIPS pure multi-arch.
1370 It is only at the multi-arch enabled stage.
1376 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1378 Can the tm.h and nm.h files be eliminated by multi-arch.
1382 Architectural Change: MI, LIBGDB and scripting languages
1383 ========================================================
1385 See also architectural changes related to the event loop. LIBGDB
1386 can't be finished until there is a generic event loop being used by
1389 The long term objective is it to be possible to integrate GDB into
1390 scripting languages.
1394 Implement generic ``(gdb) commmand > file''
1396 Once everything is going through ui_file it should be come fairly
1399 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1403 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1405 gdb_stdtarg is easily confused with gdb_stdarg.
1409 Extra ui_file methods - dump.
1411 Very useful for whitebox testing.
1415 Eliminate error_begin().
1417 With ui_file, there is no need for the statefull error_begin ()
1422 Send normal output to gdb_stdout.
1423 Send error messages to gdb_stderror.
1424 Send debug and log output log gdb_stdlog.
1426 GDB still contains many cases where (f)printf or printf_filtered () is
1427 used when it should be sending the messages to gdb_stderror or
1428 gdb_stdlog. The thought of #defining printf to something has crossed
1433 Re-do GDB's output pager.
1435 GDB's output pager still relies on people correctly using *_filtered
1436 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1437 Hopefully, with all normal output going to gdb_stdout, the pager can
1438 just look at the ui_file that the output is on and then use that to
1439 decide what to do about paging. Sounds good in theory.
1443 Check/cleanup MI documentation.
1445 The list of commands specified in the documentation needs to be
1446 checked against the mi-cmds.c table in a mechanical way (so that they
1447 two can be kept up-to-date).
1451 Convert MI into libgdb
1453 MI provides a text interface into what should be many of the libgdb
1454 functions. The implementation of those functions should be separated
1455 into the MI interface and the functions proper. Those functions being
1456 moved to gdb/lib say.
1462 The first part can already be found in defs.h.
1466 MI's input does not use buffering.
1468 At present the MI interface reads raw characters of from an unbuffered
1469 FD. This is to avoid several nasty buffer/race conditions. That code
1470 should be changed so that it registers its self with the event loop
1471 (on the input FD) and then push commands up to MI as they arrive.
1473 The serial code already does this.
1477 Make MI interface accessible from existing CLI.
1481 Add a breakpoint-edit command to MI.
1483 It would be similar to MI's breakpoint create but would apply to an
1484 existing breakpoint. It saves the need to delete/create breakpoints
1485 when ever they are changed.
1489 Add directory path to MI breakpoint.
1491 That way the GUI's task of finding the file within which the
1492 breakpoint was set is simplified.
1496 Add a mechanism to reject certain expression classes to MI
1498 There are situtations where you don't want GDB's expression
1499 parser/evaluator to perform inferior function calls or variable
1500 assignments. A way of restricting the expression parser so that such
1501 operations are not accepted would be very helpful.
1505 Remove sideffects from libgdb breakpoint create function.
1507 The user can use the CLI to create a breakpoint with partial
1508 information - no file (gdb would use the file from the last
1511 The libgdb interface currently affects that environment which can lead
1512 to confusion when a user is setting breakpoints via both the MI and
1515 This is also a good example of how getting the CLI ``right'' will be
1520 Move gdb_lasterr to ui_out?
1522 The way GDB throws errors and records them needs a re-think. ui_out
1523 handles the correct output well. It doesn't resolve what to do with
1524 output / error-messages when things go wrong.
1528 do_setshow_command contains a 1024 byte buffer.
1530 The function assumes that there will never be any more than 1024 bytes
1531 of enum. It should use mem_file.
1535 Should struct cmd_list_element . completer take the command as an
1540 Should the bulk of top.c:line_completion_function() be moved to
1541 command.[hc]? complete_on_cmdlist() and complete_on_enums() could
1542 then be made private.
1546 top.c (execute_command): Should a command being valid when the target
1547 is running be made an attribute (predicate) to the command rather than
1548 an explicit set of tests.
1552 top.c (execute_command): Should the bulk of this function be moved
1553 into command.[hc] so that top.c doesn't grub around in the command
1558 Architectural Change: Async
1559 ===========================
1561 While GDB uses an event loop when prompting the user for input. That
1562 event loop is not exploited by targets when they allow the target
1563 program to continue. Typically targets still block in (target_wait())
1564 until the program again halts.
1566 The closest a target comes to supporting full asynchronous mode are
1567 the remote targets ``async'' and ``extended-async''.
1571 Asynchronous expression evaluator
1573 Inferior function calls hang GDB.
1577 Fix implementation of ``target xxx''.
1579 At present when the user specifies ``target xxxx'', the CLI maps that
1580 directly onto a target open method. It is then assumed that the
1581 target open method should do all sorts of complicated things as this
1582 is the only chance it has. Check how the various remote targets
1583 duplicate the target operations. Check also how the various targets
1584 behave differently for purely arbitrary reasons.
1586 What should happen is that ``target xxxx'' should call a generic
1587 ``target'' function and that should then co-ordinate the opening of
1588 ``xxxx''. This becomes especially important when you're trying to
1589 open an asynchronous target that may need to perform background tasks
1590 as part of the ``attach'' phase.
1592 Unfortunately, due to limitations in the old/creaking command.h
1593 interface, that isn't possible. The function being called isn't told
1594 of the ``xxx'' or any other context information.
1596 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1597 CLI code so that it passes to the callback function (attatched to a
1598 command) useful information such as the actual command and a context
1599 for that command. Other changes such as making ``struct command''
1600 opaque may also help.
1603 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1607 Make "target xxx" command interruptible.
1609 As things become async this becomes possible. A target would start
1610 the connect and then return control to the event loop. A cntrl-c
1611 would notify the target that the operation is to be abandoned and the
1612 target code could respond.
1616 Add a "suspend" subcommand of the "continue" command to suspend gdb
1617 while continuing execution of the subprocess. Useful when you are
1618 debugging servers and you want to dodge out and initiate a connection
1619 to a server running under gdb.
1628 Frequently requested but not approved requests.
1632 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1634 The benefits on this one are thought to be marginal - GDBs design
1635 means that unused parameters are very common. GCC 3.0 will also
1636 include the option -Wno-unused-parameter which means that ``-Wall
1637 -Wno-unused-parameters -Werror'' can be specified.
1646 This list is not up to date, and opinions vary about the importance or
1647 even desirability of some of the items. If you do fix something, it
1648 always pays to check the below.
1652 @c This does not work (yet if ever). FIXME.
1653 @c @item --parse=@var{lang} @dots{}
1654 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1655 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1656 @c list of all supported languages, omit the argument. Without this
1657 @c option, @value{GDBN} is configured to parse all supported languages.
1661 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1662 is its default value. Clean this up.
1666 It should be possible to use symbols from shared libraries before we know
1667 exactly where the libraries will be loaded. E.g. "b perror" before running
1668 the program. This could maybe be done as an extension of the "breakpoint
1669 re-evaluation" after new symbols are loaded.
1673 Make single_step() insert and remove breakpoints in one operation.
1675 [If this is talking about having single_step() insert the breakpoints,
1676 run the target then pull the breakpoints then it is wrong. The
1677 function has to return as control has to eventually be passed back to
1678 the main event loop.]
1682 Speed up single stepping by avoiding extraneous ptrace calls.
1686 Speed up single stepping by not inserting and removing breakpoints
1687 each time the inferior starts and stops.
1689 Breakpoints should not be inserted and deleted all the time. Only the
1690 one(s) there should be removed when we have to step over one. Support
1691 breakpoints that don't have to be removed to step over them.
1693 [this has resulted in numerous debates. The issue isn't clear cut]
1697 Provide "voodoo" debugging of core files. This creates a zombie
1698 process as a child of the debugger, and loads it up with the data,
1699 stack, and regs of the core file. This allows you to call functions
1700 in the executable, to manipulate the data in the core file.
1706 GDB reopens the source file on every line, as you "next" through it.
1708 [still true? I've a memory of this being fixed]
1712 Perhaps "i source" should take an argument like that of "list".
1716 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1717 it matches the source line indicated.
1721 The prompt at end of screen should accept space as well as CR.
1725 Backtrace should point out what the currently selected frame is, in
1726 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1727 ...)" rather than "#3 foo (bar, ...)".
1731 "i program" should work for core files, and display more info, like what
1732 actually caused it to die.
1736 "x/10i" should shorten the long name, if any, on subsequent lines.
1740 "next" over a function that longjumps, never stops until next time you happen
1741 to get to that spot by accident. E.g. "n" over execute_command which has
1746 "set zeroprint off", don't bother printing members of structs which
1747 are entirely zero. Useful for those big structs with few useful
1752 GDB does four ioctl's for every command, probably switching terminal modes
1753 to/from inferior or for readline or something.
1757 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1758 if the state is the same, too.
1762 "i frame" shows wrong "arglist at" location, doesn't show where the args
1763 should be found, only their actual values.
1767 There should be a way for "set" commands to validate the new setting
1768 before it takes effect.
1772 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1776 i line VAR produces "Line number not known for symbol ``var''.". I
1777 thought we were stashing that info now!
1781 We should be able to write to random files at hex offsets like adb.
1785 [elena - delete this]
1787 Handle add_file with separate text, data, and bss addresses. Maybe
1788 handle separate addresses for each segment in the object file?
1792 [Jimb/Elena delete this one]
1794 Handle free_named_symtab to cope with multiply-loaded object files
1795 in a dynamic linking environment. Should remember the last copy loaded,
1796 but not get too snowed if it finds references to the older copy.
1800 [elena delete this also]
1802 Remove all references to:
1809 now that we have BFD. All remaining are in machine dependent files.
1813 Re-organize help categories into things that tend to fit on a screen
1818 Add in commands like ADB's for searching for patterns, etc. We should
1819 be able to examine and patch raw unsymboled binaries as well in gdb as
1820 we can in adb. (E.g. increase the timeout in /bin/login without source).
1822 [actually, add ADB interface :-]
1826 When doing "step" or "next", if a few lines of source are skipped between
1827 the previous line and the current one, print those lines, not just the
1828 last line of a multiline statement.
1832 Handling of "&" address-of operator needs some serious overhaul
1833 for ANSI C and consistency on arrays and functions.
1834 For "float point[15];":
1835 ptype &point[4] ==> Attempt to take address of non-lvalue.
1836 For "char *malloc();":
1837 ptype malloc ==> "char *()"; should be same as
1838 ptype &malloc ==> "char *(*)()"
1839 call printf ("%x\n", malloc) ==> weird value, should be same as
1840 call printf ("%x\n", &malloc) ==> correct value
1844 Fix dbxread.c symbol reading in the presence of interrupts. It
1845 currently leaves a cleanup to blow away the entire symbol table when a
1846 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1848 [I suspect that the grype was that, on a slow system, you might want
1849 to cntrl-c and get just half the symbols and then load the rest later
1850 - scary to be honest]
1854 Mipsread.c reads include files depth-first, because the dependencies
1855 in the psymtabs are way too inclusive (it seems to me). Figure out what
1856 really depends on what, to avoid recursing 20 or 30 times while reading
1861 value_add() should be subtracting the lower bound of arrays, if known,
1862 and possibly checking against the upper bound for error reporting.
1866 When listing source lines, check for a preceding \n, to verify that
1867 the file hasn't changed out from under us.
1869 [fixed by some other means I think. That hack wouldn't actually work
1870 reliably - the file might move such that another \n appears. ]
1874 Get all the remote systems (where the protocol allows it) to be able to
1875 stop the remote system when the GDB user types ^C (like remote.c
1876 does). For ebmon, use ^Ak.
1880 Possible feature: A version of the "disassemble" command which shows
1881 both source and assembly code ("set symbol-filename on" is a partial
1884 [has this been done? It was certainly done for MI and GDBtk]
1888 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1889 using a 0 address for bad purposes internally).
1893 Make "info path" and path_command work again (but independent of the
1894 environment either of gdb or that we'll pass to the inferior).
1898 Make GDB understand the GCC feature for putting octal constants in
1899 enums. Make it so overflow on an enum constant does not error_type
1900 the whole type. Allow arbitrarily large enums with type attributes.
1901 Put all this stuff in the testsuite.
1905 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1906 the value in hex; process type attributes). Add this to the
1907 testsuite. This way future compilers can add new types and old
1908 versions of GDB can do something halfway reasonable.
1912 Fix mdebugread.c:parse_type to do fundamental types right (see
1913 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1914 that the debug format fixes the sizes of these things and it shouldn't
1915 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1916 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1917 be aware of that). Also use a switch statement for clarity and speed.
1921 Investigate adding symbols in target_load--some targets do, some
1926 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1927 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1932 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1937 Make a watchpoint which contains a function call an error (it is
1938 broken now, making it work is probably not worth the effort).
1942 New test case based on weird.exp but in which type numbers are not
1943 renumbered (thus multiply defining a type). This currently causes an
1944 infinite loop on "p v_comb".
1948 [Hey! Hint Hint Delete Delete!!!]
1950 Fix 386 floating point so that floating point registers are real
1951 registers (but code can deal at run-time if they are missing, like
1952 mips and 68k). This would clean up "info float" and related stuff.
1956 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1957 about not being able to access memory location 0.
1959 -------------------- enummask.c
1980 If try to modify value in file with "set write off" should give
1981 appropriate error not "cannot access memory at address 0x65e0".
1985 Allow core file without exec file on RS/6000.
1989 Make sure "shell" with no arguments works right on DOS.
1993 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1994 the same directory can be NFS-mounted on unix or DOS, and work the
1999 [Is this another delete???]
2001 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
2002 get RS/6000 to work right, might not be immediately relevant).
2006 Work out some kind of way to allow running the inferior to be done as
2007 a sub-execution of, eg. breakpoint command lists. Currently running
2008 the inferior interupts any command list execution. This would require
2009 some rewriting of wait_for_inferior & friends, and hence should
2010 probably be done in concert with the above.
2014 Add function arguments to gdb user defined functions.
2018 Add convenience variables that refer to exec file, symbol file,
2019 selected frame source file, selected frame function, selected frame
2024 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2025 allocate a partial symtab for each BINCL/EINCL grouping. This will
2026 seriously decrease the size of inter-psymtab dependencies and hence
2027 lessen the amount that needs to be read in when a new source file is
2032 Add a command for searching memory, a la adb. It specifies size,
2033 mask, value, start address. ADB searches until it finds it or hits
2034 an error (or is interrupted).
2038 Remove the range and type checking code and documentation, if not