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@
204 configure.{in,host,tgt}
206 *-tdep.c *-nat.c *-xdep.c
213 Change documentation to GFDL license.
215 ``It is time to make an effort to start using the GFDL more
216 thoroughly. Would all GNU maintainers please change the license to
217 the GFDL, for all manuals and other major documentation files?
219 The GFDL and some instructions for using it can be found in
220 http://www.gnu.org/copyleft/''
226 Fix copyright notices.
228 Turns out that ``1998-2000'' isn't considered valid :-(
230 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
234 GDB 5.1 - Known Problems
235 ========================
241 The z8k has suffered bit rot and is known to not build. The problem
242 was occuring in the opcodes directory.
248 The m88k has suffered bit rot and is known to not build.
252 Solaris 8 x86 CURSES_H problem
253 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
255 The original problem was worked around with:
257 2000-06-06 Michael Snyder <msnyder@cygnus.com>
259 * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
260 * configure: Regenerate.
262 When building both GDB and SID using the same source tree the problem
263 will still occure. sid/component/configure.in mis-configures
264 <curses.h> and leaves wrong information in the config cache.
273 Thread support. Right now, as soon as a thread finishes and exits,
274 you're hosed. This problem is reported once a week or so.
278 GDB 5.2 - New features
279 ======================
283 GCC 3.0 ABI support (but hopefully sooner...).
287 Objective C/C++ support (but hopefully sooner...).
294 The following cleanups have been identified as part of GDB 5.2.
298 Remove old code that does not use ui_out functions and all the related
299 "ifdef"s. This also allows the elimination of -DUI_OUT from
300 Makefile.in and configure.in.
306 Eliminate all warnings for at least one host/target for the flags:
307 -Wimplicit -Wreturn-type -Wcomment -Wtrigraphs -Wformat -Wparentheses
308 -Wpointer-arith -Wuninitialized
312 Deprecate, if not delete, the following:
318 Replaced by, on the target side
320 and on core-gdb side:
321 {read,write}_register_gen()
322 Remote.c will need to use something
323 other than REGISTER_BYTE() and
324 REGISTER_RAW_SIZE() when unpacking
327 STORE_PSEUDO_REGISTER
328 FETCH_PSEUDO_REGISTER
329 Now handed by the methods
330 gdbarch_{read,write}_register()
331 which sits between core GDB and
335 REGISTER_CONVERT_TO_RAW
336 REGISTER_CONVERT_TO_VIRTUAL
337 I think these three are redundant.
338 gdbarch_register_{read,write} can
339 do any conversion it likes.
341 REGISTER_VIRTUAL_SIZE
342 MAX_REGISTER_VIRTUAL_SIZE
343 REGISTER_VIRTUAL_TYPE
344 I think these can be replaced by
346 FRAME_REGISTER_TYPE(frame, regnum)
347 REGISTER_TYPE(regnum)
351 FRAME_REGISTER_INFO (frame, ...)
354 If nothing else rename this so that
355 how it relates to rawreg and the
359 The size of the cache can be computed
364 Restructure gdb directory tree so that it avoids any 8.3 and 14
369 Convert GDB build process to AUTOMAKE.
371 See also sub-directory configure below.
373 The current convention is (kind of) to use $(<header>_h) in all
374 dependency lists. It isn't done in a consistent way.
378 GDB 5.2 - Known Problems
379 ========================
383 Code Cleanups: General
384 ======================
386 The following are more general cleanups and fixes. They are not tied
387 to any specific release.
391 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
395 Can't build IRIX -> arm GDB.
396 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
399 > Now I'm building for an embedded arm target. If there is a way of turning
400 > remote-rdi off, I couldn't find it. It looks like it gets built by default
401 > in gdb/configure.tgt(line 58) Anyway, the build dies in
402 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
403 > aren't one of the architectures supported.
407 Problem with weak functions
408 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
410 Dan Nicolaescu writes:
411 > It seems that gdb-4.95.1 does not display correctly the function when
412 > stoping in weak functions.
414 > It stops in a function that is defined as weak, not in the function
415 > that is actually run...
419 Follow through `make check' with --enable-shared.
421 When the srcware tree is configured with --enable-shared, the `expect'
422 program won't run properly. Jim Wilson found out gdb has a local hack
423 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
426 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
430 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
432 Patches in the database.
436 printcmd.c (print_address_numeric):
438 NOTE: This assumes that the significant address information is kept in
439 the least significant bits of ADDR - the upper bits were either zero
440 or sign extended. Should ADDRESS_TO_POINTER() or some
441 ADDRESS_TO_PRINTABLE() be used to do the conversion?
445 The BFD directory requires bug-fixed AUTOMAKE et.al.
447 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
448 contained the full path to texinfo.tex when it should have only
449 contained the directory. The bug has been fixed in the current
450 AUTOMAKE sources. Automake snapshots can be found in:
451 ftp://sourceware.cygnus.com/pub/gdb/snapshots
452 and ftp://sourceware.cygnus.com/pub/binutils
456 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
457 determine the default isa/byte-order.
461 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
462 BIG_ENDIAN and LITTLE_ENDIAN.
466 Eliminate more compiler warnings.
468 Of course there also needs to be the usual debate over which warnings
469 are valid and how to best go about this.
471 One method: choose a single option; get agreement that it is
472 reasonable; try it out to see if there isn't anything silly about it
473 (-Wunused-parameters is an example of that) then incrementally hack
476 The other method is to enable all warnings and eliminate them from one
481 Elimination of ``(catch_errors_ftype *) func''.
483 Like make_cleanup_func it isn't portable.
484 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
485 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
489 Nuke #define CONST_PTR.
497 [PATCH/5] src/intl/Makefile.in:distclean additions
498 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
500 Do not forget to merge the patch back into the trunk.
504 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
506 At present defs.h includes <endian.h> (which is linux specific) yet
507 almost nothing depends on it. Suggest "gdb_endian.h" which can also
508 handle <machine/endian.h> and only include that where it is really
513 Replace savestring() with something from libiberty.
515 An xstrldup()? but that would have different semantics.
519 Rationalize use of floatformat_unknown in GDB sources.
521 Instead of defaulting to floatformat_unknown, should hosts/targets
522 specify the value explicitly?
524 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
528 Add a ``name'' member to include/floatformat.h:struct floatformat.
529 Print that name in gdbarch.c.
533 Sort out the harris mess in include/floatformat.h (it hardwires two
534 different floating point formats).
538 See of the GDB local floatformat_do_doublest() and libiberty's
539 floatformat_to_double (which was once GDB's ...) can be merged some
544 Eliminate mmalloc(), mstrsave() et.al. from GDB.
546 Also eliminate it from defs.h.
550 Eliminate PTR. ISO-C allows ``void *''.
556 GDB should never abort. GDB should either throw ``error ()'' or
557 ``internal_error ()''. Better still GDB should naturally unwind with
562 GDB probably doesn't build on FreeBSD pre 2.2.x
563 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
565 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
570 Deprecate "fg". Apparently ``fg'' is actually continue.
572 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
576 Deprecate current use of ``floatformat_unknown''.
578 Require all targets to explicitly provide their float format instead
579 of defaulting to floatformat unknown. Doing the latter leads to nasty
582 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
586 Rationalize floatformat_to_double() vs floatformat_to_doublest().
588 Looks like GDB migrated floatformat_to_double() to libiberty but then
589 turned around and created a ..._to_doublest() the latter containing
592 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
596 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
598 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
602 The ``maintenance deprecate set endian big'' command doesn't notice
603 that it is deprecating ``set endian'' and not ``set endian big'' (big
604 is implemented using an enum). Is anyone going to notice this?
608 When tab expanding something like ``set arch<tab>'' ignore the
609 deprecated ``set archdebug'' and expand to ``set architecture''.
613 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
614 like from arm-tdep.c.
618 Fix uses of ->function.cfunc = set_function().
620 The command.c code calls sfunc() when a set command. Rather than
621 change it suggest fixing the callback function so that it is more
624 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
626 See also ``Fix implementation of ``target xxx''.'' below.
630 IRIX 3.x support is probably broken.
634 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
635 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
637 Apart from the d30v, are there any sim/common simulators that make use
640 A brief summary of what happened is that sim/common/sim-break.c was
641 created as a good idea. It turned out a better idea was to use
642 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
646 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
647 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
651 Eliminate ``extern'' from C files.
655 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
657 Extreme care is recommeded - perhaps only modify tests that are
658 exercised by the testsuite (as determined using some type of code
663 Replace the file gdb/CONTRIBUTE with a file that is generated from the
664 gdb/doc/*.texinfo directory.
668 New Features and Fixes
669 ======================
671 These are harder than cleanups but easier than work involving
672 fundamental architectural change.
676 Hardware watchpoint problems on x86 OSes, including Linux:
678 1. Delete/disable hardware watchpoints should free hardware debug
680 2. Watch for different values on a viariable with one hardware debug
683 According to Eli Zaretskii <eliz@delorie.com>:
685 These are not GDB/ia32 issues per se: the above features are all
686 implemented in the DJGPP port of GDB and work in v5.0. Every
687 x86-based target should be able to lift the relevant parts of
688 go32-nat.c and use them almost verbatim. You get debug register
689 sharing through reference counts, and the ability to watch large
690 regions (up to 16 bytes) using multiple registers. (The required
691 infrastructure in high-level GDB application code, mostly in
692 breakpoint.c, is also working since v5.0.)
696 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
697 so that you can see how the GDB was created.
701 Add an "info bfd" command that displays supported object formats,
702 similarly to objdump -i.
704 Is there a command already?
708 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
710 This requires internationalization.
716 (gdb) p fwprintf(stdout,L"%S\n", f)
717 No symbol "L" in current context.
721 Cleanup configury support for optional sub-directories.
723 Check how GCC handles multiple front ends for an example of how things
724 could work. A tentative first step is to rationalize things so that
725 all sub directories are handled in a fashion similar to gdb/mi.
727 See also automake above.
731 Add a transcript mechanism to GDB.
733 Such a mechanism might log all gdb input and output to a file in a
734 form that would allow it to be replayed. It could involve ``gdb
735 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
739 Can the xdep files be replaced by autoconf?
743 Document trace machinery
747 Document ui-out and ui-file.
749 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
753 Update texinfo.tex to latest?
757 Incorporate agentexpr.texi into gdb.texinfo
759 agentexpr.texi mostly describes the details of the byte code used for
760 tracepoints, not the internals of the support for this in GDB. So it
761 looks like gdb.texinfo is a better place for this information.
763 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
767 Document overlay machinery.
771 ``(gdb) catch signal SIGNAL''
773 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
774 You can attach commands to a catch but not a handle. A handle has a
775 limited number of hardwired actions.
781 o readline/*.h bitrot
783 The TUI isn't up-to-date with
784 respect to the readline currently
785 bundled with GDB. Importing a
786 new readline is on the 5.1 wish
787 list so this can only get worse.
789 Grep for things like term_cursor_move.
791 (To be honest, I don't see anyone
792 importing a new readline before 5.1 is
795 o tui.c:va_catch_errors() bitrot
797 This nasty piece of work used knowledge
798 of the internals of GDBs error functions :-(
799 Ever since those internals were cleaned
800 up this code has been broken. :-(
802 o tuiWin.c:c_makeVisibleWithNewHeight() broken
803 tuiLayout.c:_extractDisplayStartAddr() broken
805 Both these function call find_line_pc()
806 incorrectly (wrong args, wrong return value).
808 I suspect this bug has always been there!
809 It had been hidden because those files
810 didn't include the necessary header files
813 o tuiRegs() host dependant
815 Not suprisingly, this isn't a very portable
816 section of code. However, I'm sure people
817 could live with no regs in the short to
820 o defs.h: #include "tui.h" et.al.
822 I'm not sure where this came from.
823 It was a really bad idea.
825 To get things to compile I did a nasty
826 hack (Just declare what was needed and
827 replace any expressions like xx->y.z()
828 in GDB proper with function calls). I
829 could commit it slightly cleaned up if
832 Medium Term. the #ifdef TUI and TuiDo()
833 should be changed to hooks (like GDBTK).
834 The gdb-events.[hc] is there for that
837 o tui.c:_tuiReset() host dependant
839 tui.c contains a lump of termio[s]
840 I suspect an equivalent block of
841 code can be lifted from readline.
842 An equivalent readline function may
845 o curses.h vs ncurses.h.
847 Simple portability problem.
851 This function is a mystery - where is it?
853 o tui-file.[hc] cleanup
855 This can be significantly simplified.
857 o The code should be pacified. (-Werror -W...)
859 There are plenty of #includes,
860 duplicate #includes, missing function decls
863 Some of the problems I found were through
864 fixing a few of the warnings.
866 o The code should be GNUtified.
868 It would be very nice to have this code
869 look like the rest of GDB. That way people
870 would be more accepting of it as a true
873 Until it is GNUtified it is going to stick
874 out like a sore thumb to the programmer.
876 o The code should be clearly copyrighted
878 (FSF, with due credit to HP)
882 Add support for ``gdb --- PROGRAM ARGS ...''.
883 Add support for ``gdb -cmd=...''
885 Along with many variations. Check:
887 ????? for a full discussion.
893 Implement ``(gdb) !ls''.
895 Which is very different from ``(gdb) ! ls''. Implementing the latter
898 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
902 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
907 When tab expanding something like ``set arch<tab>'' ignore the
908 deprecated ``set archdebug'' and expand to ``set architecture''.
912 Replace the code that uses the host FPU with an emulator of the target
917 The "ocd reset" command needs to flush the dcache, which requires breaking
918 the abstraction layer between the target independent and target code. One
919 way to address this is provide a generic "reset" command and target vector.
921 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
930 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
931 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
933 The thread_db assisted debugging code doesn't handle exiting threads
934 properly, at least in combination with glibc 2.1.3 (the framework is
935 there, just not the actual code). There are at least two problems
936 that prevent this from working.
938 As an additional reference point, the pre thread_db code did not work
943 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
945 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
947 Christopher Blizzard writes:
949 So, I've done some more digging into this and it looks like Jim
950 Kingdon has reported this problem in the past:
952 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
954 I can reproduce this problem both with and without Tom's patch. Has
955 anyone seen this before? Maybe have a solution for it hanging around?
958 There's a test case for this documented at:
960 when debugging threaded applications you get extra SIGTRAPs
961 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
963 [There should be a GDB testcase - cagney]
967 GDB5 TOT on unixware 7
968 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
971 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
972 > practical matter, the current thread support is somewhat more annoying
973 > than when GDB was thread-unaware.
980 New languages come onto the scene all the time.
984 Re: Various C++ things
986 value_headof/value_from_vtable_info are worthless, and should be
987 removed. The one place in printcmd.c that uses it should use the RTTI
990 RTTI for g++ should be using the typeinfo functions rather than the
991 vtables. The typeinfo functions are always at offset 4 from the
992 beginning of the vtable, and are always right. The vtables will have
993 weird names like E::VB sometimes. The typeinfo function will always
994 be "E type_info function", or somesuch.
996 value_virtual_fn_field needs to be fixed so there are no failures for
997 virtual functions for C++ using g++.
999 Testsuite cases are the major priority right now for C++ support,
1000 since i have to make a lot of changes that could potentially break
1005 Add support for Modula3
1007 Get DEC/Compaq to contribute their Modula-3 support.
1011 Remote Protocol Support
1012 =======================
1016 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
1017 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
1019 (Broken) support for GDB's remote protocol across UDP is to be
1020 included in the follow-on release.
1022 It should be noted that UDP can only work when the [Gg] packet fits in
1023 a single UDP packet.
1025 There is also much debate over the merit of this.
1029 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
1031 Add support for packet enable/disable commands with these thread
1032 packets. General cleanup.
1034 [PATCH] Document the ThreadInfo remote protocol queries
1035 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
1037 [PATCH] "info threads" queries for remote.c
1038 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
1042 Remote protocol doco feedback.
1044 Too much feedback to mention needs to be merged in (901660). Search
1045 for the word ``remote''.
1048 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
1049 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
1050 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
1054 GDB doesn't recover gracefully from remote protocol errors.
1056 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
1057 ignored and a timeout is required before GDB retries. A pre-cursor to
1058 fixing this this is making GDB's remote protocol packet more robust.
1060 While downloading to a remote protocol target, gdb ignores packet
1061 errors in so far as it will continue to download with chunk N+1 even
1062 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
1063 take a painfully long time to run. As a PS that test needs to be
1064 fixed so that it builds on 16 bit machines.
1068 Fix the ``!'' packet.
1070 JT reported that the existing targets do, in fact return ``OK'' so it
1071 is possible to merge remote and extended-remote targets.
1075 Drop ``<address>'' from the [SsCc] packets.
1077 I don't think that GDB generates them so having it in the protocol is
1082 Fix doco on the ``q'' packet.
1084 It has evolved into a generic RPC. The notes should reflect this and,
1085 perhaps, the ``Q'' packet can be deprecated.
1087 The doco should mention that ``OK'' is a valid packet response.
1089 The doco should explain why ``OK'' needs to be a valid packet
1094 Add the cycle step command.
1096 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1100 Resolve how to scale things to support very large packets.
1104 Resolve how to handle a target that changes things like its endianess
1105 on the fly - should it be returned in the ``T'' packet?
1107 Underlying problem is that the register file is target endian. If the
1108 target endianess changes gdb doesn't know.
1115 If / when GDB starts to support the debugging of multi-processor
1116 (rather than multi-thread) applications the symtab code will need to
1117 be updated a little so that several independent symbol tables are
1118 active at a given time.
1120 The other interesting change is a clarification of the exact meaning
1121 of CORE_ADDR and that has had consequences for a few targets (that
1122 were abusing that data type).
1126 Investiagate ways of reducing memory.
1130 Investigate ways of improving load time.
1134 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1136 Consequence of recent symtab clarification. No marks for figuring out
1137 who maintains the d10v.
1141 Get the MIPS to correctly sign extend all address <-> pointer
1144 Consequence of recent symtab clarification. No marks for figuring out
1145 who maintains the MIPS.
1149 GDB truncates 64 bit enums.
1151 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1158 There are never to many testcases.
1162 Better thread testsuite.
1166 Better C++ testsuite.
1170 Look at adding a GDB specific testsuite directory so that white box
1171 tests of key internals can be added (eg ui_file).
1175 Separate out tests that involve the floating point (FP).
1177 (Something for people brining up new targets). FP and non-fp tests
1178 are combined. I think there should be set of basic tests that
1179 exercise pure integer support and then a more expanded set that
1180 exercise FP and FP/integer interactions.
1182 As an example, the MIPS, for n32 as problems with passing FP's and
1183 structs. Since most inferior call tests include FP it is difficult to
1184 determine of the integer tests are ok.
1188 Architectural Changes: General
1189 ==============================
1191 These are harder than simple cleanups / fixes and, consequently
1192 involve more work. Typically an Architectural Change will be broken
1193 down into a more digestible set of cleanups and fixes.
1197 Cleanup software single step.
1199 At present many targets implement software single step by directly
1200 blatting memory (see rs6000-tdep.c). Those targets should register
1201 the applicable breakpoints using the breakpoint framework. Perhaphs a
1202 new internal breakpoint class ``step'' is needed.
1206 Replace READ_FP() with FRAME_HANDLE().
1208 READ_FP() is a hangover from the days of the vax when the ABI really
1209 did have a frame pointer register. Modern architectures typically
1210 construct a virtual frame-handle from the stack pointer and various
1211 other bits of string.
1213 Unfortunately GDB still treats this synthetic FP register as though it
1214 is real. That in turn really confuses users (arm and ``print $fp'' VS
1215 ``info registers fp''). The synthetic FP should be separated out of
1216 the true register set presented to the user.
1220 Register Cache Cleanup (below from Andrew Cagney)
1222 I would depict the current register architecture as something like:
1224 High GDB --> Low GDB
1229 register + REGISTER_BYTE(reg_nr)
1232 -------------------------
1233 | extern register[] |
1234 -------------------------
1236 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1237 really clear on what mechanisms they should be using to manipulate that
1238 buffer. Further, much code assumes, dangerously, that registers are
1239 contigious. Having got mips-tdep.c to support multiple ABIs, believe
1240 me, that is a bad assumption. Finally, that register cache layout is
1241 determined by the current remote/local target and _not_ the less
1242 specific target ISA. In fact, in many cases it is determined by the
1243 somewhat arbitrary layout of the [gG] packets!
1246 How I would like the register file to work is more like:
1274 The main objectives being:
1276 o a clear separation between the low
1277 level target and the high level GDB
1279 o a mechanism that solves the general
1280 problem of register aliases, overlaps
1281 etc instead of treating them as optional
1282 extras that can be wedged in as an after
1283 thought (that is a reasonable description
1284 of the current code).
1286 Identify then solve the hard case and the
1287 rest just falls out. GDB solved the easy
1288 case and then tried to ignore the real
1291 o a removal of the assumption that the
1292 mapping between the register cache
1293 and virtual registers is largely static.
1294 If you flip the USR/SSR stack register
1295 select bit in the status-register then
1296 the corresponding stack registers should
1299 o a mechanism that clearly separates the
1300 gdb internal register cache from any
1301 target (not architecture) dependent
1302 specifics such as [gG] packets.
1304 Of course, like anything, it sounds good in theory. In reality, it
1305 would have to contend with many<->many relationships at both the
1306 virt<->cache and cache<->target level. For instance:
1309 Modifying an mmx register may involve
1310 scattering values across both FP and
1311 mmpx specific parts of a buffer
1314 When writing back a SP it may need to
1315 both be written to both SP and USP.
1320 Rather than let this like the last time it was discussed, just slip, I'm
1321 first going to add this e-mail (+ references) to TODO. I'd then like to
1322 sketch out a broad strategy I think could get us there.
1325 First thing I'd suggest is separating out the ``extern registers[]''
1326 code so that we can at least identify what is using it. At present
1327 things are scattered across many files. That way we can at least
1328 pretend that there is a cache instead of a global array :-)
1330 I'd then suggest someone putting up a proposal for the pseudo-reg /
1331 high-level side interface so that code can be adopted to it. For old
1332 code, initially a blanket rename of write_register_bytes() to
1333 deprecated_write_register_bytes() would help.
1335 Following that would, finaly be the corresponding changes to the target.
1339 Check that GDB can handle all BFD architectures (Andrew Cagney)
1341 There should be a test that checks that BFD/GDB are in sync with
1342 regard to architecture changes. Something like a test that first
1343 queries GDB for all supported architectures and then feeds each back
1344 to GDB.. Anyone interested in learning how to write tests? :-)
1348 Architectural Change: Multi-arch et al.
1349 =======================================
1351 The long term objective is to remove all assumptions that there is a
1352 single target with a single address space with a single instruction
1353 set architecture and single application binary interface.
1355 This is an ongoing effort. The first milestone is to enable
1356 ``multi-arch'' where by all architectural decisions are made at
1359 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1360 ``gdbisa''. Once things are multi-arched breaking that down correctly
1361 will become much easier.
1365 GDBARCH cleanup (Andrew Cagney)
1367 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1368 into arch-utils.[hc].
1370 Document that gdbarch_init_ftype could easily fail because it didn't
1371 identify an architecture.
1375 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
1377 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1382 Fix target_signal_from_host() etc.
1384 The name is wrong for starters. ``target_signal'' should probably be
1385 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
1386 After that it needs to be multi-arched and made independent of any
1387 host signal numbering.
1391 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1394 This is a barrier to replacing mips_extra_func_info with something
1395 that works with multi-arch.
1399 Multi-arch mips_extra_func_info.
1401 This first needs the alpha to be updated so that it uses ``struct
1406 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1408 Surely one of them is redundant.
1412 Convert ALL architectures to MULTI-ARCH.
1416 Select the initial multi-arch ISA / ABI based on --target or similar.
1418 At present the default is based on what ever is first in the BFD
1419 archures table. It should be determined based on the ``--target=...''
1424 Make MIPS pure multi-arch.
1426 It is only at the multi-arch enabled stage.
1432 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1434 Can the tm.h and nm.h files be eliminated by multi-arch.
1438 Architectural Change: MI, LIBGDB and scripting languages
1439 ========================================================
1441 See also architectural changes related to the event loop. LIBGDB
1442 can't be finished until there is a generic event loop being used by
1445 The long term objective is it to be possible to integrate GDB into
1446 scripting languages.
1450 Implement generic ``(gdb) commmand > file''
1452 Once everything is going through ui_file it should be come fairly
1455 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1459 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1461 gdb_stdtarg is easily confused with gdb_stdarg.
1465 Extra ui_file methods - dump.
1467 Very useful for whitebox testing.
1471 Eliminate error_begin().
1473 With ui_file, there is no need for the statefull error_begin ()
1478 Send normal output to gdb_stdout.
1479 Send error messages to gdb_stderror.
1480 Send debug and log output log gdb_stdlog.
1482 GDB still contains many cases where (f)printf or printf_filtered () is
1483 used when it should be sending the messages to gdb_stderror or
1484 gdb_stdlog. The thought of #defining printf to something has crossed
1489 Re-do GDB's output pager.
1491 GDB's output pager still relies on people correctly using *_filtered
1492 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1493 Hopefully, with all normal output going to gdb_stdout, the pager can
1494 just look at the ui_file that the output is on and then use that to
1495 decide what to do about paging. Sounds good in theory.
1499 Check/cleanup MI documentation.
1501 The list of commands specified in the documentation needs to be
1502 checked against the mi-cmds.c table in a mechanical way (so that they
1503 two can be kept up-to-date).
1507 Convert MI into libgdb
1509 MI provides a text interface into what should be many of the libgdb
1510 functions. The implementation of those functions should be separated
1511 into the MI interface and the functions proper. Those functions being
1512 moved to gdb/lib say.
1518 The first part can already be found in defs.h.
1522 MI's input does not use buffering.
1524 At present the MI interface reads raw characters of from an unbuffered
1525 FD. This is to avoid several nasty buffer/race conditions. That code
1526 should be changed so that it registers its self with the event loop
1527 (on the input FD) and then push commands up to MI as they arrive.
1529 The serial code already does this.
1533 Make MI interface accessible from existing CLI.
1537 Add a breakpoint-edit command to MI.
1539 It would be similar to MI's breakpoint create but would apply to an
1540 existing breakpoint. It saves the need to delete/create breakpoints
1541 when ever they are changed.
1545 Add directory path to MI breakpoint.
1547 That way the GUI's task of finding the file within which the
1548 breakpoint was set is simplified.
1552 Add a mechanism to reject certain expression classes to MI
1554 There are situtations where you don't want GDB's expression
1555 parser/evaluator to perform inferior function calls or variable
1556 assignments. A way of restricting the expression parser so that such
1557 operations are not accepted would be very helpful.
1561 Remove sideffects from libgdb breakpoint create function.
1563 The user can use the CLI to create a breakpoint with partial
1564 information - no file (gdb would use the file from the last
1567 The libgdb interface currently affects that environment which can lead
1568 to confusion when a user is setting breakpoints via both the MI and
1571 This is also a good example of how getting the CLI ``right'' will be
1576 Move gdb_lasterr to ui_out?
1578 The way GDB throws errors and records them needs a re-think. ui_out
1579 handles the correct output well. It doesn't resolve what to do with
1580 output / error-messages when things go wrong.
1584 do_setshow_command contains a 1024 byte buffer.
1586 The function assumes that there will never be any more than 1024 bytes
1587 of enum. It should use mem_file.
1591 Should struct cmd_list_element . completer take the command as an
1596 Should the bulk of top.c:line_completion_function() be moved to
1597 command.[hc]? complete_on_cmdlist() and complete_on_enums() could
1598 then be made private.
1602 top.c (execute_command): Should a command being valid when the target
1603 is running be made an attribute (predicate) to the command rather than
1604 an explicit set of tests.
1608 top.c (execute_command): Should the bulk of this function be moved
1609 into command.[hc] so that top.c doesn't grub around in the command
1614 Architectural Change: Async
1615 ===========================
1617 While GDB uses an event loop when prompting the user for input. That
1618 event loop is not exploited by targets when they allow the target
1619 program to continue. Typically targets still block in (target_wait())
1620 until the program again halts.
1622 The closest a target comes to supporting full asynchronous mode are
1623 the remote targets ``async'' and ``extended-async''.
1627 Asynchronous expression evaluator
1629 Inferior function calls hang GDB.
1633 Fix implementation of ``target xxx''.
1635 At present when the user specifies ``target xxxx'', the CLI maps that
1636 directly onto a target open method. It is then assumed that the
1637 target open method should do all sorts of complicated things as this
1638 is the only chance it has. Check how the various remote targets
1639 duplicate the target operations. Check also how the various targets
1640 behave differently for purely arbitrary reasons.
1642 What should happen is that ``target xxxx'' should call a generic
1643 ``target'' function and that should then co-ordinate the opening of
1644 ``xxxx''. This becomes especially important when you're trying to
1645 open an asynchronous target that may need to perform background tasks
1646 as part of the ``attach'' phase.
1648 Unfortunately, due to limitations in the old/creaking command.h
1649 interface, that isn't possible. The function being called isn't told
1650 of the ``xxx'' or any other context information.
1652 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1653 CLI code so that it passes to the callback function (attatched to a
1654 command) useful information such as the actual command and a context
1655 for that command. Other changes such as making ``struct command''
1656 opaque may also help.
1659 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1663 Make "target xxx" command interruptible.
1665 As things become async this becomes possible. A target would start
1666 the connect and then return control to the event loop. A cntrl-c
1667 would notify the target that the operation is to be abandoned and the
1668 target code could respond.
1672 Add a "suspend" subcommand of the "continue" command to suspend gdb
1673 while continuing execution of the subprocess. Useful when you are
1674 debugging servers and you want to dodge out and initiate a connection
1675 to a server running under gdb.
1684 Frequently requested but not approved requests.
1688 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1690 The benefits on this one are thought to be marginal - GDBs design
1691 means that unused parameters are very common. GCC 3.0 will also
1692 include the option -Wno-unused-parameter which means that ``-Wall
1693 -Wno-unused-parameters -Werror'' can be specified.
1702 This list is not up to date, and opinions vary about the importance or
1703 even desirability of some of the items. If you do fix something, it
1704 always pays to check the below.
1708 @c This does not work (yet if ever). FIXME.
1709 @c @item --parse=@var{lang} @dots{}
1710 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1711 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1712 @c list of all supported languages, omit the argument. Without this
1713 @c option, @value{GDBN} is configured to parse all supported languages.
1717 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1718 is its default value. Clean this up.
1722 It should be possible to use symbols from shared libraries before we know
1723 exactly where the libraries will be loaded. E.g. "b perror" before running
1724 the program. This could maybe be done as an extension of the "breakpoint
1725 re-evaluation" after new symbols are loaded.
1729 Make single_step() insert and remove breakpoints in one operation.
1731 [If this is talking about having single_step() insert the breakpoints,
1732 run the target then pull the breakpoints then it is wrong. The
1733 function has to return as control has to eventually be passed back to
1734 the main event loop.]
1738 Speed up single stepping by avoiding extraneous ptrace calls.
1742 Speed up single stepping by not inserting and removing breakpoints
1743 each time the inferior starts and stops.
1745 Breakpoints should not be inserted and deleted all the time. Only the
1746 one(s) there should be removed when we have to step over one. Support
1747 breakpoints that don't have to be removed to step over them.
1749 [this has resulted in numerous debates. The issue isn't clear cut]
1753 Provide "voodoo" debugging of core files. This creates a zombie
1754 process as a child of the debugger, and loads it up with the data,
1755 stack, and regs of the core file. This allows you to call functions
1756 in the executable, to manipulate the data in the core file.
1762 GDB reopens the source file on every line, as you "next" through it.
1764 [still true? I've a memory of this being fixed]
1768 Perhaps "i source" should take an argument like that of "list".
1772 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1773 it matches the source line indicated.
1777 The prompt at end of screen should accept space as well as CR.
1781 Backtrace should point out what the currently selected frame is, in
1782 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1783 ...)" rather than "#3 foo (bar, ...)".
1787 "i program" should work for core files, and display more info, like what
1788 actually caused it to die.
1792 "x/10i" should shorten the long name, if any, on subsequent lines.
1796 "next" over a function that longjumps, never stops until next time you happen
1797 to get to that spot by accident. E.g. "n" over execute_command which has
1802 "set zeroprint off", don't bother printing members of structs which
1803 are entirely zero. Useful for those big structs with few useful
1808 GDB does four ioctl's for every command, probably switching terminal modes
1809 to/from inferior or for readline or something.
1813 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1814 if the state is the same, too.
1818 "i frame" shows wrong "arglist at" location, doesn't show where the args
1819 should be found, only their actual values.
1823 There should be a way for "set" commands to validate the new setting
1824 before it takes effect.
1828 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1832 i line VAR produces "Line number not known for symbol ``var''.". I
1833 thought we were stashing that info now!
1837 We should be able to write to random files at hex offsets like adb.
1841 [elena - delete this]
1843 Handle add_file with separate text, data, and bss addresses. Maybe
1844 handle separate addresses for each segment in the object file?
1848 [Jimb/Elena delete this one]
1850 Handle free_named_symtab to cope with multiply-loaded object files
1851 in a dynamic linking environment. Should remember the last copy loaded,
1852 but not get too snowed if it finds references to the older copy.
1856 [elena delete this also]
1858 Remove all references to:
1865 now that we have BFD. All remaining are in machine dependent files.
1869 Re-organize help categories into things that tend to fit on a screen
1874 Add in commands like ADB's for searching for patterns, etc. We should
1875 be able to examine and patch raw unsymboled binaries as well in gdb as
1876 we can in adb. (E.g. increase the timeout in /bin/login without source).
1878 [actually, add ADB interface :-]
1882 When doing "step" or "next", if a few lines of source are skipped between
1883 the previous line and the current one, print those lines, not just the
1884 last line of a multiline statement.
1888 Handling of "&" address-of operator needs some serious overhaul
1889 for ANSI C and consistency on arrays and functions.
1890 For "float point[15];":
1891 ptype &point[4] ==> Attempt to take address of non-lvalue.
1892 For "char *malloc();":
1893 ptype malloc ==> "char *()"; should be same as
1894 ptype &malloc ==> "char *(*)()"
1895 call printf ("%x\n", malloc) ==> weird value, should be same as
1896 call printf ("%x\n", &malloc) ==> correct value
1900 Fix dbxread.c symbol reading in the presence of interrupts. It
1901 currently leaves a cleanup to blow away the entire symbol table when a
1902 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1904 [I suspect that the grype was that, on a slow system, you might want
1905 to cntrl-c and get just half the symbols and then load the rest later
1906 - scary to be honest]
1910 Mipsread.c reads include files depth-first, because the dependencies
1911 in the psymtabs are way too inclusive (it seems to me). Figure out what
1912 really depends on what, to avoid recursing 20 or 30 times while reading
1917 value_add() should be subtracting the lower bound of arrays, if known,
1918 and possibly checking against the upper bound for error reporting.
1922 When listing source lines, check for a preceding \n, to verify that
1923 the file hasn't changed out from under us.
1925 [fixed by some other means I think. That hack wouldn't actually work
1926 reliably - the file might move such that another \n appears. ]
1930 Get all the remote systems (where the protocol allows it) to be able to
1931 stop the remote system when the GDB user types ^C (like remote.c
1932 does). For ebmon, use ^Ak.
1936 Possible feature: A version of the "disassemble" command which shows
1937 both source and assembly code ("set symbol-filename on" is a partial
1940 [has this been done? It was certainly done for MI and GDBtk]
1944 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1945 using a 0 address for bad purposes internally).
1949 Make "info path" and path_command work again (but independent of the
1950 environment either of gdb or that we'll pass to the inferior).
1954 Make GDB understand the GCC feature for putting octal constants in
1955 enums. Make it so overflow on an enum constant does not error_type
1956 the whole type. Allow arbitrarily large enums with type attributes.
1957 Put all this stuff in the testsuite.
1961 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1962 the value in hex; process type attributes). Add this to the
1963 testsuite. This way future compilers can add new types and old
1964 versions of GDB can do something halfway reasonable.
1968 Fix mdebugread.c:parse_type to do fundamental types right (see
1969 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1970 that the debug format fixes the sizes of these things and it shouldn't
1971 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1972 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1973 be aware of that). Also use a switch statement for clarity and speed.
1977 Investigate adding symbols in target_load--some targets do, some
1982 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1983 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1988 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1993 Make a watchpoint which contains a function call an error (it is
1994 broken now, making it work is probably not worth the effort).
1998 New test case based on weird.exp but in which type numbers are not
1999 renumbered (thus multiply defining a type). This currently causes an
2000 infinite loop on "p v_comb".
2004 [Hey! Hint Hint Delete Delete!!!]
2006 Fix 386 floating point so that floating point registers are real
2007 registers (but code can deal at run-time if they are missing, like
2008 mips and 68k). This would clean up "info float" and related stuff.
2012 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
2013 about not being able to access memory location 0.
2015 -------------------- enummask.c
2036 If try to modify value in file with "set write off" should give
2037 appropriate error not "cannot access memory at address 0x65e0".
2041 Allow core file without exec file on RS/6000.
2045 Make sure "shell" with no arguments works right on DOS.
2049 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
2050 the same directory can be NFS-mounted on unix or DOS, and work the
2055 [Is this another delete???]
2057 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
2058 get RS/6000 to work right, might not be immediately relevant).
2062 Work out some kind of way to allow running the inferior to be done as
2063 a sub-execution of, eg. breakpoint command lists. Currently running
2064 the inferior interupts any command list execution. This would require
2065 some rewriting of wait_for_inferior & friends, and hence should
2066 probably be done in concert with the above.
2070 Add function arguments to gdb user defined functions.
2074 Add convenience variables that refer to exec file, symbol file,
2075 selected frame source file, selected frame function, selected frame
2080 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2081 allocate a partial symtab for each BINCL/EINCL grouping. This will
2082 seriously decrease the size of inter-psymtab dependencies and hence
2083 lessen the amount that needs to be read in when a new source file is
2088 Add a command for searching memory, a la adb. It specifies size,
2089 mask, value, start address. ADB searches until it finds it or hits
2090 an error (or is interrupted).
2094 Remove the range and type checking code and documentation, if not