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.
7 Known problems in GDB 5.0
8 =========================
10 Below is a list of problems identified during the GDB 5.0 release
11 cycle. People hope to have these problems fixed in a follow-on
14 (The names in paren indicate people that posted the original problem.)
18 GDB doesn't build under IRIX6.4
22 Has anyone successfully built the latest (from cvs) gdb on IRIX6.4 or
23 later? The first problem I hit is that proc-api.c includes
24 sys/user.h, which no longer exists under IRIX6.4. If I comment out
25 that include, the next problem I hit is that PIOCGETPR and PIOCGETU
26 are no longer defined in IRIX6.4 (presumably related to the
27 disappearance of user.h).
31 The BFD directory requires bug-fixed AUTOMAKE et.al.
33 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable. It
34 contained the full path to texinfo.tex when it should have only
35 contained the directory. The bug has been fixed in the current
36 AUTOMAKE sources. Automake snapshots can be found in:
37 ftp://sourceware.cygnus.com/pub/gdb/snapshots
38 and ftp://sourceware.cygnus.com/pub/binutils
42 gdb-cvs fails to build on freebsd-elf
43 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00004.html
45 Either the FreeBSD group need to contribute their local GDB changes
46 back to the master sources or someone needs to provides a new
47 (clean-room) implementation. Since the former involves a fairly
48 complicated assignment the latter may be easier. [cagney]
52 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
53 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
55 The thread_db assisted debugging code doesn't handle exiting threads
56 properly, at least in combination with glibc 2.1.3 (the framework is
57 there, just not the actual code). There are at least two problems
58 that prevent this from working.
60 As an additional reference point, the pre thread_db code did not work
65 Java (Anthony Green, David Taylor)
67 Anthony Green has a number of Java patches that did not make it into
71 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
74 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
76 Patch: handle N_MAIN stab
77 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
81 Pascal (Pierre Muller, David Taylor)
83 Pierre Muller has contributed patches for adding Pascal Language
86 2 pascal language patches inserted in database
87 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
90 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
94 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
96 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
98 Christopher Blizzard writes:
100 So, I've done some more digging into this and it looks like Jim
101 Kingdon has reported this problem in the past:
103 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
105 I can reproduce this problem both with and without Tom's patch. Has
106 anyone seen this before? Maybe have a solution for it hanging around?
109 There's a test case for this documented at:
111 when debugging threaded applications you get extra SIGTRAPs
112 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
114 [There should be a GDB testcase - cagney]
118 Possible regressions with some devel GCCs.
119 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00475.html
121 gcc-2.95.2 outputs a line note *before* the prologue (and one for the
122 closing brace after the epilogue, instead of before it, as it used to
123 be). By disabling the RTL-style prologue generating mechanism
124 (undocumented GCC option -mno-schedule-prologue), you get back the
125 traditional behaviour.
126 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00510.html
128 This should now be fixed.
132 RFD: infrun.c: No bpstat_stop_status call after proceed over break?
134 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00665.html
136 GDB misses watchpoint triggers after proceeding over a breakpoint on
141 x86 linux GDB and SIGALRM (???)
142 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00803.html
144 I know there are problems with single stepping through signal
145 handlers. These problems were present in 4.18. They were just masked
146 because 4.18 failed to recognize signal handlers. Fixing it is not
147 easy, and will require changes to handle_inferior_event(), that I
148 prefer not to make before the 5.0 release.
154 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
155 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
157 (Broken) support for GDB's remote protocol across UDP is to be
158 included in the follow-on release.
162 Can't build IRIX -> arm GDB.
163 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
166 > Now I'm building for an embedded arm target. If there is a way of turning
167 > remote-rdi off, I couldn't find it. It looks like it gets built by default
168 > in gdb/configure.tgt(line 58) Anyway, the build dies in
169 > gdb/rdi-share/unixcomm.c. SERPORT1 et. al. never get defined because we
170 > aren't one of the architectures supported.
174 Problem with weak functions
175 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
177 Dan Nicolaescu writes:
178 > It seems that gdb-4.95.1 does not display correctly the function when
179 > stoping in weak functions.
181 > It stops in a function that is defined as weak, not in the function
182 > that is actualy run...
186 GDB5 TOT on unixware 7
187 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
190 > I just spun the top of tree of the GDB5 branch on UnixWare 7. As a
191 > practical matter, the current thread support is somewhat more annoying
192 > than when GDB was thread-unaware.
199 The following are small cleanups that will hopefully be completed by
200 the follow on to 5.0.
206 The need for this as almost been eliminated. The next version of GCC
207 (assuming cagney gets the relevant patch committed) will be able to
208 supress unused parameter warnings.
212 Eliminate more compiler warnings.
214 Of course there also needs to be the usual debate over which warnings
215 are valid and how to best go about this.
217 One method: choose a single option; get agreement that it is
218 reasonable; try it out to see if there isn't anything silly about it
219 (-Wunused-parameters is an example of that) then incrementally hack
222 The other method is to enable all warnings and eliminate them from one
227 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
229 Patches in the database.
235 Readline 4.? is out. A merge wouldn't hurt.
241 Eliminate all uses of PARAMS in GDB's source code.
245 Elimination of make_cleanup_func. (Andrew Cagney)
247 make_cleanup_func elimination
248 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
249 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
253 Elimination of ``(catch_errors_ftype *) func''.
255 Like make_cleanup_func it isn't portable.
259 Re: Various C++ things
261 value_headof/value_from_vtable_info are worthless, and should be removed.
262 The one place in printcmd.c that uses it should use the RTTI functions.
264 RTTI for g++ should be using the typeinfo functions rather than the vtables.
265 The typeinfo functions are always at offset 4 from the beginning of the vtable,
266 and are always right. The vtables will have weird names like E::VB sometimes.
267 The typeinfo function will always be "E type_info function", or somesuch.
269 value_virtual_fn_field needs to be fixed so there are no failures for virtual
270 functions for C++ using g++.
272 Testsuite cases are the major priority right now for C++ support, since i have
273 to make a lot of changes that could potentially break each other.
277 Fix ``set architecture <tab>''
279 This command should expand to a list of all supported architectures.
280 At present ``info architecture'' needs to be used. That is simply
281 wrong. It involves the use of add_set_enum_cmd().
285 GDBARCH cleanup (Andrew Cagney)
287 The non-generated parts of gdbarch.{sh,h,c} should be separated out
288 into arch-utils.[hc].
290 Document that gdbarch_init_ftype could easily fail because it didn't
291 identify an architecture.
295 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
297 Add support for packet enable/disable commands with these thread
298 packets. General cleanup.
300 [PATCH] Document the ThreadInfo remote protocol queries
301 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
303 [PATCH] "info threads" queries for remote.c
304 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
308 General Cleanups / Fixes
309 ========================
311 The following are more general cleanups and fixes. They are not tied
312 to any specific release.
320 Eliminate gdb/tui/Makefile.in.
321 Cleanup configury support for optional sub-directories.
323 Check how GCC handles multiple front ends for an example of how things
324 could work. A tentative first step is to rationalize things so that
325 all sub directories are handled in a fashion similar to gdb/mi.
329 [PATCH/5] src/intl/Makefile.in:distclean additions
330 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
332 Do not forget to merge the patch back into the trunk.
336 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
339 This is a barrier to replacing mips_extra_func_info with something
340 that works with multi-arch.
344 Multi-arch mips_extra_func_info.
346 This first needs the alpha to be updated so that it uses ``struct
351 Send normal output to gdb_stdout.
352 Send error messages to gdb_stderror.
353 Send debug and log output log gdb_stdlog.
355 GDB still contains many cases where (f)printf or printf_filtered () is
356 used when it should be sending the messages to gdb_stderror or
357 gdb_stdlog. The thought of #defining printf to something has crossed
362 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
364 At present defs.h includes <endian.h> (which is linux specific) yet
365 almost nothing depends on it. Suggest "gdb_endian.h" which can also
366 handle <machine/endian.h> and only include that where it is really
371 Replace asprintf() calls with xasprintf() calls.
373 As with things like strdup() most calls to asprintf() don't check the
378 Rationaize savestring(), msavestring() and mstrsave().
380 In general libiberty's xstrdup () can be used.
384 Eliminate mmalloc() from GDB.
386 Also eliminate it from defs.h.
390 Check/cleanup MI documentation.
392 The list of commands specified in the documentation needs to be
393 checked against the mi-cmds.c table in a mechanical way (so that they
394 two can be kept up-to-date).
398 Eliminate error_begin().
400 With ui_file, there is no need for the statefull error_begin ()
405 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
406 so that you can see how the GDB was created.
408 Some of these (*m.h) would be added to the generated config.h. That
409 in turn would fix a long standing bug where by the build process many
410 not notice a changed tm.h file. Since everything depends on config.h,
411 a change to *m.h forces a change to config.h and, consequently forces
416 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
418 gdb_stdtarg is easily confused with gdb_stdarg.
422 Remote protocol doco feedback.
424 Too much feedback to mention needs to be merged in (901660). Search
425 for the word ``remote''.
429 set/show remote X-packet ...
431 ``(gdb) help set remote X-packet'' doesn't list the applicable
432 responses. The help message needs to be expanded.
436 Extra ui_file methods - dump.
438 These are for debugging / testing. An aside is to set up a whitebox
439 testsuite for key internals such as ui_file.
443 Add an "info bfd" command that displays supported object formats,
444 similarly to objdump -i.
446 Is there a command already?
450 Eliminate PTR. ISO-C allows ``void *''.
456 GDB should never abort. GDB should either throw ``error ()'' or
457 ``internal_error ()''. Better still GDB should naturally unwind with
462 Architectural Changes
463 =====================
465 These are harder than simple cleanups / fixes and, consequently
466 involve more work. Typically an Architectural Change will be broken
467 down into a more digestible set of cleanups and fixes.
471 Replace READ_FP() with FRAME_HANDLE().
473 READ_FP() is a hangover from the days of the vax when the ABI really
474 did have a frame pointer register. Modern architectures typically
475 construct a virtual frame-handle from the stack pointer and various
476 other bits of string.
478 Unfortunatly GDB still treats this synthetic FP register as though it
479 is real. That in turn really confuses users (arm and ``print $fp'' VS
480 ``info registers fp''). The synthetic FP should be separated out of
481 the true register set presented to the user.
485 MI's input does not use buffering.
487 At present the MI interface reads raw characters of from an unbuffered
488 FD. This is to avoid several nasty buffer/race conditions. That code
489 should be changed so that it registers its self with the event loop
490 (on the input FD) and then push commands up to MI as they arrive.
492 The serial code already does this.
496 Register Cache Cleanup (below from Andrew Cagney)
498 I would depict the current register architecture as something like:
505 register + REGISTER_BYTE(reg_nr)
508 -------------------------
509 | extern register[] |
510 -------------------------
512 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
513 really clear on what mechanisms they should be using to manipulate that
514 buffer. Further, much code assumes, dangerously, that registers are
515 contigious. Having got mips-tdep.c to support multiple ABIs, believe
516 me, that is a bad assumption. Finally, that register cache layout is
517 determined by the current remote/local target and _not_ the less
518 specific target ISA. In fact, in many cases it is determined by the
519 somewhat arbitrary layout of the [gG] packets!
522 How I would like the register file to work is more like:
550 The main objectives being:
552 o a clear separation between the low
553 level target and the high level GDB
555 o a mechanism that solves the general
556 problem of register aliases, overlaps
557 etc instead of treating them as optional
558 extras that can be wedged in as an after
559 thought (that is a reasonable description
560 of the current code).
562 Identify then solve the hard case and the
563 rest just falls out. GDB solved the easy
564 case and then tried to ignore the real
567 o a removal of the assumption that the
568 mapping between the register cache
569 and virtual registers is largely static.
570 If you flip the USR/SSR stack register
571 select bit in the status-register then
572 the corresponding stack registers should
575 o a mechanism that clearly separates the
576 gdb internal register cache from any
577 target (not architecture) dependant
578 specifics such as [gG] packets.
580 Of course, like anything, it sounds good in theory. In reality, it
581 would have to contend with many<->many relationships at both the
582 virt<->cache and cache<->target level. For instance:
585 Modifying an mmx register may involve
586 scattering values across both FP and
587 mmpx specific parts of a buffer
590 When writing back a SP it may need to
591 both be written to both SP and USP.
596 Rather than let this like the last time it was discussed, just slip, I'm
597 first going to add this e-mail (+ references) to TODO. I'd then like to
598 sketch out a broad strategy I think could get us there.
601 First thing I'd suggest is separating out the ``extern registers[]''
602 code so that we can at least identify what is using it. At present
603 things are scattered across many files. That way we can at least
604 pretend that there is a cache instead of a global array :-)
606 I'd then suggest someone putting up a proposal for the pseudo-reg /
607 high-level side interface so that code can be adopted to it. For old
608 code, initially a blanket rename of write_register_bytes() to
609 deprecated_write_register_bytes() would help.
611 Following that would, finaly be the corresponding changes to the target.
615 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
617 This requires internationalization.
621 Check that GDB can handle all BFD architectures (Andrew Cagney)
623 There should be a test that checks that BFD/GDB are in sync with
624 regard to architecture changes. Something like a test that first
625 queries GDB for all supported architectures and then feeds each back
626 to GDB.. Anyone interested in learning how to write tests? :-)
630 Add support for Modula3
632 Get DEC/Compaq to contribute their Modula-3 support.
636 Convert ALL architectures to MULTI-ARCH.
640 Convert GDB build process to AUTOMAKE.
644 Restructure gdb directory tree so that it avoids any 8.3 and 14
649 Can the xdep files be replaced by autoconf?
650 Can the tm.h and nm.h files be eliminated by multi-arch.
654 Add a transcript mechanism to GDB.
656 Such a mechanism might log all gdb input and output to a file in a
657 form that would allow it to be replayed. It could involve ``gdb
658 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
662 Make MI interface accessable from existing CLI.
666 Select the initial multi-arch ISA / ABI based on --target or similar.
668 At present the default is based on what ever is first in the BFD
669 archures table. It should be determined based on the ``--target=...''
676 Enable the code to recognize --enable-targets=.... like BINUTILS does.
680 Add a breakpoint-edit command to MI.
682 It would be similar to MI's breakpoint create but would apply to an
683 existing breakpoint. It saves the need to delete/create breakpoints
684 when ever they are changed.
688 Add directory path to MI breakpoint.
690 That way the GUI's task of finding the file within which the
691 breakpoint was set is simplified.
695 Re-do GDB's output pager.
697 GDB's output pager still relies on people correctly using *_filtered
698 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
699 Hopefully, with all normal output going to gdb_stdout, the pager can
700 just look at the ui_file that the output is on and then use that to
701 decide what to do about paging. Sounds good in theory.
705 Add mechanism to reject expression classes to MI
707 There are situtations where you don't want GDB's expression
708 parser/evaluator to perform inferior function calls or variable
713 Remove sideffects from libgdb breakpoint create function.
715 The user can use the CLI to create a breakpoint with partial
716 information - no file (gdb would use the file from the last
719 The libgdb interface currently affects that environment which can lead
720 to confusion when a user is setting breakpoints via both the MI and
723 This is also a good example of how getting the CLI ``right'' will be
728 GDB doesn't recover gracefully from remote protocol errors.
730 GDB wasn't checking for NAKs from the remote target. Instead a NAK is
731 ignored and a timeout is required before GDB retries. A pre-cursor to
732 fixing this this is making GDB's remote protocol packet more robust.
734 While downloading to a remote protocol target, gdb ignores packet
735 errors in so far as it will continue to edownload with chunk N+1 even
736 if chunk N was not correctly sent. This causes gdb.base/remote.exp to
737 take a painfully long time to run. As a PS that test needs to be
738 fixed so that it builds on 16 bit machines.
742 Move gdb_lasterr to ui_out?
744 The way GDB throws errors and records them needs a re-think. ui_out
745 handles the correct output well. It doesn't resolve what to do with
746 output / error-messages when things go wrong.
750 Fix implementation of ``target xxx''.
752 At present when the user specifies ``target xxxx'', the CLI maps that
753 directly onto a target open method. It is then assumed that the
754 target open method should do all sorts of complicated things as this
755 is the only chance it has. Check how the various remote targets
756 duplicate the target operations. Check also how the various targets
757 behave differently for purely arbitrary reasons.
759 What should happen is that ``target xxxx'' should call a generic
760 ``target'' function and that should then co-ordinate the opening of
761 ``xxxx''. This becomes especially important when you're trying to
762 open an asynchronous target that may need to perform background tasks
763 as part of the ``attach'' phase.
765 Unfortunatly, due to limitations in the old/creaking command.h
766 interface, that isn't possible. The function being called isn't told
767 of the ``xxx'' or any other context information.
769 Consequently a precursor to fixing ``target xxxx'' is to clean up the
770 CLI code so that it passes to the callback function (attatched to a
771 command) useful information such as the actual command and a context
772 for that command. Other changes such as making ``struct command''
773 opaque may also help.
777 Document trace machinery
781 Document overlay machinery.
788 This list is not up to date, and opinions vary about the importance or
789 even desirability of some of the items. If you do fix something, it
790 always pays to check the below.
794 @c This does not work (yet if ever). FIXME.
795 @c @item --parse=@var{lang} @dots{}
796 @c Configure the @value{GDBN} expression parser to parse the listed languages.
797 @c @samp{all} configures @value{GDBN} for all supported languages. To get a
798 @c list of all supported languages, omit the argument. Without this
799 @c option, @value{GDBN} is configured to parse all supported languages.
803 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
804 is its default value. Clean this up.
808 It should be possible to use symbols from shared libraries before we know
809 exactly where the libraries will be loaded. E.g. "b perror" before running
810 the program. This could maybe be done as an extension of the "breakpoint
811 re-evaluation" after new symbols are loaded.
815 Make single_step() insert and remove breakpoints in one operation.
819 Speed up single stepping by avoiding extraneous ptrace calls.
823 Speed up single stepping by not inserting and removing breakpoints
824 each time the inferior starts and stops.
826 Breakpoints should not be inserted and deleted all the time. Only the
827 one(s) there should be removed when we have to step over one. Support
828 breakpoints that don't have to be removed to step over them.
830 [this has resulted in numerous debates. The issue isn't clear cut]
834 Provide "voodoo" debugging of core files. This creates a zombie
835 process as a child of the debugger, and loads it up with the data,
836 stack, and regs of the core file. This allows you to call functions
837 in the executable, to manipulate the data in the core file.
843 GDB reopens the source file on every line, as you "next" through it.
845 [still true? I've a memory of this being fixed]
849 Perhaps "i source" should take an argument like that of "list".
853 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
854 it matches the source line indicated.
858 The prompt at end of screen should accept space as well as CR.
862 Backtrace should point out what the currently selected frame is, in
863 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
864 ...)" rather than "#3 foo (bar, ...)".
868 "i program" should work for core files, and display more info, like what
869 actually caused it to die.
873 "x/10i" should shorten the long name, if any, on subsequent lines.
877 "next" over a function that longjumps, never stops until next time you happen
878 to get to that spot by accident. E.g. "n" over execute_command which has
883 "set zeroprint off", don't bother printing members of structs which
884 are entirely zero. Useful for those big structs with few useful
889 GDB does four ioctl's for every command, probably switching terminal modes
890 to/from inferior or for readline or something.
894 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
895 if the state is the same, too.
899 "i frame" shows wrong "arglist at" location, doesn't show where the args
900 should be found, only their actual values.
904 There should be a way for "set" commands to validate the new setting
905 before it takes effect.
909 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
913 i line VAR produces "Line number not known for symbol ``var''.". I
914 thought we were stashing that info now!
918 We should be able to write to random files at hex offsets like adb.
922 Make "target xxx" command interruptible.
926 [elena - delete this]
928 Handle add_file with separate text, data, and bss addresses. Maybe
929 handle separate addresses for each segment in the object file?
933 [Jimb/Elena delete this one]
935 Handle free_named_symtab to cope with multiply-loaded object files
936 in a dynamic linking environment. Should remember the last copy loaded,
937 but not get too snowed if it finds references to the older copy.
941 [elena delete this also]
943 Remove all references to:
950 now that we have BFD. All remaining are in machine dependent files.
954 Re-organize help categories into things that tend to fit on a screen
959 Add in commands like ADB's for searching for patterns, etc. We should
960 be able to examine and patch raw unsymboled binaries as well in gdb as
961 we can in adb. (E.g. increase the timeout in /bin/login without source).
963 [actually, add ADB interface :-]
967 When doing "step" or "next", if a few lines of source are skipped between
968 the previous line and the current one, print those lines, not just the
969 last line of a multiline statement.
973 Handling of "&" address-of operator needs some serious overhaul
974 for ANSI C and consistency on arrays and functions.
975 For "float point[15];":
976 ptype &point[4] ==> Attempt to take address of non-lvalue.
977 For "char *malloc();":
978 ptype malloc ==> "char *()"; should be same as
979 ptype &malloc ==> "char *(*)()"
980 call printf ("%x\n", malloc) ==> weird value, should be same as
981 call printf ("%x\n", &malloc) ==> correct value
985 Fix dbxread.c symbol reading in the presence of interrupts. It
986 currently leaves a cleanup to blow away the entire symbol table when a
987 QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993).
989 [I suspect that the grype was that, on a slow system, you might want
990 to cntrl-c and get just half the symbols and then load the rest later
991 - scary to be honest]
995 Mipsread.c reads include files depth-first, because the dependencies
996 in the psymtabs are way too inclusive (it seems to me). Figure out what
997 really depends on what, to avoid recursing 20 or 30 times while reading
1002 value_add() should be subtracting the lower bound of arrays, if known,
1003 and possibly checking against the upper bound for error reporting.
1007 When listing source lines, check for a preceding \n, to verify that
1008 the file hasn't changed out from under us.
1010 [fixed by some other means I think. That hack wouldn't actually work
1011 reliably - the file might move such that another \n appears. ]
1015 Get all the remote systems (where the protocol allows it) to be able to
1016 stop the remote system when the GDB user types ^C (like remote.c
1017 does). For ebmon, use ^Ak.
1021 Possible feature: A version of the "disassemble" command which shows
1022 both source and assembly code ("set symbol-filename on" is a partial
1025 [has this been done? It was certainly done for MI and GDBtk]
1029 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1030 using a 0 address for bad purposes internally).
1034 Make "info path" and path_command work again (but independent of the
1035 environment either of gdb or that we'll pass to the inferior).
1039 Make GDB understand the GCC feature for putting octal constants in
1040 enums. Make it so overflow on an enum constant does not error_type
1041 the whole type. Allow arbitrarily large enums with type attributes.
1042 Put all this stuff in the testsuite.
1046 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1047 the value in hex; process type attributes). Add this to the
1048 testsuite. This way future compilers can add new types and old
1049 versions of GDB can do something halfway reasonable.
1053 Fix mdebugread.c:parse_type to do fundamental types right (see
1054 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1055 that the debug format fixes the sizes of these things and it shouldn't
1056 depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem
1057 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1058 be aware of that). Also use a switch statement for clarity and speed.
1062 Investigate adding symbols in target_load--some targets do, some
1067 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1068 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1073 Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of
1078 Make a watchpoint which contains a function call an error (it is
1079 broken now, making it work is probably not worth the effort).
1083 New test case based on weird.exp but in which type numbers are not
1084 renumbered (thus multiply defining a type). This currently causes an
1085 infinite loop on "p v_comb".
1089 [Hey! Hint Hint Delete Delete!!!]
1091 Fix 386 floating point so that floating point registers are real
1092 registers (but code can deal at run-time if they are missing, like
1093 mips and 68k). This would clean up "info float" and related stuff.
1097 gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains
1098 about not being able to access memory location 0.
1100 -------------------- enummask.c
1121 If try to modify value in file with "set write off" should give
1122 appropriate error not "cannot access memory at address 0x65e0".
1126 Allow core file without exec file on RS/6000.
1130 Make sure "shell" with no arguments works right on DOS.
1134 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1135 the same directory can be NFS-mounted on unix or DOS, and work the
1140 [Is this another delete???]
1142 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
1143 get RS/6000 to work right, might not be immediately relevant).
1147 Work out some kind of way to allow running the inferior to be done as
1148 a sub-execution of, eg. breakpoint command lists. Currently running
1149 the inferior interupts any command list execution. This would require
1150 some rewriting of wait_for_inferior & friends, and hence should
1151 probably be done in concert with the above.
1155 Add function arguments to gdb user defined functions.
1159 Add convenience variables that refer to exec file, symbol file,
1160 selected frame source file, selected frame function, selected frame
1165 Add a "suspend" subcommand of the "continue" command to suspend gdb
1166 while continuing execution of the subprocess. Useful when you are
1167 debugging servers and you want to dodge out and initiate a connection
1168 to a server running under gdb.
1174 Modify the handling of symbols grouped through BINCL/EINCL stabs to
1175 allocate a partial symtab for each BINCL/EINCL grouping. This will
1176 seriously decrease the size of inter-psymtab dependencies and hence
1177 lessen the amount that needs to be read in when a new source file is
1184 Modify gdb to work correctly with Pascal.
1188 Add a command for searching memory, a la adb. It specifies size,
1189 mask, value, start address. ADB searches until it finds it or hits
1190 an error (or is interrupted).
1194 Remove the range and type checking code and documentation, if not