gdb/TODO

   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.
   5
   6
   7                         GDB 5.1 - Fixes
   8                         ===============
   9
  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.
  12
  13 -- 2001-03-08
  14
  15 Update GDB's coding standard documentation.  Known topics:
  16
  17        o     alloca/malloc et.al.
  18
  19        o     typedef and structs
  20
  21        o        ISO-C
  22
  23 and most likely also:
  24
  25        o         include conventions
  26
  27 --
  28
  29 Wow, three bug reports for the same problem in one day!  We should
  30 probably make fixing this a real priority :-).
  31
  32 Anyway, thanks for reporting.
  33
  34 The following patch will fix the problems with setting breakpoints in
  35 dynamically loaded objects:
  36
  37    http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00230.html
  38
  39 This patch isn't checked in yet (ping Michael/JimB), but I hope this
  40 will be in the next GDB release.
  41
  42 There should really be a test in the testsuite for this problem, since
  43 it keeps coming up :-(.  Any volunteers?
  44
  45 Mark
  46
  47 --
  48
  49 x86 linux GDB and SIGALRM (???)
  50 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00803.html
  51
  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
  55
  56 Mark
  57
  58 [The test has been submitted for approval - cagney]
  59
  60 --
  61
  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
  64
  65 GDB misses watchpoint triggers after proceeding over a breakpoint on
  66 x86 targets.
  67
  68 --
  69
  70 GDB 5.0 doesn't work on Linux/SPARC
  71
  72 There are two parts to this.
  73
  74       o   GDB 5.0 doesn't work on GNU/Linux/SPARC32
  75
  76       o   GDB 5.0 doesn't work on the new target
  77           GNU/Linux/SPARC64
  78
  79 GDB does build on both these targets.
  80
  81 The first problem is the one that should be fixed.
  82
  83 --
  84
  85                 GDB 5.1 - New features
  86                 ======================
  87
  88 The following new features should be included in 5.1.
  89
  90 --
  91
  92 Enable MI by default.  Old code can be deleted after 5.1 is out.
  93
  94 Issues:
  95
  96         o       syntax change where a list would
  97                 look like:
  98                      [ foo=a, foo=b, foo=c ]
  99                 instead of
 100                      { foo=a, foo=b, foo=c }
 101
 102         o       kill off the idea of a reverse
 103                 query.
 104
 105         o       review test cases
 106
 107         o       enable it
 108
 109 --
 110
 111 Pascal (Pierre Muller, David Taylor)
 112
 113 Pierre Muller has contributed patches for adding Pascal Language
 114 support to GDB.
 115
 116 2 pascal language patches inserted in database
 117 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
 118
 119 Indent -gnu ?
 120 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
 121
 122 [I think this has been merged, need to confirm - cagney]
 123
 124 --
 125
 126 Java (Anthony Green, David Taylor)
 127
 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.
 131
 132 Patch: java tests
 133 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
 134
 135 Patch: java booleans
 136 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
 137
 138 Patch: handle N_MAIN stab
 139 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
 140
 141 -- 2001-03-08
 142
 143 Add CRIS target.
 144
 145 A predicate to this is the multi-arching of SOFTWARE_SINGLE_STEP().  A
 146 patch has been submitted.
 147
 148 --
 149
 150                 GDB 5.1 - Cleanups
 151                 ==================
 152
 153 The following code cleanups will hopefully be applied to GDB 5.1.
 154
 155 -- 2001-03-26
 156
 157 Resolve the build status of all broken targets as identified by the
 158 MAINTAINERS file.
 159
 160         o       arm-* vs NetBSD's lack of ``unix''
 161         o       arm-* vs IRIX (see below)
 162
 163         o       delete mpw?
 164
 165 -- 2001-03-15
 166
 167                         Obsolete some targets.
 168
 169 Possible selection criteria are:
 170
 171         o       uses a deprecated feature
 172
 173         o       doesn't build
 174
 175         o       doesn't have a maintainer
 176
 177 Steps:
 178
 179         o       post proposals to gdb@ (DONE)
 180
 181         o       post announcement to gdb-announce@
 182                 crossed with gdb@ reply-to to gdb@
 183                 (DONE)
 184
 185                         ns32k-*-mach3*
 186                         ns32k-umax-*
 187                         ns32k-utek-sysv*
 188                         tic80-*
 189                         m68*-isi-*
 190                         m68*-sony-*
 191                         m68*-rom68k-*
 192                         m68*-*bug-*
 193                         m68*-monitor-*
 194                         m68*-est-*
 195                         a29k-ultra3
 196                         powerpcle-*-solaris*
 197                         powerpcle-*-cygwin*
 198                         powerpc-*-netware*
 199                         w65-*-*
 200                         i[3456]86-*-sunos*
 201
 202         o       clobber the files:
 203
 204                         configure.{in,host,tgt}
 205                         Makefile.in
 206                         *-tdep.c *-nat.c *-xdep.c
 207                         configure/*/*
 208
 209         o       update NEWS
 210
 211 --
 212
 213 Fix copyright notices.
 214
 215 Turns out that ``1998-2000'' isn't considered valid :-(
 216
 217 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
 218
 219 --
 220
 221                 GDB 5.1 - Known Problems
 222                 ========================
 223
 224 --
 225
 226 z8k
 227
 228 The z8k has suffered bit rot and is known to not build.  The problem
 229 was occuring in the opcodes directory.
 230
 231 --
 232
 233 m88k
 234
 235 The m88k has suffered bit rot and is known to not build.
 236
 237 --
 238
 239 Solaris 8 x86 CURSES_H problem
 240 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
 241
 242 The original problem was worked around with:
 243
 244     2000-06-06  Michael Snyder  <msnyder@cygnus.com>
 245
 246         * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
 247         * configure: Regenerate.
 248
 249 When building both GDB and SID using the same source tree the problem
 250 will still occure. sid/component/configure.in mis-configures
 251 <curses.h> and leaves wrong information in the config cache.
 252
 253 --
 254
 255                 GDB 5.2 - Fixes
 256                 ===============
 257
 258 --
 259
 260 Thread support.  Right now, as soon as a thread finishes and exits,
 261 you're hosed.  This problem is reported once a week or so.
 262
 263 --
 264
 265                 GDB 5.2 - New features
 266                 ======================
 267
 268 --
 269
 270 GCC 3.0 ABI support (but hopefully sooner...).
 271
 272 --
 273
 274 Objective C/C++ support (but hopefully sooner...).
 275
 276 --
 277
 278                 GDB 5.2 - Cleanups
 279                 ==================
 280
 281 The following cleanups have been identified as part of GDB 5.2.
 282
 283 --
 284
 285 Remove old code that does not use ui_out functions and all the related
 286 "ifdef"s.  This also allows the elimination of -DUI_OUT from
 287 Makefile.in and configure.in.
 288
 289 --
 290
 291 Compiler warnings.
 292
 293 Eliminate all warnings for at least one host/target for the flags:
 294 -Wimplicit -Wreturn-type -Wcomment -Wtrigraphs -Wformat -Wparentheses
 295 -Wpointer-arith -Wuninitialized
 296
 297 --
 298
 299 Deprecate, if not delete, the following:
 300
 301         register[]
 302         register_valid[]
 303         register_buffer()
 304         REGISTER_BYTE()
 305                 Replaced by, on the target side
 306                   supply_register()
 307                 and on core-gdb side:
 308                   {read,write}_register_gen()
 309                 Remote.c will need to use something
 310                 other than REGISTER_BYTE() and
 311                 REGISTER_RAW_SIZE() when unpacking
 312                 [gG] packets.
 313
 314         STORE_PSEUDO_REGISTER
 315         FETCH_PSEUDO_REGISTER
 316                 Now handed by the methods
 317                   gdbarch_{read,write}_register()
 318                 which sits between core GDB and
 319                 the register cache.
 320
 321         REGISTER_CONVERTIBLE
 322         REGISTER_CONVERT_TO_RAW
 323         REGISTER_CONVERT_TO_VIRTUAL
 324                 I think these three are redundant.
 325                 gdbarch_register_{read,write} can
 326                 do any conversion it likes.
 327
 328         REGISTER_VIRTUAL_SIZE
 329         MAX_REGISTER_VIRTUAL_SIZE
 330         REGISTER_VIRTUAL_TYPE
 331                 I think these can be replaced by
 332                 the pair:
 333                   FRAME_REGISTER_TYPE(frame, regnum)
 334                   REGISTER_TYPE(regnum)
 335
 336         DO_REGISTERS_INFO
 337                 Replace with
 338                  FRAME_REGISTER_INFO (frame, ...)
 339
 340         REGISTER_SIM_REGNO()
 341                 If nothing else rename this so that
 342                 how it relates to rawreg and the
 343                 regnum is clear.
 344
 345         REGISTER_BYTES
 346                 The size of the cache can be computed
 347                 on the fly.
 348
 349 --
 350
 351 Restructure gdb directory tree so that it avoids any 8.3 and 14
 352 filename problems.
 353
 354 --
 355
 356 Convert GDB build process to AUTOMAKE.
 357
 358 See also sub-directory configure below.
 359
 360 The current convention is (kind of) to use $(<header>_h) in all
 361 dependency lists.  It isn't done in a consistent way.
 362
 363 --
 364
 365                 GDB 5.2 - Known Problems
 366                 ========================
 367
 368 --
 369
 370                 Code Cleanups: General
 371                 ======================
 372
 373 The following are more general cleanups and fixes.  They are not tied
 374 to any specific release.
 375
 376 --
 377
 378 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
 379
 380 --
 381
 382 Can't build IRIX -> arm GDB.
 383 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
 384
 385 David Whedon writes:
 386 > Now I'm building for an embedded arm target.  If there is a way of turning
 387 > remote-rdi off, I couldn't find it. It looks like it gets built by default
 388 > in gdb/configure.tgt(line 58) Anyway, the build dies in
 389 > gdb/rdi-share/unixcomm.c.  SERPORT1 et. al. never get defined because we
 390 > aren't one of the architectures supported.
 391
 392 --
 393
 394 Problem with weak functions
 395 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
 396
 397 Dan Nicolaescu writes:
 398 > It seems that gdb-4.95.1  does not display correctly the function when
 399 > stoping in weak functions.
 400 >
 401 > It stops in a function that is defined as weak, not in the function
 402 > that is actually run...
 403
 404 --
 405
 406 Follow through `make check' with --enable-shared.
 407
 408 When the srcware tree is configured with --enable-shared, the `expect'
 409 program won't run properly.  Jim Wilson found out gdb has a local hack
 410 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
 411 similarly.
 412
 413 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
 414
 415 --
 416
 417 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
 418
 419 Patches in the database.
 420
 421 --
 422
 423 printcmd.c (print_address_numeric):
 424
 425 NOTE: This assumes that the significant address information is kept in
 426 the least significant bits of ADDR - the upper bits were either zero
 427 or sign extended.  Should ADDRESS_TO_POINTER() or some
 428 ADDRESS_TO_PRINTABLE() be used to do the conversion?
 429
 430 --
 431
 432 The BFD directory requires bug-fixed AUTOMAKE et.al.
 433
 434 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable.  It
 435 contained the full path to texinfo.tex when it should have only
 436 contained the directory.  The bug has been fixed in the current
 437 AUTOMAKE sources.  Automake snapshots can be found in:
 438         ftp://sourceware.cygnus.com/pub/gdb/snapshots
 439 and     ftp://sourceware.cygnus.com/pub/binutils
 440
 441 --
 442
 443 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
 444 determine the default isa/byte-order.
 445
 446 --
 447
 448 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
 449 BIG_ENDIAN and LITTLE_ENDIAN.
 450
 451 --
 452
 453 Eliminate more compiler warnings.
 454
 455 Of course there also needs to be the usual debate over which warnings
 456 are valid and how to best go about this.
 457
 458 One method: choose a single option; get agreement that it is
 459 reasonable; try it out to see if there isn't anything silly about it
 460 (-Wunused-parameters is an example of that) then incrementally hack
 461 away.
 462
 463 The other method is to enable all warnings and eliminate them from one
 464 file at a time.
 465
 466 --
 467
 468 Elimination of ``(catch_errors_ftype *) func''.
 469
 470 Like make_cleanup_func it isn't portable.
 471 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
 472 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
 473
 474 --
 475
 476 Nuke #define CONST_PTR.
 477
 478 --
 479
 480 Nuke USG define.
 481
 482 --
 483
 484 [PATCH/5] src/intl/Makefile.in:distclean additions
 485 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
 486
 487 Do not forget to merge the patch back into the trunk.
 488
 489 --
 490
 491 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
 492
 493 At present defs.h includes <endian.h> (which is linux specific) yet
 494 almost nothing depends on it.  Suggest "gdb_endian.h" which can also
 495 handle <machine/endian.h> and only include that where it is really
 496 needed.
 497
 498 --
 499
 500 Replace savestring() with something from libiberty.
 501
 502 An xstrldup()? but that would have different semantics.
 503
 504 --
 505
 506 Rationalize use of floatformat_unknown in GDB sources.
 507
 508 Instead of defaulting to floatformat_unknown, should hosts/targets
 509 specify the value explicitly?
 510
 511 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
 512
 513 --
 514
 515 Add a ``name'' member to include/floatformat.h:struct floatformat.
 516 Print that name in gdbarch.c.
 517
 518 --
 519
 520 Sort out the harris mess in include/floatformat.h (it hardwires two
 521 different floating point formats).
 522
 523 --
 524
 525 See of the GDB local floatformat_do_doublest() and libiberty's
 526 floatformat_to_double (which was once GDB's ...) can be merged some
 527 how.
 528
 529 --
 530
 531 Eliminate mmalloc(), mstrsave() et.al. from GDB.
 532
 533 Also eliminate it from defs.h.
 534
 535 --
 536
 537 Eliminate PTR.  ISO-C allows ``void *''.
 538
 539 --
 540
 541 Eliminate abort ().
 542
 543 GDB should never abort.  GDB should either throw ``error ()'' or
 544 ``internal_error ()''.  Better still GDB should naturally unwind with
 545 an error status.
 546
 547 --
 548
 549 GDB probably doesn't build on FreeBSD pre 2.2.x
 550 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
 551
 552 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
 553 suffer bit rot.
 554
 555 --
 556
 557 Deprecate "fg".  Apparently ``fg'' is actually continue.
 558
 559 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
 560
 561 --
 562
 563 Deprecate current use of ``floatformat_unknown''.
 564
 565 Require all targets to explicitly provide their float format instead
 566 of defaulting to floatformat unknown.  Doing the latter leads to nasty
 567 bugs.
 568
 569 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
 570
 571 --
 572
 573 Rationalize floatformat_to_double() vs floatformat_to_doublest().
 574
 575 Looks like GDB migrated floatformat_to_double() to libiberty but then
 576 turned around and created a ..._to_doublest() the latter containing
 577 several bug fixes.
 578
 579 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
 580
 581 --
 582
 583 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
 584
 585 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
 586
 587 --
 588
 589 The ``maintenance deprecate set endian big'' command doesn't notice
 590 that it is deprecating ``set endian'' and not ``set endian big'' (big
 591 is implemented using an enum).  Is anyone going to notice this?
 592
 593 --
 594
 595 When tab expanding something like ``set arch<tab>'' ignore the
 596 deprecated ``set archdebug'' and expand to ``set architecture''.
 597
 598 --
 599
 600 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
 601 like from arm-tdep.c.
 602
 603 --
 604
 605 Fix uses of ->function.cfunc = set_function().
 606
 607 The command.c code calls sfunc() when a set command.  Rather than
 608 change it suggest fixing the callback function so that it is more
 609 useful.  See:
 610
 611 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
 612
 613 See also ``Fix implementation of  ``target xxx''.'' below.
 614
 615 --
 616
 617 IRIX 3.x support is probably broken.
 618
 619 --
 620
 621 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
 622 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
 623
 624 Apart from the d30v, are there any sim/common simulators that make use
 625 of this?
 626
 627 A brief summary of what happened is that sim/common/sim-break.c was
 628 created as a good idea.  It turned out a better idea was to use
 629 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
 630
 631 --
 632
 633 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
 634 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
 635
 636 --
 637
 638 Eliminate ``extern'' from C files.
 639
 640 --
 641
 642 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
 643
 644 Extreme care is recommeded - perhaps only modify tests that are
 645 exercised by the testsuite (as determined using some type of code
 646 coverage analysis).
 647
 648 --
 649
 650 Replace the file gdb/CONTRIBUTE with a file that is generated from the
 651 gdb/doc/*.texinfo directory.
 652
 653 --
 654
 655                         New Features and Fixes
 656                         ======================
 657
 658 These are harder than cleanups but easier than work involving
 659 fundamental architectural change.
 660
 661 --
 662
 663 Hardware watchpoint problems on x86 OSes, including Linux:
 664
 665 1. Delete/disable hardware watchpoints should free hardware debug
 666 registers.
 667 2. Watch for different values on a viariable with one hardware debug
 668 register.
 669
 670 According to Eli Zaretskii <eliz@delorie.com>:
 671
 672 These are not GDB/ia32 issues per se: the above features are all
 673 implemented in the DJGPP port of GDB and work in v5.0.  Every
 674 x86-based target should be able to lift the relevant parts of
 675 go32-nat.c and use them almost verbatim.  You get debug register
 676 sharing through reference counts, and the ability to watch large
 677 regions (up to 16 bytes) using multiple registers.  (The required
 678 infrastructure in high-level GDB application code, mostly in
 679 breakpoint.c, is also working since v5.0.)
 680
 681 --
 682
 683 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
 684 so that you can see how the GDB was created.
 685
 686 --
 687
 688 Add an "info bfd" command that displays supported object formats,
 689 similarly to objdump -i.
 690
 691 Is there a command already?
 692
 693 --
 694
 695 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
 696
 697 This requires internationalization.
 698
 699 --
 700
 701 Add support for:
 702
 703 (gdb) p fwprintf(stdout,L"%S\n", f)
 704 No symbol "L" in current context.
 705
 706 --
 707
 708 Cleanup configury support for optional sub-directories.
 709
 710 Check how GCC handles multiple front ends for an example of how things
 711 could work.  A tentative first step is to rationalize things so that
 712 all sub directories are handled in a fashion similar to gdb/mi.
 713
 714 See also automake above.
 715
 716 --
 717
 718 Add a transcript mechanism to GDB.
 719
 720 Such a mechanism might log all gdb input and output to a file in a
 721 form that would allow it to be replayed.  It could involve ``gdb
 722 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
 723
 724 --
 725
 726 Can the xdep files be replaced by autoconf?
 727
 728 --
 729
 730 Document trace machinery
 731
 732 --
 733
 734 Document ui-out and ui-file.
 735
 736 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
 737
 738 --
 739
 740 Update texinfo.tex to latest?
 741
 742 --
 743
 744 Incorporate agentexpr.texi into gdb.texinfo
 745
 746 agentexpr.texi mostly describes the details of the byte code used for
 747 tracepoints, not the internals of the support for this in GDB.  So it
 748 looks like gdb.texinfo is a better place for this information.
 749
 750 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
 751
 752 --
 753
 754 Document overlay machinery.
 755
 756 --
 757
 758 ``(gdb) catch signal SIGNAL''
 759
 760 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
 761 You can attach commands to a catch but not a handle.  A handle has a
 762 limited number of hardwired actions.
 763
 764 --
 765
 766 Fix TUI
 767
 768      o  readline/*.h bitrot
 769
 770         The TUI isn't up-to-date with
 771         respect to the readline currently
 772         bundled with GDB.  Importing a
 773         new readline is on the 5.1 wish
 774         list so this can only get worse.
 775
 776         Grep for things like term_cursor_move.
 777
 778         (To be honest, I don't see anyone
 779         importing a new readline before 5.1 is
 780         out)
 781
 782      o  tui.c:va_catch_errors() bitrot
 783
 784         This nasty piece of work used knowledge
 785         of the internals of GDBs error functions :-(
 786         Ever since those internals were cleaned
 787         up this code has been broken. :-(
 788
 789      o  tuiWin.c:c_makeVisibleWithNewHeight() broken
 790         tuiLayout.c:_extractDisplayStartAddr() broken
 791
 792         Both these function call find_line_pc()
 793         incorrectly (wrong args, wrong return value).
 794
 795         I suspect this bug has always been there!
 796         It had been hidden because those files
 797         didn't include the necessary header files
 798         from gdb proper :-(
 799
 800      o  tuiRegs() host dependant
 801
 802         Not suprisingly, this isn't a very portable
 803         section of code.  However, I'm sure people
 804         could live with no regs in the short to
 805         medium term.
 806
 807      o  defs.h: #include "tui.h" et.al.
 808
 809         I'm not sure where this came from.
 810         It was a really bad idea.
 811
 812         To get things to compile I did a nasty
 813         hack (Just declare what was needed and
 814         replace any expressions like xx->y.z()
 815         in GDB proper with function calls).  I
 816         could commit it slightly cleaned up if
 817         you like.
 818
 819         Medium Term. the #ifdef TUI and TuiDo()
 820         should be changed to hooks (like GDBTK).
 821         The gdb-events.[hc] is there for that
 822         purpose (1)
 823
 824      o  tui.c:_tuiReset() host dependant
 825
 826         tui.c contains a lump of termio[s]
 827         I suspect an equivalent block of
 828         code can be lifted from readline.
 829         An equivalent readline function may
 830         even be available.
 831
 832      o  curses.h vs ncurses.h.
 833
 834         Simple portability problem.
 835
 836      o  subsetCompare()
 837
 838         This function is a mystery - where is it?
 839
 840      o  tui-file.[hc] cleanup
 841
 842         This can be significantly simplified.
 843
 844      o  The code should be pacified. (-Werror -W...)
 845
 846         There are plenty of #includes,
 847         duplicate #includes, missing function decls
 848         and the like.
 849
 850         Some of the problems I found were through
 851         fixing a few of the warnings.
 852
 853      o  The code should be GNUtified.
 854
 855         It would be very nice to have this code
 856         look like the rest of GDB.  That way people
 857         would be more accepting of it as a true
 858         gdb component.
 859
 860         Until it is GNUtified it is going to stick
 861         out like a sore thumb to the programmer.
 862
 863      o  The code should be clearly copyrighted
 864
 865         (FSF, with due credit to HP)
 866
 867 --
 868
 869 Add support for ``gdb --- PROGRAM ARGS ...''.
 870 Add support for ``gdb -cmd=...''
 871
 872 Along with many variations.  Check:
 873
 874 ????? for a full discussion.
 875
 876 for a discussion.
 877
 878 --
 879
 880 Implement ``(gdb) !ls''.
 881
 882 Which is very different from ``(gdb) ! ls''.  Implementing the latter
 883 is trivial.
 884
 885 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
 886
 887 --
 888
 889 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
 890 be passed.
 891
 892 --
 893
 894 When tab expanding something like ``set arch<tab>'' ignore the
 895 deprecated ``set archdebug'' and expand to ``set architecture''.
 896
 897 --
 898
 899 Replace the code that uses the host FPU with an emulator of the target
 900 FPU.
 901
 902 --
 903
 904 The "ocd reset" command needs to flush the dcache, which requires breaking
 905 the abstraction layer between the target independent and target code.  One
 906 way to address this is provide a generic "reset" command and target vector.
 907
 908 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
 909
 910 --
 911
 912                         Thread Support
 913                         ==============
 914
 915 --
 916
 917 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
 918 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
 919
 920 The thread_db assisted debugging code doesn't handle exiting threads
 921 properly, at least in combination with glibc 2.1.3 (the framework is
 922 there, just not the actual code).  There are at least two problems
 923 that prevent this from working.
 924
 925 As an additional reference point, the pre thread_db code did not work
 926 either.
 927
 928 --
 929
 930 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
 931 Solaris/x86).
 932 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
 933
 934 Christopher Blizzard writes:
 935
 936 So, I've done some more digging into this and it looks like Jim
 937 Kingdon has reported this problem in the past:
 938
 939 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
 940
 941 I can reproduce this problem both with and without Tom's patch.  Has
 942 anyone seen this before?  Maybe have a solution for it hanging around?
 943 :)
 944
 945 There's a test case for this documented at:
 946
 947 when debugging threaded applications you get extra SIGTRAPs
 948 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
 949
 950 [There should be a GDB testcase - cagney]
 951
 952 --
 953
 954 GDB5 TOT on unixware 7
 955 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
 956
 957 Robert Lipe writes:
 958 > I just spun the top of tree of the GDB5 branch on UnixWare 7.  As a
 959 > practical matter, the current thread support is somewhat more annoying
 960 > than when GDB was thread-unaware.
 961
 962 --
 963
 964                         Language Support
 965                         ================
 966
 967 New languages come onto the scene all the time.
 968
 969 --
 970
 971 Re: Various C++ things
 972
 973 value_headof/value_from_vtable_info are worthless, and should be
 974 removed.  The one place in printcmd.c that uses it should use the RTTI
 975 functions.
 976
 977 RTTI for g++ should be using the typeinfo functions rather than the
 978 vtables.  The typeinfo functions are always at offset 4 from the
 979 beginning of the vtable, and are always right. The vtables will have
 980 weird names like E::VB sometimes.  The typeinfo function will always
 981 be "E type_info function", or somesuch.
 982
 983 value_virtual_fn_field needs to be fixed so there are no failures for
 984 virtual functions for C++ using g++.
 985
 986 Testsuite cases are the major priority right now for C++ support,
 987 since i have to make a lot of changes that could potentially break
 988 each other.
 989
 990 --
 991
 992 Add support for Modula3
 993
 994 Get DEC/Compaq to contribute their Modula-3 support.
 995
 996 --
 997
 998                 Remote Protocol Support
 999                 =======================
1000
1001 --
1002
1003 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
1004 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
1005
1006 (Broken) support for GDB's remote protocol across UDP is to be
1007 included in the follow-on release.
1008
1009 It should be noted that UDP can only work when the [Gg] packet fits in
1010 a single UDP packet.
1011
1012 There is also much debate over the merit of this.
1013
1014 --
1015
1016 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
1017
1018 Add support for packet enable/disable commands with these thread
1019 packets.  General cleanup.
1020
1021 [PATCH] Document the ThreadInfo remote protocol queries
1022 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
1023
1024 [PATCH] "info threads" queries for remote.c
1025 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
1026
1027 --
1028
1029 Remote protocol doco feedback.
1030
1031 Too much feedback to mention needs to be merged in (901660).  Search
1032 for the word ``remote''.
1033
1034
1035 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
1036 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
1037 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
1038
1039 --
1040
1041 GDB doesn't recover gracefully from remote protocol errors.
1042
1043 GDB wasn't checking for NAKs from the remote target.  Instead a NAK is
1044 ignored and a timeout is required before GDB retries.  A pre-cursor to
1045 fixing this this is making GDB's remote protocol packet more robust.
1046
1047 While downloading to a remote protocol target, gdb ignores packet
1048 errors in so far as it will continue to download with chunk N+1 even
1049 if chunk N was not correctly sent.  This causes gdb.base/remote.exp to
1050 take a painfully long time to run.  As a PS that test needs to be
1051 fixed so that it builds on 16 bit machines.
1052
1053 --
1054
1055 Fix the ``!'' packet.
1056
1057 JT reported that the existing targets do, in fact return ``OK'' so it
1058 is possible to merge remote and extended-remote targets.
1059
1060 --
1061
1062 Drop ``<address>'' from the [SsCc] packets.
1063
1064 I don't think that GDB generates them so having it in the protocol is
1065 silly.
1066
1067 --
1068
1069 Fix doco on the ``q'' packet.
1070
1071 It has evolved into a generic RPC.  The notes should reflect this and,
1072 perhaps, the ``Q'' packet can be deprecated.
1073
1074 The doco should mention that ``OK'' is a valid packet response.
1075
1076 The doco should explain why ``OK'' needs to be a valid packet
1077 response.
1078
1079 --
1080
1081 Add the cycle step command.
1082
1083 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1084
1085 --
1086
1087 Resolve how to scale things to support very large packets.
1088
1089 --
1090
1091 Resolve how to handle a target that changes things like its endianess
1092 on the fly - should it be returned in the ``T'' packet?
1093
1094 Underlying problem is that the register file is target endian.  If the
1095 target endianess changes gdb doesn't know.
1096
1097 --
1098
1099                         Symbol Support
1100                         ==============
1101
1102 If / when GDB starts to support the debugging of multi-processor
1103 (rather than multi-thread) applications the symtab code will need to
1104 be updated a little so that several independent symbol tables are
1105 active at a given time.
1106
1107 The other interesting change is a clarification of the exact meaning
1108 of CORE_ADDR and that has had consequences for a few targets (that
1109 were abusing that data type).
1110
1111 --
1112
1113 Investiagate ways of reducing memory.
1114
1115 --
1116
1117 Investigate ways of improving load time.
1118
1119 --
1120
1121 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1122
1123 Consequence of recent symtab clarification.  No marks for figuring out
1124 who maintains the d10v.
1125
1126 --
1127
1128 Get the MIPS to correctly sign extend all address <-> pointer
1129 conversions.
1130
1131 Consequence of recent symtab clarification.  No marks for figuring out
1132 who maintains the MIPS.
1133
1134 --
1135
1136 GDB truncates 64 bit enums.
1137
1138 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1139
1140 --
1141
1142                         Testsuite Support
1143                         =================
1144
1145 There are never to many testcases.
1146
1147 --
1148
1149 Better thread testsuite.
1150
1151 --
1152
1153 Better C++ testsuite.
1154
1155 --
1156
1157 Look at adding a GDB specific testsuite directory so that white box
1158 tests of key internals can be added (eg ui_file).
1159
1160 --
1161
1162 Separate out tests that involve the floating point (FP).
1163
1164 (Something for people brining up new targets).  FP and non-fp tests
1165 are combined.  I think there should be set of basic tests that
1166 exercise pure integer support and then a more expanded set that
1167 exercise FP and FP/integer interactions.
1168
1169 As an example, the MIPS, for n32 as problems with passing FP's and
1170 structs.  Since most inferior call tests include FP it is difficult to
1171 determine of the integer tests are ok.
1172
1173 --
1174
1175                 Architectural Changes: General
1176                 ==============================
1177
1178 These are harder than simple cleanups / fixes and, consequently
1179 involve more work.  Typically an Architectural Change will be broken
1180 down into a more digestible set of cleanups and fixes.
1181
1182 --
1183
1184 Cleanup software single step.
1185
1186 At present many targets implement software single step by directly
1187 blatting memory (see rs6000-tdep.c).  Those targets should register
1188 the applicable breakpoints using the breakpoint framework.  Perhaphs a
1189 new internal breakpoint class ``step'' is needed.
1190
1191 --
1192
1193 Replace READ_FP() with FRAME_HANDLE().
1194
1195 READ_FP() is a hangover from the days of the vax when the ABI really
1196 did have a frame pointer register.  Modern architectures typically
1197 construct a virtual frame-handle from the stack pointer and various
1198 other bits of string.
1199
1200 Unfortunately GDB still treats this synthetic FP register as though it
1201 is real.  That in turn really confuses users (arm and ``print $fp'' VS
1202 ``info registers fp'').  The synthetic FP should be separated out of
1203 the true register set presented to the user.
1204
1205 --
1206
1207 Register Cache Cleanup (below from Andrew Cagney)
1208
1209 I would depict the current register architecture as something like:
1210
1211         High GDB  -->   Low GDB
1212             |              |
1213            \|/            \|/
1214             --- REG NR -----
1215                    |
1216     register + REGISTER_BYTE(reg_nr)
1217                    |
1218                   \|/
1219         -------------------------
1220         |  extern register[]    |
1221         -------------------------
1222
1223 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1224 really clear on what mechanisms they should be using to manipulate that
1225 buffer.   Further, much code assumes, dangerously, that registers are
1226 contigious.  Having got mips-tdep.c to support multiple ABIs, believe
1227 me, that is a bad assumption.  Finally, that register cache layout is
1228 determined by the current remote/local target and _not_ the less
1229 specific target ISA.  In fact, in many cases it is determined by the
1230 somewhat arbitrary layout of the [gG] packets!
1231
1232
1233 How I would like the register file to work is more like:
1234
1235
1236           High GDB
1237              |
1238             \|/
1239        pseudo reg-nr
1240              |
1241         map pseudo <->
1242         random cache
1243         bytes
1244              |
1245             \|/
1246         ------------
1247         | register |
1248         |  cache   |
1249         ------------
1250             /|\
1251              |
1252         map random cache
1253         bytes to target
1254         dependent i-face
1255             /|\
1256              |
1257        target dependent
1258         such as [gG] packet
1259         or ptrace buffer
1260
1261 The main objectives being:
1262
1263         o       a clear separation between the low
1264                 level target and the high level GDB
1265
1266         o       a mechanism that solves the general
1267                 problem of register aliases, overlaps
1268                 etc instead of treating them as optional
1269                 extras that can be wedged in as an after
1270                 thought (that is a reasonable description
1271                 of the current code).
1272
1273                 Identify then solve the hard case and the
1274                 rest just falls out.  GDB solved the easy
1275                 case and then tried to ignore the real
1276                 world :-)
1277
1278         o       a removal of the assumption that the
1279                 mapping between the register cache
1280                 and virtual registers is largely static.
1281                 If you flip the USR/SSR stack register
1282                 select bit in the status-register then
1283                 the corresponding stack registers should
1284                 reflect the change.
1285
1286         o       a mechanism that clearly separates the
1287                 gdb internal register cache from any
1288                 target (not architecture) dependent
1289                 specifics such as [gG] packets.
1290
1291 Of course, like anything, it sounds good in theory.  In reality, it
1292 would have to contend with many<->many relationships at both the
1293 virt<->cache and cache<->target level.  For instance:
1294
1295         virt<->cache
1296                 Modifying an mmx register may involve
1297                 scattering values across both FP and
1298                 mmpx specific parts of a buffer
1299
1300         cache<->target
1301                 When writing back a SP it may need to
1302                 both be written to both SP and USP.
1303
1304
1305 Hmm,
1306
1307 Rather than let this like the last time it was discussed, just slip, I'm
1308 first going to add this e-mail (+ references) to TODO.  I'd then like to
1309 sketch out a broad strategy I think could get us there.
1310
1311
1312 First thing I'd suggest is separating out the ``extern registers[]''
1313 code so that we can at least identify what is using it.  At present
1314 things are scattered across many files.  That way we can at least
1315 pretend that there is a cache instead of a global array :-)
1316
1317 I'd then suggest someone putting up a proposal for the pseudo-reg /
1318 high-level side interface so that code can be adopted to it.  For old
1319 code, initially a blanket rename of write_register_bytes() to
1320 deprecated_write_register_bytes() would help.
1321
1322 Following that would, finaly be the corresponding changes to the target.
1323
1324 --
1325
1326 Check that GDB can handle all BFD architectures (Andrew Cagney)
1327
1328 There should be a test that checks that BFD/GDB are in sync with
1329 regard to architecture changes.  Something like a test that first
1330 queries GDB for all supported architectures and then feeds each back
1331 to GDB..  Anyone interested in learning how to write tests?  :-)
1332
1333 --
1334
1335                 Architectural Change: Multi-arch et al.
1336                 =======================================
1337
1338 The long term objective is to remove all assumptions that there is a
1339 single target with a single address space with a single instruction
1340 set architecture and single application binary interface.
1341
1342 This is an ongoing effort.  The first milestone is to enable
1343 ``multi-arch'' where by all architectural decisions are made at
1344 runtime.
1345
1346 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1347 ``gdbisa''.  Once things are multi-arched breaking that down correctly
1348 will become much easier.
1349
1350 --
1351
1352 GDBARCH cleanup (Andrew Cagney)
1353
1354 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1355 into arch-utils.[hc].
1356
1357 Document that gdbarch_init_ftype could easily fail because it didn't
1358 identify an architecture.
1359
1360 --
1361
1362 Fix BELIEVE_PPC_PROMOTION.  Change it to BELIEVE_PPC_PROMOTION_P?
1363
1364 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1365 symtab file.
1366
1367 --
1368
1369 Fix target_signal_from_host() etc.
1370
1371 The name is wrong for starters.  ``target_signal'' should probably be
1372 ``gdb_signal''.  ``from_host'' should be ``from_target_signal''.
1373 After that it needs to be multi-arched and made independent of any
1374 host signal numbering.
1375
1376 --
1377
1378 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1379 EXTRA_FRAME_INFO.
1380
1381 This is a barrier to replacing mips_extra_func_info with something
1382 that works with multi-arch.
1383
1384 --
1385
1386 Multi-arch  mips_extra_func_info.
1387
1388 This first needs the alpha to be updated so that it uses ``struct
1389 frame_extra_info''.
1390
1391 --
1392
1393 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1394
1395 Surely one of them is redundant.
1396
1397 --
1398
1399 Convert ALL architectures to MULTI-ARCH.
1400
1401 --
1402
1403 Select the initial multi-arch ISA / ABI based on --target or similar.
1404
1405 At present the default is based on what ever is first in the BFD
1406 archures table.  It should be determined based on the ``--target=...''
1407 name.
1408
1409 --
1410
1411 Make MIPS pure multi-arch.
1412
1413 It is only at the multi-arch enabled stage.
1414
1415 --
1416
1417 Truly multi-arch.
1418
1419 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1420
1421 Can the tm.h and nm.h files be eliminated by multi-arch.
1422
1423 --
1424
1425         Architectural Change: MI, LIBGDB and scripting languages
1426         ========================================================
1427
1428 See also architectural changes related to the event loop.  LIBGDB
1429 can't be finished until there is a generic event loop being used by
1430 all targets.
1431
1432 The long term objective is it to be possible to integrate GDB into
1433 scripting languages.
1434
1435 --
1436
1437 Implement generic ``(gdb) commmand > file''
1438
1439 Once everything is going through ui_file it should be come fairly
1440 easy.
1441
1442 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1443
1444 --
1445
1446 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1447
1448 gdb_stdtarg is easily confused with gdb_stdarg.
1449
1450 --
1451
1452 Extra ui_file methods - dump.
1453
1454 Very useful for whitebox testing.
1455
1456 --
1457
1458 Eliminate error_begin().
1459
1460 With ui_file, there is no need for the statefull error_begin ()
1461 function.
1462
1463 --
1464
1465 Send normal output to gdb_stdout.
1466 Send error messages to gdb_stderror.
1467 Send debug and log output log gdb_stdlog.
1468
1469 GDB still contains many cases where (f)printf or printf_filtered () is
1470 used when it should be sending the messages to gdb_stderror or
1471 gdb_stdlog.  The thought of #defining printf to something has crossed
1472 peoples minds ;-)
1473
1474 --
1475
1476 Re-do GDB's output pager.
1477
1478 GDB's output pager still relies on people correctly using *_filtered
1479 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1480 Hopefully, with all normal output going to gdb_stdout, the pager can
1481 just look at the ui_file that the output is on and then use that to
1482 decide what to do about paging.  Sounds good in theory.
1483
1484 --
1485
1486 Check/cleanup MI documentation.
1487
1488 The list of commands specified in the documentation needs to be
1489 checked against the mi-cmds.c table in a mechanical way (so that they
1490 two can be kept up-to-date).
1491
1492 --
1493
1494 Convert MI into libgdb
1495
1496 MI provides a text interface into what should be many of the libgdb
1497 functions.  The implementation of those functions should be separated
1498 into the MI interface and the functions proper.  Those functions being
1499 moved to gdb/lib say.
1500
1501 --
1502
1503 Create libgdb.h
1504
1505 The first part can already be found in defs.h.
1506
1507 --
1508
1509 MI's input does not use buffering.
1510
1511 At present the MI interface reads raw characters of from an unbuffered
1512 FD.  This is to avoid several nasty buffer/race conditions.  That code
1513 should be changed so that it registers its self with the event loop
1514 (on the input FD) and then push commands up to MI as they arrive.
1515
1516 The serial code already does this.
1517
1518 --
1519
1520 Make MI interface accessible from existing CLI.
1521
1522 --
1523
1524 Add a breakpoint-edit command to MI.
1525
1526 It would be similar to MI's breakpoint create but would apply to an
1527 existing breakpoint.  It saves the need to delete/create breakpoints
1528 when ever they are changed.
1529
1530 --
1531
1532 Add directory path to MI breakpoint.
1533
1534 That way the GUI's task of finding the file within which the
1535 breakpoint was set is simplified.
1536
1537 --
1538
1539 Add a mechanism to reject certain expression classes to MI
1540
1541 There are situtations where you don't want GDB's expression
1542 parser/evaluator to perform inferior function calls or variable
1543 assignments.  A way of restricting the expression parser so that such
1544 operations are not accepted would be very helpful.
1545
1546 --
1547
1548 Remove sideffects from libgdb breakpoint create function.
1549
1550 The user can use the CLI to create a breakpoint with partial
1551 information - no file (gdb would use the file from the last
1552 breakpoint).
1553
1554 The libgdb interface currently affects that environment which can lead
1555 to confusion when a user is setting breakpoints via both the MI and
1556 the CLI.
1557
1558 This is also a good example of how getting the CLI ``right'' will be
1559 hard.
1560
1561 --
1562
1563 Move gdb_lasterr to ui_out?
1564
1565 The way GDB throws errors and records them needs a re-think.  ui_out
1566 handles the correct output well.  It doesn't resolve what to do with
1567 output / error-messages when things go wrong.
1568
1569 --
1570
1571 do_setshow_command contains a 1024 byte buffer.
1572
1573 The function assumes that there will never be any more than 1024 bytes
1574 of enum.  It should use mem_file.
1575
1576 --
1577
1578 Should struct cmd_list_element . completer take the command as an
1579 argument?
1580
1581 --
1582
1583 Should the bulk of top.c:line_completion_function() be moved to
1584 command.[hc]?  complete_on_cmdlist() and complete_on_enums() could
1585 then be made private.
1586
1587 --
1588
1589 top.c (execute_command): Should a command being valid when the target
1590 is running be made an attribute (predicate) to the command rather than
1591 an explicit set of tests.
1592
1593 --
1594
1595 top.c (execute_command): Should the bulk of this function be moved
1596 into command.[hc] so that top.c doesn't grub around in the command
1597 internals?
1598
1599 --
1600
1601                 Architectural Change: Async
1602                 ===========================
1603
1604 While GDB uses an event loop when prompting the user for input.  That
1605 event loop is not exploited by targets when they allow the target
1606 program to continue.  Typically targets still block in (target_wait())
1607 until the program again halts.
1608
1609 The closest a target comes to supporting full asynchronous mode are
1610 the remote targets ``async'' and ``extended-async''.
1611
1612 --
1613
1614 Asynchronous expression evaluator
1615
1616 Inferior function calls hang GDB.
1617
1618 --
1619
1620 Fix implementation of  ``target xxx''.
1621
1622 At present when the user specifies ``target xxxx'', the CLI maps that
1623 directly onto a target open method.  It is then assumed that the
1624 target open method should do all sorts of complicated things as this
1625 is the only chance it has.  Check how the various remote targets
1626 duplicate the target operations.  Check also how the various targets
1627 behave differently for purely arbitrary reasons.
1628
1629 What should happen is that ``target xxxx'' should call a generic
1630 ``target'' function and that should then co-ordinate the opening of
1631 ``xxxx''.  This becomes especially important when you're trying to
1632 open an asynchronous target that may need to perform background tasks
1633 as part of the ``attach'' phase.
1634
1635 Unfortunately, due to limitations in the old/creaking command.h
1636 interface, that isn't possible.  The function being called isn't told
1637 of the ``xxx'' or any other context information.
1638
1639 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1640 CLI code so that it passes to the callback function (attatched to a
1641 command) useful information such as the actual command and a context
1642 for that command.  Other changes such as making ``struct command''
1643 opaque may also help.
1644
1645 See also:
1646 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1647
1648 --
1649
1650 Make "target xxx" command interruptible.
1651
1652 As things become async this becomes possible.  A target would start
1653 the connect and then return control to the event loop.  A cntrl-c
1654 would notify the target that the operation is to be abandoned and the
1655 target code could respond.
1656
1657 --
1658
1659 Add a "suspend" subcommand of the "continue" command to suspend gdb
1660 while continuing  execution of the subprocess.  Useful when you are
1661 debugging servers and you want to dodge out and initiate a connection
1662 to a server running under gdb.
1663
1664 [hey async!!]
1665
1666 --
1667
1668                         TODO FAQ
1669                         ========
1670
1671 Frequently requested but not approved requests.
1672
1673 --
1674
1675 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1676
1677 The benefits on this one are thought to be marginal - GDBs design
1678 means that unused parameters are very common.  GCC 3.0 will also
1679 include the option -Wno-unused-parameter which means that ``-Wall
1680 -Wno-unused-parameters -Werror'' can be specified.
1681
1682 --
1683
1684
1685
1686                         Legacy Wish List
1687                         ================
1688
1689 This list is not up to date, and opinions vary about the importance or
1690 even desirability of some of the items.  If you do fix something, it
1691 always pays to check the below.
1692
1693 --
1694
1695 @c This does not work (yet if ever).  FIXME.
1696 @c @item --parse=@var{lang} @dots{}
1697 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1698 @c @samp{all} configures @value{GDBN} for all supported languages.  To get a
1699 @c list of all supported languages, omit the argument.  Without this
1700 @c option, @value{GDBN} is configured to parse all supported languages.
1701
1702 --
1703
1704 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1705 is its default value.  Clean this up.
1706
1707 --
1708
1709 It should be possible to use symbols from shared libraries before we know
1710 exactly where the libraries will be loaded.  E.g. "b perror" before running
1711 the program.  This could maybe be done as an extension of the "breakpoint
1712 re-evaluation" after new symbols are loaded.
1713
1714 --
1715
1716 Make single_step() insert and remove breakpoints in one operation.
1717
1718 [If this is talking about having single_step() insert the breakpoints,
1719 run the target then pull the breakpoints then it is wrong.  The
1720 function has to return as control has to eventually be passed back to
1721 the main event loop.]
1722
1723 --
1724
1725 Speed up single stepping by avoiding extraneous ptrace calls.
1726
1727 --
1728
1729 Speed up single stepping by not inserting and removing breakpoints
1730 each time the inferior starts and stops.
1731
1732 Breakpoints should not be inserted and deleted all the time.  Only the
1733 one(s) there should be removed when we have to step over one.  Support
1734 breakpoints that don't have to be removed to step over them.
1735
1736 [this has resulted in numerous debates.  The issue isn't clear cut]
1737
1738 --
1739
1740 Provide "voodoo" debugging of core files.  This creates a zombie
1741 process as a child of the debugger, and loads it up with the data,
1742 stack, and regs of the core file.  This allows you to call functions
1743 in the executable, to manipulate the data in the core file.
1744
1745 [you wish]
1746
1747 --
1748
1749 GDB reopens the source file on every line, as you "next" through it.
1750
1751 [still true? I've a memory of this being fixed]
1752
1753 --
1754
1755 Perhaps "i source" should take an argument like that of "list".
1756
1757 --
1758
1759 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1760 it matches the source line indicated.
1761
1762 --
1763
1764 The prompt at end of screen should accept space as well as CR.
1765
1766 --
1767
1768 Backtrace should point out what the currently selected frame is, in
1769 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1770 ...)" rather than "#3 foo (bar, ...)".
1771
1772 --
1773
1774 "i program" should work for core files, and display more info, like what
1775 actually caused it to die.
1776
1777 --
1778
1779 "x/10i" should shorten the long name, if any, on subsequent lines.
1780
1781 --
1782
1783 "next" over a function that longjumps, never stops until next time you happen
1784 to get to that spot by accident.  E.g. "n" over execute_command which has
1785 an error.
1786
1787 --
1788
1789 "set zeroprint off", don't bother printing members of structs which
1790 are entirely zero.  Useful for those big structs with few useful
1791 members.
1792
1793 --
1794
1795 GDB does four ioctl's for every command, probably switching terminal modes
1796 to/from inferior or for readline or something.
1797
1798 --
1799
1800 terminal_ours versus terminal_inferior: cache state.  Switch should be a noop
1801 if the state is the same, too.
1802
1803 --
1804
1805 "i frame" shows wrong "arglist at" location, doesn't show where the args
1806 should be found, only their actual values.
1807
1808 --
1809
1810 There should be a way for "set" commands to validate the new setting
1811 before it takes effect.
1812
1813 --
1814
1815 "ena d" is ambiguous, why?  "ena delete" seems to think it is a command!
1816
1817 --
1818
1819 i line VAR produces "Line number not known for symbol ``var''.".  I
1820 thought we were stashing that info now!
1821
1822 --
1823
1824 We should be able to write to random files at hex offsets like adb.
1825
1826 --
1827
1828 [elena - delete this]
1829
1830 Handle add_file with separate text, data, and bss addresses.  Maybe
1831 handle separate addresses for each segment in the object file?
1832
1833 --
1834
1835 [Jimb/Elena delete this one]
1836
1837 Handle free_named_symtab to cope with multiply-loaded object files
1838 in a dynamic linking environment.  Should remember the last copy loaded,
1839 but not get too snowed if it finds references to the older copy.
1840
1841 --
1842
1843 [elena delete this also]
1844
1845 Remove all references to:
1846         text_offset
1847         data_offset
1848         text_data_start
1849         text_end
1850         exec_data_offset
1851         ...
1852 now that we have BFD.  All remaining are in machine dependent files.
1853
1854 --
1855
1856 Re-organize help categories into things that tend to fit on a screen
1857 and hang together.
1858
1859 --
1860
1861 Add in commands like ADB's for searching for patterns, etc.  We should
1862 be able to examine and patch raw unsymboled binaries as well in gdb as
1863 we can in adb.  (E.g. increase the timeout in /bin/login without source).
1864
1865 [actually, add ADB interface :-]
1866
1867 --
1868
1869 When doing "step" or "next", if a few lines of source are skipped between
1870 the previous line and the current one, print those lines, not just the
1871 last line of a multiline statement.
1872
1873 --
1874
1875 Handling of "&" address-of operator needs some serious overhaul
1876 for ANSI C and consistency on arrays and functions.
1877   For "float point[15];":
1878 ptype &point[4]   ==> Attempt to take address of non-lvalue.
1879   For "char *malloc();":
1880 ptype malloc      ==> "char *()";  should be same as
1881 ptype &malloc     ==> "char *(*)()"
1882 call printf ("%x\n", malloc)   ==> weird value, should be same as
1883 call printf ("%x\n", &malloc)  ==> correct value
1884
1885 --
1886
1887 Fix dbxread.c symbol reading in the presence of interrupts.  It
1888 currently leaves a cleanup to blow away the entire symbol table when a
1889 QUIT occurs.  (What's wrong with that? -kingdon, 28 Oct 1993).
1890
1891 [I suspect that the grype was that, on a slow system, you might want
1892 to cntrl-c and get just half the symbols and then load the rest later
1893 - scary to be honest]
1894
1895 --
1896
1897 Mipsread.c reads include files depth-first, because the dependencies
1898 in the psymtabs are way too inclusive (it seems to me).  Figure out what
1899 really depends on what, to avoid recursing 20 or 30 times while reading
1900 real symtabs.
1901
1902 --
1903
1904 value_add() should be subtracting the lower bound of arrays, if known,
1905 and possibly checking against the upper bound for error reporting.
1906
1907 --
1908
1909 When listing source lines, check for a preceding \n, to verify that
1910 the file hasn't changed out from under us.
1911
1912 [fixed by some other means I think.  That hack wouldn't actually work
1913 reliably - the file might move such that another \n appears. ]
1914
1915 --
1916
1917 Get all the remote systems (where the protocol allows it) to be able to
1918 stop the remote system when the GDB user types ^C (like remote.c
1919 does).  For ebmon, use ^Ak.
1920
1921 --
1922
1923 Possible feature: A version of the "disassemble" command which shows
1924 both source and assembly code ("set symbol-filename on" is a partial
1925 solution).
1926
1927 [has this been done?  It was certainly done for MI and GDBtk]
1928
1929 --
1930
1931 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1932 using a 0 address for bad purposes internally).
1933
1934 --
1935
1936 Make "info path" and path_command work again (but independent of the
1937 environment either of gdb or that we'll pass to the inferior).
1938
1939 --
1940
1941 Make GDB understand the GCC feature for putting octal constants in
1942 enums.  Make it so overflow on an enum constant does not error_type
1943 the whole type.  Allow arbitrarily large enums with type attributes.
1944 Put all this stuff in the testsuite.
1945
1946 --
1947
1948 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1949 the value in hex; process type attributes).  Add this to the
1950 testsuite.  This way future compilers can add new types and old
1951 versions of GDB can do something halfway reasonable.
1952
1953 --
1954
1955 Fix mdebugread.c:parse_type to do fundamental types right (see
1956 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1957 that the debug format fixes the sizes of these things and it shouldn't
1958 depend on stuff like TARGET_PTR_BIT and so on.  For mdebug, there seem
1959 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1960 be aware of that).  Also use a switch statement for clarity and speed.
1961
1962 --
1963
1964 Investigate adding symbols in target_load--some targets do, some
1965 don't.
1966
1967 --
1968
1969 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1970 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1971 bar.c).
1972
1973 --
1974
1975 Merge xcoffread.c and coffread.c.  Use breakpoint_re_set instead of
1976 fixup_breakpoints.
1977
1978 --
1979
1980 Make a watchpoint which contains a function call an error (it is
1981 broken now, making it work is probably not worth the effort).
1982
1983 --
1984
1985 New test case based on weird.exp but in which type numbers are not
1986 renumbered (thus multiply defining a type).  This currently causes an
1987 infinite loop on "p v_comb".
1988
1989 --
1990
1991 [Hey! Hint Hint Delete Delete!!!]
1992
1993 Fix 386 floating point so that floating point registers are real
1994 registers (but code can deal at run-time if they are missing, like
1995 mips and 68k).  This would clean up "info float" and related stuff.
1996
1997 --
1998
1999 gcc -g -c enummask.c then gdb enummask.o, then "p v".  GDB complains
2000 about not being able to access memory location 0.
2001
2002 -------------------- enummask.c
2003 enum mask
2004 {
2005   ANIMAL = 0,
2006   VEGETABLE = 1,
2007   MINERAL = 2,
2008   BASIC_CATEGORY = 3,
2009
2010   WHITE = 0,
2011   BLUE = 4,
2012   GREEN = 8,
2013   BLACK = 0xc,
2014   COLOR = 0xc,
2015
2016   ALIVE = 0x10,
2017
2018   LARGE = 0x20
2019 } v;
2020
2021 --
2022
2023 If try to modify value in file with "set write off" should give
2024 appropriate error not "cannot access memory at address 0x65e0".
2025
2026 --
2027
2028 Allow core file without exec file on RS/6000.
2029
2030 --
2031
2032 Make sure "shell" with no arguments works right on DOS.
2033
2034 --
2035
2036 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
2037 the same directory can be NFS-mounted on unix or DOS, and work the
2038 same way.
2039
2040 --
2041
2042 [Is this another delete???]
2043
2044 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
2045 get RS/6000 to work right, might not be immediately relevant).
2046
2047 --
2048
2049 Work out some kind of way to allow running the inferior to be done as
2050 a sub-execution of, eg. breakpoint command lists.  Currently running
2051 the inferior interupts any command list execution.  This would require
2052 some rewriting of wait_for_inferior & friends, and hence should
2053 probably be done in concert with the above.
2054
2055 --
2056
2057 Add function arguments to gdb user defined functions.
2058
2059 --
2060
2061 Add convenience variables that refer to exec file, symbol file,
2062 selected frame source file, selected frame function, selected frame
2063 line number, etc.
2064
2065 --
2066
2067 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2068 allocate a partial symtab for each BINCL/EINCL grouping.  This will
2069 seriously decrease the size of inter-psymtab dependencies and hence
2070 lessen the amount that needs to be read in when a new source file is
2071 accessed.
2072
2073 --
2074
2075 Add a command for searching memory, a la adb.  It specifies size,
2076 mask, value, start address.  ADB searches until it finds it or hits
2077 an error (or is interrupted).
2078
2079 --
2080
2081 Remove the range and type checking code and documentation, if not
2082 going to implement.
2083
2084 # Local Variables:
2085 # mode: text
2086 # End: