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 Pascal (Pierre Muller, David Taylor)
94 Pierre Muller has contributed patches for adding Pascal Language
97 2 pascal language patches inserted in database
98 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
101 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
103 [I think this has been merged, need to confirm - cagney]
107 Java (Anthony Green, David Taylor)
109 Anthony Green has a number of Java patches that did not make it into
110 the 5.0 release. The first two are in cvs now, but the third needs
111 some fixing up before it can go in.
114 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
117 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
119 Patch: handle N_MAIN stab
120 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
126 A predicate to this is the multi-arching of SOFTWARE_SINGLE_STEP(). A
127 patch has been submitted.
134 The following code cleanups will hopefully be applied to GDB 5.1.
138 Resolve the build status of all broken targets as identified by the
141 o arm-* vs NetBSD's lack of ``unix''
142 o arm-* vs IRIX (see below)
146 Fix copyright notices.
148 Turns out that ``1998-2000'' isn't considered valid :-(
150 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
154 GDB 5.1 - Known Problems
155 ========================
161 The z8k has suffered bit rot and is known to not build. The problem
162 was occuring in the opcodes directory.
168 The m88k has suffered bit rot and is known to not build.
172 Solaris 8 x86 CURSES_H problem
173 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
175 The original problem was worked around with:
177 2000-06-06 Michael Snyder <msnyder@cygnus.com>
179 * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
180 * configure: Regenerate.
182 When building both GDB and SID using the same source tree the problem
183 will still occure. sid/component/configure.in mis-configures
184 <curses.h> and leaves wrong information in the config cache.
193 Thread support. Right now, as soon as a thread finishes and exits,
194 you're hosed. This problem is reported once a week or so.
198 GDB 5.2 - New features
199 ======================
203 GCC 3.0 ABI support (but hopefully sooner...).
207 Objective C/C++ support (but hopefully sooner...).
214 The following cleanups have been identified as part of GDB 5.2.
218 Remove old code that does not use ui_out functions and all the related
219 "ifdef"s. This also allows the elimination of -DUI_OUT from
220 Makefile.in and configure.in.
226 Eliminate warnings for all targets on at least one host for one of the
227 -W flags. Flags up for debate include: -Wswitch -Wcomment -trigraphs
228 -Wtrigraphs -Wunused-function -Wunused-label -Wunused-variable
229 -Wunused-value -Wchar-subscripts -Wtraditional -Wshadow -Wcast-qual
230 -Wcast-align -Wwrite-strings -Wconversion -Wstrict-prototypes
231 -Wmissing-prototypes -Wmissing-declarations -Wredundant-decls
232 -Woverloaded-virtual -Winline
236 Deprecate, if not delete, the following:
242 Replaced by, on the target side
244 and on core-gdb side:
245 {read,write}_register_gen()
246 Remote.c will need to use something
247 other than REGISTER_BYTE() and
248 REGISTER_RAW_SIZE() when unpacking
251 STORE_PSEUDO_REGISTER
252 FETCH_PSEUDO_REGISTER
253 Now handed by the methods
254 gdbarch_{read,write}_register()
255 which sits between core GDB and
259 REGISTER_CONVERT_TO_RAW
260 REGISTER_CONVERT_TO_VIRTUAL
261 I think these three are redundant.
262 gdbarch_register_{read,write} can
263 do any conversion it likes.
265 REGISTER_VIRTUAL_SIZE
266 MAX_REGISTER_VIRTUAL_SIZE
267 REGISTER_VIRTUAL_TYPE
268 I think these can be replaced by
270 FRAME_REGISTER_TYPE(frame, regnum)
271 REGISTER_TYPE(regnum)
275 FRAME_REGISTER_INFO (frame, ...)
278 If nothing else rename this so that
279 how it relates to rawreg and the
283 The size of the cache can be computed
286 IS_TRAPPED_INTERNALVAR
287 The pseudo registers should eventually make
292 Obsolete the targets.
298 Obsolete the protocols:
302 ``As of version 5.3, WindRiver has removed the RDB server (RDB
303 protocol support is built into gdb).'' -- Till.
307 Restructure gdb directory tree so that it avoids any 8.3 and 14
312 Convert GDB build process to AUTOMAKE.
314 See also sub-directory configure below.
316 The current convention is (kind of) to use $(<header>_h) in all
317 dependency lists. It isn't done in a consistent way.
321 GDB 5.2 - Known Problems
322 ========================
326 Code Cleanups: General
327 ======================
329 The following are more general cleanups and fixes. They are not tied
330 to any specific release.
334 Investigate changing --target=a29k-amd-udi to a29k-*-coff* and
335 rationalize *.mt files. The got-ya is in remote-eb.c - it has its own
336 custom tty manipulation - it should be using the serial object.
340 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
344 Can't build IRIX -> arm GDB.
345 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
348 > Now I'm building for an embedded arm target. If there is a way of turning
349 > remote-rdi off, I couldn't find it. It looks like it gets built by default
350 > in gdb/configure.tgt(line 58) Anyway, the build dies in
351 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
352 > aren't one of the architectures supported.
356 Problem with weak functions
357 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
359 Dan Nicolaescu writes:
360 > It seems that gdb-4.95.1 does not display correctly the function when
361 > stoping in weak functions.
363 > It stops in a function that is defined as weak, not in the function
364 > that is actually run...
368 Follow through `make check' with --enable-shared.
370 When the srcware tree is configured with --enable-shared, the `expect'
371 program won't run properly. Jim Wilson found out gdb has a local hack
372 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
375 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
379 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
381 Patches in the database.
385 printcmd.c (print_address_numeric):
387 NOTE: This assumes that the significant address information is kept in
388 the least significant bits of ADDR - the upper bits were either zero
389 or sign extended. Should ADDRESS_TO_POINTER() or some
390 ADDRESS_TO_PRINTABLE() be used to do the conversion?
394 The BFD directory requires bug-fixed AUTOMAKE et.al.
396 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
397 contained the full path to texinfo.tex when it should have only
398 contained the directory. The bug has been fixed in the current
399 AUTOMAKE sources. Automake snapshots can be found in:
400 ftp://sourceware.cygnus.com/pub/gdb/snapshots
401 and ftp://sourceware.cygnus.com/pub/binutils
405 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
406 determine the default isa/byte-order.
410 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
411 BIG_ENDIAN and LITTLE_ENDIAN.
415 Eliminate more compiler warnings.
417 Of course there also needs to be the usual debate over which warnings
418 are valid and how to best go about this.
420 One method: choose a single option; get agreement that it is
421 reasonable; try it out to see if there isn't anything silly about it
422 (-Wunused-parameters is an example of that) then incrementally hack
425 The other method is to enable all warnings and eliminate them from one
430 Elimination of ``(catch_errors_ftype *) func''.
432 Like make_cleanup_func it isn't portable.
433 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
434 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
438 Nuke #define CONST_PTR.
446 [PATCH/5] src/intl/Makefile.in:distclean additions
447 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
449 Do not forget to merge the patch back into the trunk.
453 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
455 At present defs.h includes <endian.h> (which is linux specific) yet
456 almost nothing depends on it. Suggest "gdb_endian.h" which can also
457 handle <machine/endian.h> and only include that where it is really
462 Replace savestring() with something from libiberty.
464 An xstrldup()? but that would have different semantics.
468 Rationalize use of floatformat_unknown in GDB sources.
470 Instead of defaulting to floatformat_unknown, should hosts/targets
471 specify the value explicitly?
473 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
477 Add a ``name'' member to include/floatformat.h:struct floatformat.
478 Print that name in gdbarch.c.
482 Sort out the harris mess in include/floatformat.h (it hardwires two
483 different floating point formats).
487 See of the GDB local floatformat_do_doublest() and libiberty's
488 floatformat_to_double (which was once GDB's ...) can be merged some
493 Eliminate mmalloc(), mstrsave() et.al. from GDB.
495 Also eliminate it from defs.h.
499 Eliminate PTR. ISO-C allows ``void *''.
505 GDB should never abort. GDB should either throw ``error ()'' or
506 ``internal_error ()''. Better still GDB should naturally unwind with
511 GDB probably doesn't build on FreeBSD pre 2.2.x
512 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
514 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
519 Deprecate "fg". Apparently ``fg'' is actually continue.
521 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
525 Deprecate current use of ``floatformat_unknown''.
527 Require all targets to explicitly provide their float format instead
528 of defaulting to floatformat unknown. Doing the latter leads to nasty
531 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
535 Rationalize floatformat_to_double() vs floatformat_to_doublest().
537 Looks like GDB migrated floatformat_to_double() to libiberty but then
538 turned around and created a ..._to_doublest() the latter containing
541 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
545 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
547 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
551 The ``maintenance deprecate set endian big'' command doesn't notice
552 that it is deprecating ``set endian'' and not ``set endian big'' (big
553 is implemented using an enum). Is anyone going to notice this?
557 When tab expanding something like ``set arch<tab>'' ignore the
558 deprecated ``set archdebug'' and expand to ``set architecture''.
562 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
563 like from arm-tdep.c.
567 Fix uses of ->function.cfunc = set_function().
569 The command.c code calls sfunc() when a set command. Rather than
570 change it suggest fixing the callback function so that it is more
573 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
575 See also ``Fix implementation of ``target xxx''.'' below.
579 IRIX 3.x support is probably broken.
583 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
584 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
586 Apart from the d30v, are there any sim/common simulators that make use
589 A brief summary of what happened is that sim/common/sim-break.c was
590 created as a good idea. It turned out a better idea was to use
591 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
595 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
596 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
600 Eliminate ``extern'' from C files.
604 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
606 Extreme care is recommeded - perhaps only modify tests that are
607 exercised by the testsuite (as determined using some type of code
612 Replace the file gdb/CONTRIBUTE with a file that is generated from the
613 gdb/doc/*.texinfo directory.
617 Rewrite/break up sparcl-tdep.c so that it uses ser*.c as the mechanism
618 for accessing either the serial or UDP port.
622 New Features and Fixes
623 ======================
625 These are harder than cleanups but easier than work involving
626 fundamental architectural change.
630 Hardware watchpoint problems on x86 OSes, including Linux:
632 1. Delete/disable hardware watchpoints should free hardware debug
634 2. Watch for different values on a viariable with one hardware debug
637 According to Eli Zaretskii <eliz@delorie.com>:
639 These are not GDB/ia32 issues per se: the above features are all
640 implemented in the DJGPP port of GDB and work in v5.0. Every
641 x86-based target should be able to lift the relevant parts of
642 go32-nat.c and use them almost verbatim. You get debug register
643 sharing through reference counts, and the ability to watch large
644 regions (up to 16 bytes) using multiple registers. (The required
645 infrastructure in high-level GDB application code, mostly in
646 breakpoint.c, is also working since v5.0.)
650 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
651 so that you can see how the GDB was created.
655 Add an "info bfd" command that displays supported object formats,
656 similarly to objdump -i.
658 Is there a command already?
662 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
664 This requires internationalization.
670 (gdb) p fwprintf(stdout,L"%S\n", f)
671 No symbol "L" in current context.
675 Cleanup configury support for optional sub-directories.
677 Check how GCC handles multiple front ends for an example of how things
678 could work. A tentative first step is to rationalize things so that
679 all sub directories are handled in a fashion similar to gdb/mi.
681 See also automake above.
685 Add a transcript mechanism to GDB.
687 Such a mechanism might log all gdb input and output to a file in a
688 form that would allow it to be replayed. It could involve ``gdb
689 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
693 Can the xdep files be replaced by autoconf?
697 Document trace machinery
701 Document ui-out and ui-file.
703 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
707 Update texinfo.tex to latest?
711 Incorporate agentexpr.texi into gdb.texinfo
713 agentexpr.texi mostly describes the details of the byte code used for
714 tracepoints, not the internals of the support for this in GDB. So it
715 looks like gdb.texinfo is a better place for this information.
717 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
721 Document overlay machinery.
725 ``(gdb) catch signal SIGNAL''
727 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
728 You can attach commands to a catch but not a handle. A handle has a
729 limited number of hardwired actions.
735 o readline/*.h bitrot
737 The TUI isn't up-to-date with
738 respect to the readline currently
739 bundled with GDB. Importing a
740 new readline is on the 5.1 wish
741 list so this can only get worse.
743 Grep for things like term_cursor_move.
745 (To be honest, I don't see anyone
746 importing a new readline before 5.1 is
749 o tui.c:va_catch_errors() bitrot
751 This nasty piece of work used knowledge
752 of the internals of GDBs error functions :-(
753 Ever since those internals were cleaned
754 up this code has been broken. :-(
756 o tuiWin.c:c_makeVisibleWithNewHeight() broken
757 tuiLayout.c:_extractDisplayStartAddr() broken
759 Both these function call find_line_pc()
760 incorrectly (wrong args, wrong return value).
762 I suspect this bug has always been there!
763 It had been hidden because those files
764 didn't include the necessary header files
767 o tuiRegs() host dependant
769 Not suprisingly, this isn't a very portable
770 section of code. However, I'm sure people
771 could live with no regs in the short to
774 o defs.h: #include "tui.h" et.al.
776 I'm not sure where this came from.
777 It was a really bad idea.
779 To get things to compile I did a nasty
780 hack (Just declare what was needed and
781 replace any expressions like xx->y.z()
782 in GDB proper with function calls). I
783 could commit it slightly cleaned up if
786 Medium Term. the #ifdef TUI and TuiDo()
787 should be changed to hooks (like GDBTK).
788 The gdb-events.[hc] is there for that
791 o tui.c:_tuiReset() host dependant
793 tui.c contains a lump of termio[s]
794 I suspect an equivalent block of
795 code can be lifted from readline.
796 An equivalent readline function may
799 o curses.h vs ncurses.h.
801 Simple portability problem.
805 This function is a mystery - where is it?
807 o tui-file.[hc] cleanup
809 This can be significantly simplified.
811 o The code should be pacified. (-Werror -W...)
813 There are plenty of #includes,
814 duplicate #includes, missing function decls
817 Some of the problems I found were through
818 fixing a few of the warnings.
820 o The code should be GNUtified.
822 It would be very nice to have this code
823 look like the rest of GDB. That way people
824 would be more accepting of it as a true
827 Until it is GNUtified it is going to stick
828 out like a sore thumb to the programmer.
830 o The code should be clearly copyrighted
832 (FSF, with due credit to HP)
836 Add support for ``gdb --- PROGRAM ARGS ...''.
837 Add support for ``gdb -cmd=...''
839 Along with many variations. Check:
841 ????? for a full discussion.
847 Implement ``(gdb) !ls''.
849 Which is very different from ``(gdb) ! ls''. Implementing the latter
852 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
856 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
861 When tab expanding something like ``set arch<tab>'' ignore the
862 deprecated ``set archdebug'' and expand to ``set architecture''.
866 Replace the code that uses the host FPU with an emulator of the target
871 The "ocd reset" command needs to flush the dcache, which requires breaking
872 the abstraction layer between the target independent and target code. One
873 way to address this is provide a generic "reset" command and target vector.
875 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
884 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
885 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
887 The thread_db assisted debugging code doesn't handle exiting threads
888 properly, at least in combination with glibc 2.1.3 (the framework is
889 there, just not the actual code). There are at least two problems
890 that prevent this from working.
892 As an additional reference point, the pre thread_db code did not work
897 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
899 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
901 Christopher Blizzard writes:
903 So, I've done some more digging into this and it looks like Jim
904 Kingdon has reported this problem in the past:
906 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
908 I can reproduce this problem both with and without Tom's patch. Has
909 anyone seen this before? Maybe have a solution for it hanging around?
912 There's a test case for this documented at:
914 when debugging threaded applications you get extra SIGTRAPs
915 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
917 [There should be a GDB testcase - cagney]
921 GDB5 TOT on unixware 7
922 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
925 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
926 > practical matter, the current thread support is somewhat more annoying
927 > than when GDB was thread-unaware.
934 New languages come onto the scene all the time.
938 Re: Various C++ things
940 value_headof/value_from_vtable_info are worthless, and should be
941 removed. The one place in printcmd.c that uses it should use the RTTI
944 RTTI for g++ should be using the typeinfo functions rather than the
945 vtables. The typeinfo functions are always at offset 4 from the
946 beginning of the vtable, and are always right. The vtables will have
947 weird names like E::VB sometimes. The typeinfo function will always
948 be "E type_info function", or somesuch.
950 value_virtual_fn_field needs to be fixed so there are no failures for
951 virtual functions for C++ using g++.
953 Testsuite cases are the major priority right now for C++ support,
954 since i have to make a lot of changes that could potentially break
959 Add support for Modula3
961 Get DEC/Compaq to contribute their Modula-3 support.
965 Remote Protocol Support
966 =======================
970 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
971 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
973 (Broken) support for GDB's remote protocol across UDP is to be
974 included in the follow-on release.
976 It should be noted that UDP can only work when the [Gg] packet fits in
979 There is also much debate over the merit of this.
983 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
985 Add support for packet enable/disable commands with these thread
986 packets. General cleanup.
988 [PATCH] Document the ThreadInfo remote protocol queries
989 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
991 [PATCH] "info threads" queries for remote.c
992 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
996 Remote protocol doco feedback.
998 Too much feedback to mention needs to be merged in (901660). Search
999 for the word ``remote''.
1002 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
1003 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
1004 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
1008 GDB doesn't recover gracefully from remote protocol errors.
1010 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
1011 ignored and a timeout is required before GDB retries. A pre-cursor to
1012 fixing this this is making GDB's remote protocol packet more robust.
1014 While downloading to a remote protocol target, gdb ignores packet
1015 errors in so far as it will continue to download with chunk N+1 even
1016 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
1017 take a painfully long time to run. As a PS that test needs to be
1018 fixed so that it builds on 16 bit machines.
1022 Fix the ``!'' packet.
1024 JT reported that the existing targets do, in fact return ``OK'' so it
1025 is possible to merge remote and extended-remote targets.
1029 Drop ``<address>'' from the [SsCc] packets.
1031 I don't think that GDB generates them so having it in the protocol is
1036 Fix doco on the ``q'' packet.
1038 It has evolved into a generic RPC. The notes should reflect this and,
1039 perhaps, the ``Q'' packet can be deprecated.
1041 The doco should mention that ``OK'' is a valid packet response.
1043 The doco should explain why ``OK'' needs to be a valid packet
1048 Add the cycle step command.
1050 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1054 Resolve how to scale things to support very large packets.
1058 Resolve how to handle a target that changes things like its endianess
1059 on the fly - should it be returned in the ``T'' packet?
1061 Underlying problem is that the register file is target endian. If the
1062 target endianess changes gdb doesn't know.
1064 Suggest encoding registers as NN!VALUE.
1068 GDB should allow incomming packets to be larger than outgoing ones. A
1069 fully loaded T packet (containing all registers) can be very large -
1070 definitly larger than a corresponding Gg packet.
1077 If / when GDB starts to support the debugging of multi-processor
1078 (rather than multi-thread) applications the symtab code will need to
1079 be updated a little so that several independent symbol tables are
1080 active at a given time.
1082 The other interesting change is a clarification of the exact meaning
1083 of CORE_ADDR and that has had consequences for a few targets (that
1084 were abusing that data type).
1088 Investiagate ways of reducing memory.
1092 Investigate ways of improving load time.
1096 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1098 Consequence of recent symtab clarification. No marks for figuring out
1099 who maintains the d10v.
1103 Get the MIPS to correctly sign extend all address <-> pointer
1106 Consequence of recent symtab clarification. No marks for figuring out
1107 who maintains the MIPS.
1111 GDB truncates 64 bit enums.
1113 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1120 There are never to many testcases.
1124 Better thread testsuite.
1128 Better C++ testsuite.
1132 Look at adding a GDB specific testsuite directory so that white box
1133 tests of key internals can be added (eg ui_file).
1137 Separate out tests that involve the floating point (FP).
1139 (Something for people brining up new targets). FP and non-fp tests
1140 are combined. I think there should be set of basic tests that
1141 exercise pure integer support and then a more expanded set that
1142 exercise FP and FP/integer interactions.
1144 As an example, the MIPS, for n32 as problems with passing FP's and
1145 structs. Since most inferior call tests include FP it is difficult to
1146 determine of the integer tests are ok.
1150 Architectural Changes: General
1151 ==============================
1153 These are harder than simple cleanups / fixes and, consequently
1154 involve more work. Typically an Architectural Change will be broken
1155 down into a more digestible set of cleanups and fixes.
1159 Cleanup software single step.
1161 At present many targets implement software single step by directly
1162 blatting memory (see rs6000-tdep.c). Those targets should register
1163 the applicable breakpoints using the breakpoint framework. Perhaphs a
1164 new internal breakpoint class ``step'' is needed.
1168 Replace READ_FP() with FRAME_HANDLE().
1170 READ_FP() is a hangover from the days of the vax when the ABI really
1171 did have a frame pointer register. Modern architectures typically
1172 construct a virtual frame-handle from the stack pointer and various
1173 other bits of string.
1175 Unfortunately GDB still treats this synthetic FP register as though it
1176 is real. That in turn really confuses users (arm and ``print $fp'' VS
1177 ``info registers fp''). The synthetic FP should be separated out of
1178 the true register set presented to the user.
1182 Register Cache Cleanup (below from Andrew Cagney)
1184 I would depict the current register architecture as something like:
1186 High GDB --> Low GDB
1191 register + REGISTER_BYTE(reg_nr)
1194 -------------------------
1195 | extern register[] |
1196 -------------------------
1198 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1199 really clear on what mechanisms they should be using to manipulate that
1200 buffer. Further, much code assumes, dangerously, that registers are
1201 contigious. Having got mips-tdep.c to support multiple ABIs, believe
1202 me, that is a bad assumption. Finally, that register cache layout is
1203 determined by the current remote/local target and _not_ the less
1204 specific target ISA. In fact, in many cases it is determined by the
1205 somewhat arbitrary layout of the [gG] packets!
1208 How I would like the register file to work is more like:
1236 The main objectives being:
1238 o a clear separation between the low
1239 level target and the high level GDB
1241 o a mechanism that solves the general
1242 problem of register aliases, overlaps
1243 etc instead of treating them as optional
1244 extras that can be wedged in as an after
1245 thought (that is a reasonable description
1246 of the current code).
1248 Identify then solve the hard case and the
1249 rest just falls out. GDB solved the easy
1250 case and then tried to ignore the real
1253 o a removal of the assumption that the
1254 mapping between the register cache
1255 and virtual registers is largely static.
1256 If you flip the USR/SSR stack register
1257 select bit in the status-register then
1258 the corresponding stack registers should
1261 o a mechanism that clearly separates the
1262 gdb internal register cache from any
1263 target (not architecture) dependent
1264 specifics such as [gG] packets.
1266 Of course, like anything, it sounds good in theory. In reality, it
1267 would have to contend with many<->many relationships at both the
1268 virt<->cache and cache<->target level. For instance:
1271 Modifying an mmx register may involve
1272 scattering values across both FP and
1273 mmpx specific parts of a buffer
1276 When writing back a SP it may need to
1277 both be written to both SP and USP.
1282 Rather than let this like the last time it was discussed, just slip, I'm
1283 first going to add this e-mail (+ references) to TODO. I'd then like to
1284 sketch out a broad strategy I think could get us there.
1287 First thing I'd suggest is separating out the ``extern registers[]''
1288 code so that we can at least identify what is using it. At present
1289 things are scattered across many files. That way we can at least
1290 pretend that there is a cache instead of a global array :-)
1292 I'd then suggest someone putting up a proposal for the pseudo-reg /
1293 high-level side interface so that code can be adopted to it. For old
1294 code, initially a blanket rename of write_register_bytes() to
1295 deprecated_write_register_bytes() would help.
1297 Following that would, finaly be the corresponding changes to the target.
1301 Check that GDB can handle all BFD architectures (Andrew Cagney)
1303 There should be a test that checks that BFD/GDB are in sync with
1304 regard to architecture changes. Something like a test that first
1305 queries GDB for all supported architectures and then feeds each back
1306 to GDB.. Anyone interested in learning how to write tests? :-)
1310 Architectural Change: Multi-arch et al.
1311 =======================================
1313 The long term objective is to remove all assumptions that there is a
1314 single target with a single address space with a single instruction
1315 set architecture and single application binary interface.
1317 This is an ongoing effort. The first milestone is to enable
1318 ``multi-arch'' where by all architectural decisions are made at
1321 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1322 ``gdbisa''. Once things are multi-arched breaking that down correctly
1323 will become much easier.
1327 GDBARCH cleanup (Andrew Cagney)
1329 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1330 into arch-utils.[hc].
1332 Document that gdbarch_init_ftype could easily fail because it didn't
1333 identify an architecture.
1337 Fix BELIEVE_PPC_PROMOTION. Change it to BELIEVE_PPC_PROMOTION_P?
1339 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1344 Fix target_signal_from_host() etc.
1346 The name is wrong for starters. ``target_signal'' should probably be
1347 ``gdb_signal''. ``from_host'' should be ``from_target_signal''.
1348 After that it needs to be multi-arched and made independent of any
1349 host signal numbering.
1351 Once this is done, the signal enum can probably be moved to
1352 include/gdb so that it is available to embedded stubs.
1356 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1359 This is a barrier to replacing mips_extra_func_info with something
1360 that works with multi-arch.
1364 Multi-arch mips_extra_func_info.
1366 This first needs the alpha to be updated so that it uses ``struct
1371 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1373 Surely one of them is redundant.
1377 Convert ALL architectures to MULTI-ARCH.
1381 Select the initial multi-arch ISA / ABI based on --target or similar.
1383 At present the default is based on what ever is first in the BFD
1384 archures table. It should be determined based on the ``--target=...''
1389 Make MIPS pure multi-arch.
1391 It is only at the multi-arch enabled stage.
1397 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1399 Can the tm.h and nm.h files be eliminated by multi-arch.
1403 Architectural Change: MI, LIBGDB and scripting languages
1404 ========================================================
1406 See also architectural changes related to the event loop. LIBGDB
1407 can't be finished until there is a generic event loop being used by
1410 The long term objective is it to be possible to integrate GDB into
1411 scripting languages.
1415 Implement generic ``(gdb) commmand > file''
1417 Once everything is going through ui_file it should be come fairly
1420 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1424 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1426 gdb_stdtarg is easily confused with gdb_stdarg.
1430 Extra ui_file methods - dump.
1432 Very useful for whitebox testing.
1436 Eliminate error_begin().
1438 With ui_file, there is no need for the statefull error_begin ()
1443 Send normal output to gdb_stdout.
1444 Send error messages to gdb_stderror.
1445 Send debug and log output log gdb_stdlog.
1447 GDB still contains many cases where (f)printf or printf_filtered () is
1448 used when it should be sending the messages to gdb_stderror or
1449 gdb_stdlog. The thought of #defining printf to something has crossed
1454 Re-do GDB's output pager.
1456 GDB's output pager still relies on people correctly using *_filtered
1457 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1458 Hopefully, with all normal output going to gdb_stdout, the pager can
1459 just look at the ui_file that the output is on and then use that to
1460 decide what to do about paging. Sounds good in theory.
1464 Check/cleanup MI documentation.
1466 The list of commands specified in the documentation needs to be
1467 checked against the mi-cmds.c table in a mechanical way (so that they
1468 two can be kept up-to-date).
1472 Convert MI into libgdb
1474 MI provides a text interface into what should be many of the libgdb
1475 functions. The implementation of those functions should be separated
1476 into the MI interface and the functions proper. Those functions being
1477 moved to gdb/lib say.
1483 The first part can already be found in defs.h.
1487 MI's input does not use buffering.
1489 At present the MI interface reads raw characters of from an unbuffered
1490 FD. This is to avoid several nasty buffer/race conditions. That code
1491 should be changed so that it registers its self with the event loop
1492 (on the input FD) and then push commands up to MI as they arrive.
1494 The serial code already does this.
1498 Make MI interface accessible from existing CLI.
1502 Add a breakpoint-edit command to MI.
1504 It would be similar to MI's breakpoint create but would apply to an
1505 existing breakpoint. It saves the need to delete/create breakpoints
1506 when ever they are changed.
1510 Add directory path to MI breakpoint.
1512 That way the GUI's task of finding the file within which the
1513 breakpoint was set is simplified.
1517 Add a mechanism to reject certain expression classes to MI
1519 There are situtations where you don't want GDB's expression
1520 parser/evaluator to perform inferior function calls or variable
1521 assignments. A way of restricting the expression parser so that such
1522 operations are not accepted would be very helpful.
1526 Remove sideffects from libgdb breakpoint create function.
1528 The user can use the CLI to create a breakpoint with partial
1529 information - no file (gdb would use the file from the last
1532 The libgdb interface currently affects that environment which can lead
1533 to confusion when a user is setting breakpoints via both the MI and
1536 This is also a good example of how getting the CLI ``right'' will be
1541 Move gdb_lasterr to ui_out?
1543 The way GDB throws errors and records them needs a re-think. ui_out
1544 handles the correct output well. It doesn't resolve what to do with
1545 output / error-messages when things go wrong.
1549 do_setshow_command contains a 1024 byte buffer.
1551 The function assumes that there will never be any more than 1024 bytes
1552 of enum. It should use mem_file.
1556 Should struct cmd_list_element . completer take the command as an
1561 Should the bulk of top.c:line_completion_function() be moved to
1562 command.[hc]? complete_on_cmdlist() and complete_on_enums() could
1563 then be made private.
1567 top.c (execute_command): Should a command being valid when the target
1568 is running be made an attribute (predicate) to the command rather than
1569 an explicit set of tests.
1573 top.c (execute_command): Should the bulk of this function be moved
1574 into command.[hc] so that top.c doesn't grub around in the command
1579 Architectural Change: Async
1580 ===========================
1582 While GDB uses an event loop when prompting the user for input. That
1583 event loop is not exploited by targets when they allow the target
1584 program to continue. Typically targets still block in (target_wait())
1585 until the program again halts.
1587 The closest a target comes to supporting full asynchronous mode are
1588 the remote targets ``async'' and ``extended-async''.
1592 Asynchronous expression evaluator
1594 Inferior function calls hang GDB.
1598 Fix implementation of ``target xxx''.
1600 At present when the user specifies ``target xxxx'', the CLI maps that
1601 directly onto a target open method. It is then assumed that the
1602 target open method should do all sorts of complicated things as this
1603 is the only chance it has. Check how the various remote targets
1604 duplicate the target operations. Check also how the various targets
1605 behave differently for purely arbitrary reasons.
1607 What should happen is that ``target xxxx'' should call a generic
1608 ``target'' function and that should then co-ordinate the opening of
1609 ``xxxx''. This becomes especially important when you're trying to
1610 open an asynchronous target that may need to perform background tasks
1611 as part of the ``attach'' phase.
1613 Unfortunately, due to limitations in the old/creaking command.h
1614 interface, that isn't possible. The function being called isn't told
1615 of the ``xxx'' or any other context information.
1617 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1618 CLI code so that it passes to the callback function (attatched to a
1619 command) useful information such as the actual command and a context
1620 for that command. Other changes such as making ``struct command''
1621 opaque may also help.
1624 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1628 Make "target xxx" command interruptible.
1630 As things become async this becomes possible. A target would start
1631 the connect and then return control to the event loop. A cntrl-c
1632 would notify the target that the operation is to be abandoned and the
1633 target code could respond.
1637 Add a "suspend" subcommand of the "continue" command to suspend gdb
1638 while continuing execution of the subprocess. Useful when you are
1639 debugging servers and you want to dodge out and initiate a connection
1640 to a server running under gdb.
1649 Frequently requested but not approved requests.
1653 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1655 The benefits on this one are thought to be marginal - GDBs design
1656 means that unused parameters are very common. GCC 3.0 will also
1657 include the option -Wno-unused-parameter which means that ``-Wall
1658 -Wno-unused-parameters -Werror'' can be specified.
1667 This list is not up to date, and opinions vary about the importance or
1668 even desirability of some of the items. If you do fix something, it
1669 always pays to check the below.
1673 @c This does not work (yet if ever). FIXME.
1674 @c @item --parse=@var{lang} @dots{}
1675 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1676 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
1677 @c list of all supported languages, omit the argument. Without this
1678 @c option, @value{GDBN} is configured to parse all supported languages.
1682 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1683 is its default value. Clean this up.
1687 It should be possible to use symbols from shared libraries before we know
1688 exactly where the libraries will be loaded. E.g. "b perror" before running
1689 the program. This could maybe be done as an extension of the "breakpoint
1690 re-evaluation" after new symbols are loaded.
1694 Make single_step() insert and remove breakpoints in one operation.
1696 [If this is talking about having single_step() insert the breakpoints,
1697 run the target then pull the breakpoints then it is wrong. The
1698 function has to return as control has to eventually be passed back to
1699 the main event loop.]
1703 Speed up single stepping by avoiding extraneous ptrace calls.
1707 Speed up single stepping by not inserting and removing breakpoints
1708 each time the inferior starts and stops.
1710 Breakpoints should not be inserted and deleted all the time. Only the
1711 one(s) there should be removed when we have to step over one. Support
1712 breakpoints that don't have to be removed to step over them.
1714 [this has resulted in numerous debates. The issue isn't clear cut]
1718 Provide "voodoo" debugging of core files. This creates a zombie
1719 process as a child of the debugger, and loads it up with the data,
1720 stack, and regs of the core file. This allows you to call functions
1721 in the executable, to manipulate the data in the core file.
1727 GDB reopens the source file on every line, as you "next" through it.
1729 [still true? I've a memory of this being fixed]
1733 Perhaps "i source" should take an argument like that of "list".
1737 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1738 it matches the source line indicated.
1742 The prompt at end of screen should accept space as well as CR.
1746 Backtrace should point out what the currently selected frame is, in
1747 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1748 ...)" rather than "#3 foo (bar, ...)".
1752 "i program" should work for core files, and display more info, like what
1753 actually caused it to die.
1757 "x/10i" should shorten the long name, if any, on subsequent lines.
1761 "next" over a function that longjumps, never stops until next time you happen
1762 to get to that spot by accident. E.g. "n" over execute_command which has
1767 "set zeroprint off", don't bother printing members of structs which
1768 are entirely zero. Useful for those big structs with few useful
1773 GDB does four ioctl's for every command, probably switching terminal modes
1774 to/from inferior or for readline or something.
1778 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
1779 if the state is the same, too.
1783 "i frame" shows wrong "arglist at" location, doesn't show where the args
1784 should be found, only their actual values.
1788 There should be a way for "set" commands to validate the new setting
1789 before it takes effect.
1793 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
1797 i line VAR produces "Line number not known for symbol ``var''.". I
1798 thought we were stashing that info now!
1802 We should be able to write to random files at hex offsets like adb.
1806 [elena - delete this]
1808 Handle add_file with separate text, data, and bss addresses. Maybe
1809 handle separate addresses for each segment in the object file?
1813 [Jimb/Elena delete this one]
1815 Handle free_named_symtab to cope with multiply-loaded object files
1816 in a dynamic linking environment. Should remember the last copy loaded,
1817 but not get too snowed if it finds references to the older copy.
1821 [elena delete this also]
1823 Remove all references to:
1830 now that we have BFD. All remaining are in machine dependent files.
1834 Re-organize help categories into things that tend to fit on a screen
1839 Add in commands like ADB's for searching for patterns, etc. We should
1840 be able to examine and patch raw unsymboled binaries as well in gdb as
1841 we can in adb. (E.g. increase the timeout in /bin/login without source).
1843 [actually, add ADB interface :-]
1847 When doing "step" or "next", if a few lines of source are skipped between
1848 the previous line and the current one, print those lines, not just the
1849 last line of a multiline statement.
1853 Handling of "&" address-of operator needs some serious overhaul
1854 for ANSI C and consistency on arrays and functions.
1855 For "float point[15];":
1856 ptype &point[4] ==> Attempt to take address of non-lvalue.
1857 For "char *malloc();":
1858 ptype malloc ==> "char *()"; should be same as
1859 ptype &malloc ==> "char *(*)()"
1860 call printf ("%x\n", malloc) ==> weird value, should be same as
1861 call printf ("%x\n", &malloc) ==> correct value
1865 Fix dbxread.c symbol reading in the presence of interrupts. It
1866 currently leaves a cleanup to blow away the entire symbol table when a
1867 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
1869 [I suspect that the grype was that, on a slow system, you might want
1870 to cntrl-c and get just half the symbols and then load the rest later
1871 - scary to be honest]
1875 Mipsread.c reads include files depth-first, because the dependencies
1876 in the psymtabs are way too inclusive (it seems to me). Figure out what
1877 really depends on what, to avoid recursing 20 or 30 times while reading
1882 value_add() should be subtracting the lower bound of arrays, if known,
1883 and possibly checking against the upper bound for error reporting.
1887 When listing source lines, check for a preceding \n, to verify that
1888 the file hasn't changed out from under us.
1890 [fixed by some other means I think. That hack wouldn't actually work
1891 reliably - the file might move such that another \n appears. ]
1895 Get all the remote systems (where the protocol allows it) to be able to
1896 stop the remote system when the GDB user types ^C (like remote.c
1897 does). For ebmon, use ^Ak.
1901 Possible feature: A version of the "disassemble" command which shows
1902 both source and assembly code ("set symbol-filename on" is a partial
1905 [has this been done? It was certainly done for MI and GDBtk]
1909 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1910 using a 0 address for bad purposes internally).
1914 Make "info path" and path_command work again (but independent of the
1915 environment either of gdb or that we'll pass to the inferior).
1919 Make GDB understand the GCC feature for putting octal constants in
1920 enums. Make it so overflow on an enum constant does not error_type
1921 the whole type. Allow arbitrarily large enums with type attributes.
1922 Put all this stuff in the testsuite.
1926 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1927 the value in hex; process type attributes). Add this to the
1928 testsuite. This way future compilers can add new types and old
1929 versions of GDB can do something halfway reasonable.
1933 Fix mdebugread.c:parse_type to do fundamental types right (see
1934 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1935 that the debug format fixes the sizes of these things and it shouldn't
1936 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1937 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1938 be aware of that). Also use a switch statement for clarity and speed.
1942 Investigate adding symbols in target_load--some targets do, some
1947 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1948 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1953 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1958 Make a watchpoint which contains a function call an error (it is
1959 broken now, making it work is probably not worth the effort).
1963 New test case based on weird.exp but in which type numbers are not
1964 renumbered (thus multiply defining a type). This currently causes an
1965 infinite loop on "p v_comb".
1969 [Hey! Hint Hint Delete Delete!!!]
1971 Fix 386 floating point so that floating point registers are real
1972 registers (but code can deal at run-time if they are missing, like
1973 mips and 68k). This would clean up "info float" and related stuff.
1977 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1978 about not being able to access memory location 0.
1980 -------------------- enummask.c
2001 If try to modify value in file with "set write off" should give
2002 appropriate error not "cannot access memory at address 0x65e0".
2006 Allow core file without exec file on RS/6000.
2010 Make sure "shell" with no arguments works right on DOS.
2014 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
2015 the same directory can be NFS-mounted on unix or DOS, and work the
2020 [Is this another delete???]
2022 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
2023 get RS/6000 to work right, might not be immediately relevant).
2027 Work out some kind of way to allow running the inferior to be done as
2028 a sub-execution of, eg. breakpoint command lists. Currently running
2029 the inferior interupts any command list execution. This would require
2030 some rewriting of wait_for_inferior & friends, and hence should
2031 probably be done in concert with the above.
2035 Add function arguments to gdb user defined functions.
2039 Add convenience variables that refer to exec file, symbol file,
2040 selected frame source file, selected frame function, selected frame
2045 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2046 allocate a partial symtab for each BINCL/EINCL grouping. This will
2047 seriously decrease the size of inter-psymtab dependencies and hence
2048 lessen the amount that needs to be read in when a new source file is
2053 Add a command for searching memory, a la adb. It specifies size,
2054 mask, value, start address. ADB searches until it finds it or hits
2055 an error (or is interrupted).
2059 Remove the range and type checking code and documentation, if not