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)
165 Fix copyright notices.
167 Turns out that ``1998-2000'' isn't considered valid :-(
169 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
173 GDB 5.1 - Known Problems
174 ========================
180 The z8k has suffered bit rot and is known to not build. The problem
181 was occuring in the opcodes directory.
187 The m88k has suffered bit rot and is known to not build.
191 Solaris 8 x86 CURSES_H problem
192 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
194 The original problem was worked around with:
196 2000-06-06 Michael Snyder <msnyder@cygnus.com>
198 * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
199 * configure: Regenerate.
201 When building both GDB and SID using the same source tree the problem
202 will still occure. sid/component/configure.in mis-configures
203 <curses.h> and leaves wrong information in the config cache.
212 Thread support. Right now, as soon as a thread finishes and exits,
213 you're hosed. This problem is reported once a week or so.
217 GDB 5.2 - New features
218 ======================
222 GCC 3.0 ABI support (but hopefully sooner...).
226 Objective C/C++ support (but hopefully sooner...).
233 The following cleanups have been identified as part of GDB 5.2.
237 Remove old code that does not use ui_out functions and all the related
238 "ifdef"s. This also allows the elimination of -DUI_OUT from
239 Makefile.in and configure.in.
245 Eliminate all warnings for at least one host/target for the flags:
246 -Wimplicit -Wreturn-type -Wcomment -Wtrigraphs -Wformat -Wparentheses
247 -Wpointer-arith -Wuninitialized
251 Deprecate, if not delete, the following:
257 Replaced by, on the target side
259 and on core-gdb side:
260 {read,write}_register_gen()
261 Remote.c will need to use something
262 other than REGISTER_BYTE() and
263 REGISTER_RAW_SIZE() when unpacking
266 STORE_PSEUDO_REGISTER
267 FETCH_PSEUDO_REGISTER
268 Now handed by the methods
269 gdbarch_{read,write}_register()
270 which sits between core GDB and
274 REGISTER_CONVERT_TO_RAW
275 REGISTER_CONVERT_TO_VIRTUAL
276 I think these three are redundant.
277 gdbarch_register_{read,write} can
278 do any conversion it likes.
280 REGISTER_VIRTUAL_SIZE
281 MAX_REGISTER_VIRTUAL_SIZE
282 REGISTER_VIRTUAL_TYPE
283 I think these can be replaced by
285 FRAME_REGISTER_TYPE(frame, regnum)
286 REGISTER_TYPE(regnum)
290 FRAME_REGISTER_INFO (frame, ...)
293 If nothing else rename this so that
294 how it relates to rawreg and the
298 The size of the cache can be computed
303 Restructure gdb directory tree so that it avoids any 8.3 and 14
308 Convert GDB build process to AUTOMAKE.
310 See also sub-directory configure below.
312 The current convention is (kind of) to use $(<header>_h) in all
313 dependency lists. It isn't done in a consistent way.
317 GDB 5.2 - Known Problems
318 ========================
322 Code Cleanups: General
323 ======================
325 The following are more general cleanups and fixes. They are not tied
326 to any specific release.
330 Investigate changing --target=a29k-amd-udi to a29k-*-coff* and
331 rationalize *.mt files. The got-ya is in remote-eb.c - it has its own
332 custom tty manipulation - it should be using the serial object.
336 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
340 Can't build IRIX -> arm GDB.
341 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
344 > Now I'm building for an embedded arm target. If there is a way of turning
345 > remote-rdi off, I couldn't find it. It looks like it gets built by default
346 > in gdb/configure.tgt(line 58) Anyway, the build dies in
347 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
348 > aren't one of the architectures supported.
352 Problem with weak functions
353 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
355 Dan Nicolaescu writes:
356 > It seems that gdb-4.95.1 does not display correctly the function when
357 > stoping in weak functions.
359 > It stops in a function that is defined as weak, not in the function
360 > that is actually run...
364 Follow through `make check' with --enable-shared.
366 When the srcware tree is configured with --enable-shared, the `expect'
367 program won't run properly. Jim Wilson found out gdb has a local hack
368 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
371 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
375 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
377 Patches in the database.
381 printcmd.c (print_address_numeric):
383 NOTE: This assumes that the significant address information is kept in
384 the least significant bits of ADDR - the upper bits were either zero
385 or sign extended. Should ADDRESS_TO_POINTER() or some
386 ADDRESS_TO_PRINTABLE() be used to do the conversion?
390 The BFD directory requires bug-fixed AUTOMAKE et.al.
392 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
393 contained the full path to texinfo.tex when it should have only
394 contained the directory. The bug has been fixed in the current
395 AUTOMAKE sources. Automake snapshots can be found in:
396 ftp://sourceware.cygnus.com/pub/gdb/snapshots
397 and ftp://sourceware.cygnus.com/pub/binutils
401 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
402 determine the default isa/byte-order.
406 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
407 BIG_ENDIAN and LITTLE_ENDIAN.
411 Eliminate more compiler warnings.
413 Of course there also needs to be the usual debate over which warnings
414 are valid and how to best go about this.
416 One method: choose a single option; get agreement that it is
417 reasonable; try it out to see if there isn't anything silly about it
418 (-Wunused-parameters is an example of that) then incrementally hack
421 The other method is to enable all warnings and eliminate them from one
426 Elimination of ``(catch_errors_ftype *) func''.
428 Like make_cleanup_func it isn't portable.
429 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
430 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
434 Nuke #define CONST_PTR.
442 [PATCH/5] src/intl/Makefile.in:distclean additions
443 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
445 Do not forget to merge the patch back into the trunk.
449 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
451 At present defs.h includes <endian.h> (which is linux specific) yet
452 almost nothing depends on it. Suggest "gdb_endian.h" which can also
453 handle <machine/endian.h> and only include that where it is really
458 Replace savestring() with something from libiberty.
460 An xstrldup()? but that would have different semantics.
464 Rationalize use of floatformat_unknown in GDB sources.
466 Instead of defaulting to floatformat_unknown, should hosts/targets
467 specify the value explicitly?
469 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
473 Add a ``name'' member to include/floatformat.h:struct floatformat.
474 Print that name in gdbarch.c.
478 Sort out the harris mess in include/floatformat.h (it hardwires two
479 different floating point formats).
483 See of the GDB local floatformat_do_doublest() and libiberty's
484 floatformat_to_double (which was once GDB's ...) can be merged some
489 Eliminate mmalloc(), mstrsave() et.al. from GDB.
491 Also eliminate it from defs.h.
495 Eliminate PTR. ISO-C allows ``void *''.
501 GDB should never abort. GDB should either throw ``error ()'' or
502 ``internal_error ()''. Better still GDB should naturally unwind with
507 GDB probably doesn't build on FreeBSD pre 2.2.x
508 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
510 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
515 Deprecate "fg". Apparently ``fg'' is actually continue.
517 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
521 Deprecate current use of ``floatformat_unknown''.
523 Require all targets to explicitly provide their float format instead
524 of defaulting to floatformat unknown. Doing the latter leads to nasty
527 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
531 Rationalize floatformat_to_double() vs floatformat_to_doublest().
533 Looks like GDB migrated floatformat_to_double() to libiberty but then
534 turned around and created a ..._to_doublest() the latter containing
537 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
541 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
543 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
547 The ``maintenance deprecate set endian big'' command doesn't notice
548 that it is deprecating ``set endian'' and not ``set endian big'' (big
549 is implemented using an enum). Is anyone going to notice this?
553 When tab expanding something like ``set arch<tab>'' ignore the
554 deprecated ``set archdebug'' and expand to ``set architecture''.
558 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
559 like from arm-tdep.c.
563 Fix uses of ->function.cfunc = set_function().
565 The command.c code calls sfunc() when a set command. Rather than
566 change it suggest fixing the callback function so that it is more
569 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
571 See also ``Fix implementation of ``target xxx''.'' below.
575 IRIX 3.x support is probably broken.
579 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
580 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
582 Apart from the d30v, are there any sim/common simulators that make use
585 A brief summary of what happened is that sim/common/sim-break.c was
586 created as a good idea. It turned out a better idea was to use
587 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
591 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
592 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
596 Eliminate ``extern'' from C files.
600 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
602 Extreme care is recommeded - perhaps only modify tests that are
603 exercised by the testsuite (as determined using some type of code
608 Replace the file gdb/CONTRIBUTE with a file that is generated from the
609 gdb/doc/*.texinfo directory.
613 New Features and Fixes
614 ======================
616 These are harder than cleanups but easier than work involving
617 fundamental architectural change.
621 Hardware watchpoint problems on x86 OSes, including Linux:
623 1. Delete/disable hardware watchpoints should free hardware debug
625 2. Watch for different values on a viariable with one hardware debug
628 According to Eli Zaretskii <eliz@delorie.com>:
630 These are not GDB/ia32 issues per se: the above features are all
631 implemented in the DJGPP port of GDB and work in v5.0. Every
632 x86-based target should be able to lift the relevant parts of
633 go32-nat.c and use them almost verbatim. You get debug register
634 sharing through reference counts, and the ability to watch large
635 regions (up to 16 bytes) using multiple registers. (The required
636 infrastructure in high-level GDB application code, mostly in
637 breakpoint.c, is also working since v5.0.)
641 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
642 so that you can see how the GDB was created.
646 Add an "info bfd" command that displays supported object formats,
647 similarly to objdump -i.
649 Is there a command already?
653 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
655 This requires internationalization.
661 (gdb) p fwprintf(stdout,L"%S\n", f)
662 No symbol "L" in current context.
666 Cleanup configury support for optional sub-directories.
668 Check how GCC handles multiple front ends for an example of how things
669 could work. A tentative first step is to rationalize things so that
670 all sub directories are handled in a fashion similar to gdb/mi.
672 See also automake above.
676 Add a transcript mechanism to GDB.
678 Such a mechanism might log all gdb input and output to a file in a
679 form that would allow it to be replayed. It could involve ``gdb
680 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
684 Can the xdep files be replaced by autoconf?
688 Document trace machinery
692 Document ui-out and ui-file.
694 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
698 Update texinfo.tex to latest?
702 Incorporate agentexpr.texi into gdb.texinfo
704 agentexpr.texi mostly describes the details of the byte code used for
705 tracepoints, not the internals of the support for this in GDB. So it
706 looks like gdb.texinfo is a better place for this information.
708 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
712 Document overlay machinery.
716 ``(gdb) catch signal SIGNAL''
718 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
719 You can attach commands to a catch but not a handle. A handle has a
720 limited number of hardwired actions.
726 o readline/*.h bitrot
728 The TUI isn't up-to-date with
729 respect to the readline currently
730 bundled with GDB. Importing a
731 new readline is on the 5.1 wish
732 list so this can only get worse.
734 Grep for things like term_cursor_move.
736 (To be honest, I don't see anyone
737 importing a new readline before 5.1 is
740 o tui.c:va_catch_errors() bitrot
742 This nasty piece of work used knowledge
743 of the internals of GDBs error functions :-(
744 Ever since those internals were cleaned
745 up this code has been broken. :-(
747 o tuiWin.c:c_makeVisibleWithNewHeight() broken
748 tuiLayout.c:_extractDisplayStartAddr() broken
750 Both these function call find_line_pc()
751 incorrectly (wrong args, wrong return value).
753 I suspect this bug has always been there!
754 It had been hidden because those files
755 didn't include the necessary header files
758 o tuiRegs() host dependant
760 Not suprisingly, this isn't a very portable
761 section of code. However, I'm sure people
762 could live with no regs in the short to
765 o defs.h: #include "tui.h" et.al.
767 I'm not sure where this came from.
768 It was a really bad idea.
770 To get things to compile I did a nasty
771 hack (Just declare what was needed and
772 replace any expressions like xx->y.z()
773 in GDB proper with function calls). I
774 could commit it slightly cleaned up if
777 Medium Term. the #ifdef TUI and TuiDo()
778 should be changed to hooks (like GDBTK).
779 The gdb-events.[hc] is there for that
782 o tui.c:_tuiReset() host dependant
784 tui.c contains a lump of termio[s]
785 I suspect an equivalent block of
786 code can be lifted from readline.
787 An equivalent readline function may
790 o curses.h vs ncurses.h.
792 Simple portability problem.
796 This function is a mystery - where is it?
798 o tui-file.[hc] cleanup
800 This can be significantly simplified.
802 o The code should be pacified. (-Werror -W...)
804 There are plenty of #includes,
805 duplicate #includes, missing function decls
808 Some of the problems I found were through
809 fixing a few of the warnings.
811 o The code should be GNUtified.
813 It would be very nice to have this code
814 look like the rest of GDB. That way people
815 would be more accepting of it as a true
818 Until it is GNUtified it is going to stick
819 out like a sore thumb to the programmer.
821 o The code should be clearly copyrighted
823 (FSF, with due credit to HP)
827 Add support for ``gdb --- PROGRAM ARGS ...''.
828 Add support for ``gdb -cmd=...''
830 Along with many variations. Check:
832 ????? for a full discussion.
838 Implement ``(gdb) !ls''.
840 Which is very different from ``(gdb) ! ls''. Implementing the latter
843 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
847 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
852 When tab expanding something like ``set arch<tab>'' ignore the
853 deprecated ``set archdebug'' and expand to ``set architecture''.
857 Replace the code that uses the host FPU with an emulator of the target
862 The "ocd reset" command needs to flush the dcache, which requires breaking
863 the abstraction layer between the target independent and target code. One
864 way to address this is provide a generic "reset" command and target vector.
866 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
875 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
876 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
878 The thread_db assisted debugging code doesn't handle exiting threads
879 properly, at least in combination with glibc 2.1.3 (the framework is
880 there, just not the actual code). There are at least two problems
881 that prevent this from working.
883 As an additional reference point, the pre thread_db code did not work
888 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
890 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
892 Christopher Blizzard writes:
894 So, I've done some more digging into this and it looks like Jim
895 Kingdon has reported this problem in the past:
897 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
899 I can reproduce this problem both with and without Tom's patch. Has
900 anyone seen this before? Maybe have a solution for it hanging around?
903 There's a test case for this documented at:
905 when debugging threaded applications you get extra SIGTRAPs
906 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
908 [There should be a GDB testcase - cagney]
912 GDB5 TOT on unixware 7
913 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
916 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
917 > practical matter, the current thread support is somewhat more annoying
918 > than when GDB was thread-unaware.
925 New languages come onto the scene all the time.
929 Re: Various C++ things
931 value_headof/value_from_vtable_info are worthless, and should be
932 removed. The one place in printcmd.c that uses it should use the RTTI
935 RTTI for g++ should be using the typeinfo functions rather than the
936 vtables. The typeinfo functions are always at offset 4 from the
937 beginning of the vtable, and are always right. The vtables will have
938 weird names like E::VB sometimes. The typeinfo function will always
939 be "E type_info function", or somesuch.
941 value_virtual_fn_field needs to be fixed so there are no failures for
942 virtual functions for C++ using g++.
944 Testsuite cases are the major priority right now for C++ support,
945 since i have to make a lot of changes that could potentially break
950 Add support for Modula3
952 Get DEC/Compaq to contribute their Modula-3 support.
956 Remote Protocol Support
957 =======================
961 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
962 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
964 (Broken) support for GDB's remote protocol across UDP is to be
965 included in the follow-on release.
967 It should be noted that UDP can only work when the [Gg] packet fits in
970 There is also much debate over the merit of this.
974 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
976 Add support for packet enable/disable commands with these thread
977 packets. General cleanup.
979 [PATCH] Document the ThreadInfo remote protocol queries
980 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
982 [PATCH] "info threads" queries for remote.c
983 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
987 Remote protocol doco feedback.
989 Too much feedback to mention needs to be merged in (901660). Search
990 for the word ``remote''.
993 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
994 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
995 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
999 GDB doesn't recover gracefully from remote protocol errors.
1001 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
1002 ignored and a timeout is required before GDB retries. A pre-cursor to
1003 fixing this this is making GDB's remote protocol packet more robust.
1005 While downloading to a remote protocol target, gdb ignores packet
1006 errors in so far as it will continue to download with chunk N+1 even
1007 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
1008 take a painfully long time to run. As a PS that test needs to be
1009 fixed so that it builds on 16 bit machines.
1013 Fix the ``!'' packet.
1015 JT reported that the existing targets do, in fact return ``OK'' so it
1016 is possible to merge remote and extended-remote targets.
1020 Drop ``<address>'' from the [SsCc] packets.
1022 I don't think that GDB generates them so having it in the protocol is
1027 Fix doco on the ``q'' packet.
1029 It has evolved into a generic RPC. The notes should reflect this and,
1030 perhaps, the ``Q'' packet can be deprecated.
1032 The doco should mention that ``OK'' is a valid packet response.
1034 The doco should explain why ``OK'' needs to be a valid packet
1039 Add the cycle step command.
1041 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1045 Resolve how to scale things to support very large packets.
1049 Resolve how to handle a target that changes things like its endianess
1050 on the fly - should it be returned in the ``T'' packet?
1052 Underlying problem is that the register file is target endian. If the
1053 target endianess changes gdb doesn't know.
1060 If / when GDB starts to support the debugging of multi-processor
1061 (rather than multi-thread) applications the symtab code will need to
1062 be updated a little so that several independent symbol tables are
1063 active at a given time.
1065 The other interesting change is a clarification of the exact meaning
1066 of CORE_ADDR and that has had consequences for a few targets (that
1067 were abusing that data type).
1071 Investiagate ways of reducing memory.
1075 Investigate ways of improving load time.
1079 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1081 Consequence of recent symtab clarification. No marks for figuring out
1082 who maintains the d10v.
1086 Get the MIPS to correctly sign extend all address <-> pointer
1089 Consequence of recent symtab clarification. No marks for figuring out
1090 who maintains the MIPS.
1094 GDB truncates 64 bit enums.
1096 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1103 There are never to many testcases.
1107 Better thread testsuite.
1111 Better C++ testsuite.
1115 Look at adding a GDB specific testsuite directory so that white box
1116 tests of key internals can be added (eg ui_file).
1120 Separate out tests that involve the floating point (FP).
1122 (Something for people brining up new targets). FP and non-fp tests
1123 are combined. I think there should be set of basic tests that
1124 exercise pure integer support and then a more expanded set that
1125 exercise FP and FP/integer interactions.
1127 As an example, the MIPS, for n32 as problems with passing FP's and
1128 structs. Since most inferior call tests include FP it is difficult to
1129 determine of the integer tests are ok.
1133 Architectural Changes: General
1134 ==============================
1136 These are harder than simple cleanups / fixes and, consequently
1137 involve more work. Typically an Architectural Change will be broken
1138 down into a more digestible set of cleanups and fixes.
1142 Cleanup software single step.
1144 At present many targets implement software single step by directly
1145 blatting memory (see rs6000-tdep.c). Those targets should register
1146 the applicable breakpoints using the breakpoint framework. Perhaphs a
1147 new internal breakpoint class ``step'' is needed.
1151 Replace READ_FP() with FRAME_HANDLE().
1153 READ_FP() is a hangover from the days of the vax when the ABI really
1154 did have a frame pointer register. Modern architectures typically
1155 construct a virtual frame-handle from the stack pointer and various
1156 other bits of string.
1158 Unfortunately GDB still treats this synthetic FP register as though it
1159 is real. That in turn really confuses users (arm and ``print $fp'' VS
1160 ``info registers fp''). The synthetic FP should be separated out of
1161 the true register set presented to the user.
1165 Register Cache Cleanup (below from Andrew Cagney)
1167 I would depict the current register architecture as something like:
1169 High GDB --> Low GDB
1174 register + REGISTER_BYTE(reg_nr)
1177 -------------------------
1178 | extern register[] |
1179 -------------------------
1181 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1182 really clear on what mechanisms they should be using to manipulate that
1183 buffer. Further, much code assumes, dangerously, that registers are
1184 contigious. Having got mips-tdep.c to support multiple ABIs, believe
1185 me, that is a bad assumption. Finally, that register cache layout is
1186 determined by the current remote/local target and _not_ the less
1187 specific target ISA. In fact, in many cases it is determined by the
1188 somewhat arbitrary layout of the [gG] packets!
1191 How I would like the register file to work is more like:
1219 The main objectives being:
1221 o a clear separation between the low
1222 level target and the high level GDB
1224 o a mechanism that solves the general
1225 problem of register aliases, overlaps
1226 etc instead of treating them as optional
1227 extras that can be wedged in as an after
1228 thought (that is a reasonable description
1229 of the current code).
1231 Identify then solve the hard case and the
1232 rest just falls out. GDB solved the easy
1233 case and then tried to ignore the real
1236 o a removal of the assumption that the
1237 mapping between the register cache
1238 and virtual registers is largely static.
1239 If you flip the USR/SSR stack register
1240 select bit in the status-register then
1241 the corresponding stack registers should
1244 o a mechanism that clearly separates the
1245 gdb internal register cache from any
1246 target (not architecture) dependent
1247 specifics such as [gG] packets.
1249 Of course, like anything, it sounds good in theory. In reality, it
1250 would have to contend with many<->many relationships at both the
1251 virt<->cache and cache<->target level. For instance:
1254 Modifying an mmx register may involve
1255 scattering values across both FP and
1256 mmpx specific parts of a buffer
1259 When writing back a SP it may need to
1260 both be written to both SP and USP.
1265 Rather than let this like the last time it was discussed, just slip, I'm
1266 first going to add this e-mail (+ references) to TODO. I'd then like to
1267 sketch out a broad strategy I think could get us there.
1270 First thing I'd suggest is separating out the ``extern registers[]''
1271 code so that we can at least identify what is using it. At present
1272 things are scattered across many files. That way we can at least
1273 pretend that there is a cache instead of a global array :-)
1275 I'd then suggest someone putting up a proposal for the pseudo-reg /
1276 high-level side interface so that code can be adopted to it. For old
1277 code, initially a blanket rename of write_register_bytes() to
1278 deprecated_write_register_bytes() would help.
1280 Following that would, finaly be the corresponding changes to the target.
1284 Check that GDB can handle all BFD architectures (Andrew Cagney)
1286 There should be a test that checks that BFD/GDB are in sync with
1287 regard to architecture changes. Something like a test that first
1288 queries GDB for all supported architectures and then feeds each back
1289 to GDB.. Anyone interested in learning how to write tests? :-)
1293 Architectural Change: Multi-arch et al.
1294 =======================================
1296 The long term objective is to remove all assumptions that there is a
1297 single target with a single address space with a single instruction
1298 set architecture and single application binary interface.
1300 This is an ongoing effort. The first milestone is to enable
1301 ``multi-arch'' where by all architectural decisions are made at
1304 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1305 ``gdbisa''. Once things are multi-arched breaking that down correctly
1306 will become much easier.
1310 GDBARCH cleanup (Andrew Cagney)
1312 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1313 into arch-utils.[hc].
1315 Document that gdbarch_init_ftype could easily fail because it didn't
1316 identify an architecture.
1320 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
1322 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1327 Fix target_signal_from_host() etc.
1329 The name is wrong for starters. ``target_signal'' should probably be
1330 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
1331 After that it needs to be multi-arched and made independent of any
1332 host signal numbering.
1336 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1339 This is a barrier to replacing mips_extra_func_info with something
1340 that works with multi-arch.
1344 Multi-arch mips_extra_func_info.
1346 This first needs the alpha to be updated so that it uses ``struct
1351 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1353 Surely one of them is redundant.
1357 Convert ALL architectures to MULTI-ARCH.
1361 Select the initial multi-arch ISA / ABI based on --target or similar.
1363 At present the default is based on what ever is first in the BFD
1364 archures table. It should be determined based on the ``--target=...''
1369 Make MIPS pure multi-arch.
1371 It is only at the multi-arch enabled stage.
1377 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1379 Can the tm.h and nm.h files be eliminated by multi-arch.
1383 Architectural Change: MI, LIBGDB and scripting languages
1384 ========================================================
1386 See also architectural changes related to the event loop. LIBGDB
1387 can't be finished until there is a generic event loop being used by
1390 The long term objective is it to be possible to integrate GDB into
1391 scripting languages.
1395 Implement generic ``(gdb) commmand > file''
1397 Once everything is going through ui_file it should be come fairly
1400 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1404 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1406 gdb_stdtarg is easily confused with gdb_stdarg.
1410 Extra ui_file methods - dump.
1412 Very useful for whitebox testing.
1416 Eliminate error_begin().
1418 With ui_file, there is no need for the statefull error_begin ()
1423 Send normal output to gdb_stdout.
1424 Send error messages to gdb_stderror.
1425 Send debug and log output log gdb_stdlog.
1427 GDB still contains many cases where (f)printf or printf_filtered () is
1428 used when it should be sending the messages to gdb_stderror or
1429 gdb_stdlog. The thought of #defining printf to something has crossed
1434 Re-do GDB's output pager.
1436 GDB's output pager still relies on people correctly using *_filtered
1437 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1438 Hopefully, with all normal output going to gdb_stdout, the pager can
1439 just look at the ui_file that the output is on and then use that to
1440 decide what to do about paging. Sounds good in theory.
1444 Check/cleanup MI documentation.
1446 The list of commands specified in the documentation needs to be
1447 checked against the mi-cmds.c table in a mechanical way (so that they
1448 two can be kept up-to-date).
1452 Convert MI into libgdb
1454 MI provides a text interface into what should be many of the libgdb
1455 functions. The implementation of those functions should be separated
1456 into the MI interface and the functions proper. Those functions being
1457 moved to gdb/lib say.
1463 The first part can already be found in defs.h.
1467 MI's input does not use buffering.
1469 At present the MI interface reads raw characters of from an unbuffered
1470 FD. This is to avoid several nasty buffer/race conditions. That code
1471 should be changed so that it registers its self with the event loop
1472 (on the input FD) and then push commands up to MI as they arrive.
1474 The serial code already does this.
1478 Make MI interface accessible from existing CLI.
1482 Add a breakpoint-edit command to MI.
1484 It would be similar to MI's breakpoint create but would apply to an
1485 existing breakpoint. It saves the need to delete/create breakpoints
1486 when ever they are changed.
1490 Add directory path to MI breakpoint.
1492 That way the GUI's task of finding the file within which the
1493 breakpoint was set is simplified.
1497 Add a mechanism to reject certain expression classes to MI
1499 There are situtations where you don't want GDB's expression
1500 parser/evaluator to perform inferior function calls or variable
1501 assignments. A way of restricting the expression parser so that such
1502 operations are not accepted would be very helpful.
1506 Remove sideffects from libgdb breakpoint create function.
1508 The user can use the CLI to create a breakpoint with partial
1509 information - no file (gdb would use the file from the last
1512 The libgdb interface currently affects that environment which can lead
1513 to confusion when a user is setting breakpoints via both the MI and
1516 This is also a good example of how getting the CLI ``right'' will be
1521 Move gdb_lasterr to ui_out?
1523 The way GDB throws errors and records them needs a re-think. ui_out
1524 handles the correct output well. It doesn't resolve what to do with
1525 output / error-messages when things go wrong.
1529 do_setshow_command contains a 1024 byte buffer.
1531 The function assumes that there will never be any more than 1024 bytes
1532 of enum. It should use mem_file.
1536 Should struct cmd_list_element . completer take the command as an
1541 Should the bulk of top.c:line_completion_function() be moved to
1542 command.[hc]? complete_on_cmdlist() and complete_on_enums() could
1543 then be made private.
1547 top.c (execute_command): Should a command being valid when the target
1548 is running be made an attribute (predicate) to the command rather than
1549 an explicit set of tests.
1553 top.c (execute_command): Should the bulk of this function be moved
1554 into command.[hc] so that top.c doesn't grub around in the command
1559 Architectural Change: Async
1560 ===========================
1562 While GDB uses an event loop when prompting the user for input. That
1563 event loop is not exploited by targets when they allow the target
1564 program to continue. Typically targets still block in (target_wait())
1565 until the program again halts.
1567 The closest a target comes to supporting full asynchronous mode are
1568 the remote targets ``async'' and ``extended-async''.
1572 Asynchronous expression evaluator
1574 Inferior function calls hang GDB.
1578 Fix implementation of ``target xxx''.
1580 At present when the user specifies ``target xxxx'', the CLI maps that
1581 directly onto a target open method. It is then assumed that the
1582 target open method should do all sorts of complicated things as this
1583 is the only chance it has. Check how the various remote targets
1584 duplicate the target operations. Check also how the various targets
1585 behave differently for purely arbitrary reasons.
1587 What should happen is that ``target xxxx'' should call a generic
1588 ``target'' function and that should then co-ordinate the opening of
1589 ``xxxx''. This becomes especially important when you're trying to
1590 open an asynchronous target that may need to perform background tasks
1591 as part of the ``attach'' phase.
1593 Unfortunately, due to limitations in the old/creaking command.h
1594 interface, that isn't possible. The function being called isn't told
1595 of the ``xxx'' or any other context information.
1597 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1598 CLI code so that it passes to the callback function (attatched to a
1599 command) useful information such as the actual command and a context
1600 for that command. Other changes such as making ``struct command''
1601 opaque may also help.
1604 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1608 Make "target xxx" command interruptible.
1610 As things become async this becomes possible. A target would start
1611 the connect and then return control to the event loop. A cntrl-c
1612 would notify the target that the operation is to be abandoned and the
1613 target code could respond.
1617 Add a "suspend" subcommand of the "continue" command to suspend gdb
1618 while continuing execution of the subprocess. Useful when you are
1619 debugging servers and you want to dodge out and initiate a connection
1620 to a server running under gdb.
1629 Frequently requested but not approved requests.
1633 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1635 The benefits on this one are thought to be marginal - GDBs design
1636 means that unused parameters are very common. GCC 3.0 will also
1637 include the option -Wno-unused-parameter which means that ``-Wall
1638 -Wno-unused-parameters -Werror'' can be specified.
1647 This list is not up to date, and opinions vary about the importance or
1648 even desirability of some of the items. If you do fix something, it
1649 always pays to check the below.
1653 @c This does not work (yet if ever). FIXME.
1654 @c @item --parse=@var{lang} @dots{}
1655 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1656 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1657 @c list of all supported languages, omit the argument. Without this
1658 @c option, @value{GDBN} is configured to parse all supported languages.
1662 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1663 is its default value. Clean this up.
1667 It should be possible to use symbols from shared libraries before we know
1668 exactly where the libraries will be loaded. E.g. "b perror" before running
1669 the program. This could maybe be done as an extension of the "breakpoint
1670 re-evaluation" after new symbols are loaded.
1674 Make single_step() insert and remove breakpoints in one operation.
1676 [If this is talking about having single_step() insert the breakpoints,
1677 run the target then pull the breakpoints then it is wrong. The
1678 function has to return as control has to eventually be passed back to
1679 the main event loop.]
1683 Speed up single stepping by avoiding extraneous ptrace calls.
1687 Speed up single stepping by not inserting and removing breakpoints
1688 each time the inferior starts and stops.
1690 Breakpoints should not be inserted and deleted all the time. Only the
1691 one(s) there should be removed when we have to step over one. Support
1692 breakpoints that don't have to be removed to step over them.
1694 [this has resulted in numerous debates. The issue isn't clear cut]
1698 Provide "voodoo" debugging of core files. This creates a zombie
1699 process as a child of the debugger, and loads it up with the data,
1700 stack, and regs of the core file. This allows you to call functions
1701 in the executable, to manipulate the data in the core file.
1707 GDB reopens the source file on every line, as you "next" through it.
1709 [still true? I've a memory of this being fixed]
1713 Perhaps "i source" should take an argument like that of "list".
1717 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1718 it matches the source line indicated.
1722 The prompt at end of screen should accept space as well as CR.
1726 Backtrace should point out what the currently selected frame is, in
1727 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1728 ...)" rather than "#3 foo (bar, ...)".
1732 "i program" should work for core files, and display more info, like what
1733 actually caused it to die.
1737 "x/10i" should shorten the long name, if any, on subsequent lines.
1741 "next" over a function that longjumps, never stops until next time you happen
1742 to get to that spot by accident. E.g. "n" over execute_command which has
1747 "set zeroprint off", don't bother printing members of structs which
1748 are entirely zero. Useful for those big structs with few useful
1753 GDB does four ioctl's for every command, probably switching terminal modes
1754 to/from inferior or for readline or something.
1758 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1759 if the state is the same, too.
1763 "i frame" shows wrong "arglist at" location, doesn't show where the args
1764 should be found, only their actual values.
1768 There should be a way for "set" commands to validate the new setting
1769 before it takes effect.
1773 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1777 i line VAR produces "Line number not known for symbol ``var''.". I
1778 thought we were stashing that info now!
1782 We should be able to write to random files at hex offsets like adb.
1786 [elena - delete this]
1788 Handle add_file with separate text, data, and bss addresses. Maybe
1789 handle separate addresses for each segment in the object file?
1793 [Jimb/Elena delete this one]
1795 Handle free_named_symtab to cope with multiply-loaded object files
1796 in a dynamic linking environment. Should remember the last copy loaded,
1797 but not get too snowed if it finds references to the older copy.
1801 [elena delete this also]
1803 Remove all references to:
1810 now that we have BFD. All remaining are in machine dependent files.
1814 Re-organize help categories into things that tend to fit on a screen
1819 Add in commands like ADB's for searching for patterns, etc. We should
1820 be able to examine and patch raw unsymboled binaries as well in gdb as
1821 we can in adb. (E.g. increase the timeout in /bin/login without source).
1823 [actually, add ADB interface :-]
1827 When doing "step" or "next", if a few lines of source are skipped between
1828 the previous line and the current one, print those lines, not just the
1829 last line of a multiline statement.
1833 Handling of "&" address-of operator needs some serious overhaul
1834 for ANSI C and consistency on arrays and functions.
1835 For "float point[15];":
1836 ptype &point[4] ==> Attempt to take address of non-lvalue.
1837 For "char *malloc();":
1838 ptype malloc ==> "char *()"; should be same as
1839 ptype &malloc ==> "char *(*)()"
1840 call printf ("%x\n", malloc) ==> weird value, should be same as
1841 call printf ("%x\n", &malloc) ==> correct value
1845 Fix dbxread.c symbol reading in the presence of interrupts. It
1846 currently leaves a cleanup to blow away the entire symbol table when a
1847 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1849 [I suspect that the grype was that, on a slow system, you might want
1850 to cntrl-c and get just half the symbols and then load the rest later
1851 - scary to be honest]
1855 Mipsread.c reads include files depth-first, because the dependencies
1856 in the psymtabs are way too inclusive (it seems to me). Figure out what
1857 really depends on what, to avoid recursing 20 or 30 times while reading
1862 value_add() should be subtracting the lower bound of arrays, if known,
1863 and possibly checking against the upper bound for error reporting.
1867 When listing source lines, check for a preceding \n, to verify that
1868 the file hasn't changed out from under us.
1870 [fixed by some other means I think. That hack wouldn't actually work
1871 reliably - the file might move such that another \n appears. ]
1875 Get all the remote systems (where the protocol allows it) to be able to
1876 stop the remote system when the GDB user types ^C (like remote.c
1877 does). For ebmon, use ^Ak.
1881 Possible feature: A version of the "disassemble" command which shows
1882 both source and assembly code ("set symbol-filename on" is a partial
1885 [has this been done? It was certainly done for MI and GDBtk]
1889 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1890 using a 0 address for bad purposes internally).
1894 Make "info path" and path_command work again (but independent of the
1895 environment either of gdb or that we'll pass to the inferior).
1899 Make GDB understand the GCC feature for putting octal constants in
1900 enums. Make it so overflow on an enum constant does not error_type
1901 the whole type. Allow arbitrarily large enums with type attributes.
1902 Put all this stuff in the testsuite.
1906 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1907 the value in hex; process type attributes). Add this to the
1908 testsuite. This way future compilers can add new types and old
1909 versions of GDB can do something halfway reasonable.
1913 Fix mdebugread.c:parse_type to do fundamental types right (see
1914 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1915 that the debug format fixes the sizes of these things and it shouldn't
1916 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1917 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1918 be aware of that). Also use a switch statement for clarity and speed.
1922 Investigate adding symbols in target_load--some targets do, some
1927 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1928 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1933 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1938 Make a watchpoint which contains a function call an error (it is
1939 broken now, making it work is probably not worth the effort).
1943 New test case based on weird.exp but in which type numbers are not
1944 renumbered (thus multiply defining a type). This currently causes an
1945 infinite loop on "p v_comb".
1949 [Hey! Hint Hint Delete Delete!!!]
1951 Fix 386 floating point so that floating point registers are real
1952 registers (but code can deal at run-time if they are missing, like
1953 mips and 68k). This would clean up "info float" and related stuff.
1957 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1958 about not being able to access memory location 0.
1960 -------------------- enummask.c
1981 If try to modify value in file with "set write off" should give
1982 appropriate error not "cannot access memory at address 0x65e0".
1986 Allow core file without exec file on RS/6000.
1990 Make sure "shell" with no arguments works right on DOS.
1994 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1995 the same directory can be NFS-mounted on unix or DOS, and work the
2000 [Is this another delete???]
2002 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
2003 get RS/6000 to work right, might not be immediately relevant).
2007 Work out some kind of way to allow running the inferior to be done as
2008 a sub-execution of, eg. breakpoint command lists. Currently running
2009 the inferior interupts any command list execution. This would require
2010 some rewriting of wait_for_inferior & friends, and hence should
2011 probably be done in concert with the above.
2015 Add function arguments to gdb user defined functions.
2019 Add convenience variables that refer to exec file, symbol file,
2020 selected frame source file, selected frame function, selected frame
2025 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2026 allocate a partial symtab for each BINCL/EINCL grouping. This will
2027 seriously decrease the size of inter-psymtab dependencies and hence
2028 lessen the amount that needs to be read in when a new source file is
2033 Add a command for searching memory, a la adb. It specifies size,
2034 mask, value, start address. ADB searches until it finds it or hits
2035 an error (or is interrupted).
2039 Remove the range and type checking code and documentation, if not