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 Obsolete ser-ocd.c.
  30
  31 --
  32
  33 Wow, three bug reports for the same problem in one day!  We should
  34 probably make fixing this a real priority :-).
  35
  36 Anyway, thanks for reporting.
  37
  38 The following patch will fix the problems with setting breakpoints in
  39 dynamically loaded objects:
  40
  41    http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00230.html
  42
  43 This patch isn't checked in yet (ping Michael/JimB), but I hope this
  44 will be in the next GDB release.
  45
  46 There should really be a test in the testsuite for this problem, since
  47 it keeps coming up :-(.  Any volunteers?
  48
  49 Mark
  50
  51 --
  52
  53 x86 linux GDB and SIGALRM (???)
  54 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00803.html
  55
  56 This problem has been fixed, but a regression test still needs to be
  57 added to the testsuite:
  58 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00309.html
  59
  60 Mark
  61
  62 [The test has been submitted for approval - cagney]
  63
  64 --
  65
  66 RFD: infrun.c: No bpstat_stop_status call after proceed over break?
  67 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00665.html
  68
  69 GDB misses watchpoint triggers after proceeding over a breakpoint on
  70 x86 targets.
  71
  72 --
  73
  74 GDB 5.0 doesn't work on Linux/SPARC
  75
  76 There are two parts to this.
  77
  78       o   GDB 5.0 doesn't work on GNU/Linux/SPARC32
  79
  80       o   GDB 5.0 doesn't work on the new target
  81           GNU/Linux/SPARC64
  82
  83 GDB does build on both these targets.
  84
  85 The first problem is the one that should be fixed.
  86
  87 --
  88
  89                 GDB 5.1 - New features
  90                 ======================
  91
  92 The following new features should be included in 5.1.
  93
  94 --
  95
  96 Enable MI by default.  Old code can be deleted after 5.1 is out.
  97
  98 Issues:
  99
 100         o       syntax change where a list would
 101                 look like:
 102                      [ foo=a, foo=b, foo=c ]
 103                 instead of
 104                      { foo=a, foo=b, foo=c }
 105
 106         o       kill off the idea of a reverse
 107                 query.
 108
 109         o       review test cases
 110
 111         o       enable it
 112
 113 --
 114
 115 Pascal (Pierre Muller, David Taylor)
 116
 117 Pierre Muller has contributed patches for adding Pascal Language
 118 support to GDB.
 119
 120 2 pascal language patches inserted in database
 121 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
 122
 123 Indent -gnu ?
 124 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
 125
 126 [I think this has been merged, need to confirm - cagney]
 127
 128 --
 129
 130 Java (Anthony Green, David Taylor)
 131
 132 Anthony Green has a number of Java patches that did not make it into
 133 the 5.0 release.  The first two are in cvs now, but the third needs
 134 some fixing up before it can go in.
 135
 136 Patch: java tests
 137 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
 138
 139 Patch: java booleans
 140 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
 141
 142 Patch: handle N_MAIN stab
 143 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
 144
 145 -- 2001-03-08
 146
 147 Add CRIS target.
 148
 149 A predicate to this is the multi-arching of SOFTWARE_SINGLE_STEP().  A
 150 patch has been submitted.
 151
 152 --
 153
 154                 GDB 5.1 - Cleanups
 155                 ==================
 156
 157 The following code cleanups will hopefully be applied to GDB 5.1.
 158
 159 -- 2001-03-26
 160
 161 Resolve the build status of all broken targets as identified by the
 162 MAINTAINERS file.
 163
 164         o       arm-* vs NetBSD's lack of ``unix''
 165         o       arm-* vs IRIX (see below)
 166
 167 --
 168
 169 Fix copyright notices.
 170
 171 Turns out that ``1998-2000'' isn't considered valid :-(
 172
 173 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
 174
 175 --
 176
 177                 GDB 5.1 - Known Problems
 178                 ========================
 179
 180 --
 181
 182 z8k
 183
 184 The z8k has suffered bit rot and is known to not build.  The problem
 185 was occuring in the opcodes directory.
 186
 187 --
 188
 189 m88k
 190
 191 The m88k has suffered bit rot and is known to not build.
 192
 193 --
 194
 195 Solaris 8 x86 CURSES_H problem
 196 http://sources.redhat.com/ml/gdb/2000-07/msg00038.html
 197
 198 The original problem was worked around with:
 199
 200     2000-06-06  Michael Snyder  <msnyder@cygnus.com>
 201
 202         * configure.in: Enable autoconf to find curses.h on Solaris 2.8.
 203         * configure: Regenerate.
 204
 205 When building both GDB and SID using the same source tree the problem
 206 will still occure. sid/component/configure.in mis-configures
 207 <curses.h> and leaves wrong information in the config cache.
 208
 209 --
 210
 211                 GDB 5.2 - Fixes
 212                 ===============
 213
 214 --
 215
 216 Thread support.  Right now, as soon as a thread finishes and exits,
 217 you're hosed.  This problem is reported once a week or so.
 218
 219 --
 220
 221                 GDB 5.2 - New features
 222                 ======================
 223
 224 --
 225
 226 GCC 3.0 ABI support (but hopefully sooner...).
 227
 228 --
 229
 230 Objective C/C++ support (but hopefully sooner...).
 231
 232 --
 233
 234                 GDB 5.2 - Cleanups
 235                 ==================
 236
 237 The following cleanups have been identified as part of GDB 5.2.
 238
 239 --
 240
 241 Remove old code that does not use ui_out functions and all the related
 242 "ifdef"s.  This also allows the elimination of -DUI_OUT from
 243 Makefile.in and configure.in.
 244
 245 --
 246
 247 Compiler warnings.
 248
 249 Eliminate warnings for all targets on at least one host for one of the
 250 -W flags.  Flags up for debate include: -Wswitch -Wcomment -trigraphs
 251 -Wtrigraphs -Wunused-function -Wunused-label -Wunused-variable
 252 -Wunused-value -Wchar-subscripts -Wtraditional -Wshadow -Wcast-qual
 253 -Wcast-align -Wwrite-strings -Wconversion -Wstrict-prototypes
 254 -Wmissing-prototypes -Wmissing-declarations -Wredundant-decls
 255 -Woverloaded-virtual -Winline
 256
 257 --
 258
 259 Deprecate, if not delete, the following:
 260
 261         register[]
 262         register_valid[]
 263         register_buffer()
 264         REGISTER_BYTE()
 265                 Replaced by, on the target side
 266                   supply_register()
 267                 and on core-gdb side:
 268                   {read,write}_register_gen()
 269                 Remote.c will need to use something
 270                 other than REGISTER_BYTE() and
 271                 REGISTER_RAW_SIZE() when unpacking
 272                 [gG] packets.
 273
 274         STORE_PSEUDO_REGISTER
 275         FETCH_PSEUDO_REGISTER
 276                 Now handed by the methods
 277                   gdbarch_{read,write}_register()
 278                 which sits between core GDB and
 279                 the register cache.
 280
 281         REGISTER_CONVERTIBLE
 282         REGISTER_CONVERT_TO_RAW
 283         REGISTER_CONVERT_TO_VIRTUAL
 284                 I think these three are redundant.
 285                 gdbarch_register_{read,write} can
 286                 do any conversion it likes.
 287
 288         REGISTER_VIRTUAL_SIZE
 289         MAX_REGISTER_VIRTUAL_SIZE
 290         REGISTER_VIRTUAL_TYPE
 291                 I think these can be replaced by
 292                 the pair:
 293                   FRAME_REGISTER_TYPE(frame, regnum)
 294                   REGISTER_TYPE(regnum)
 295
 296         DO_REGISTERS_INFO
 297                 Replace with
 298                  FRAME_REGISTER_INFO (frame, ...)
 299
 300         REGISTER_SIM_REGNO()
 301                 If nothing else rename this so that
 302                 how it relates to rawreg and the
 303                 regnum is clear.
 304
 305         REGISTER_BYTES
 306                 The size of the cache can be computed
 307                 on the fly.
 308
 309         IS_TRAPPED_INTERNALVAR
 310                 The pseudo registers should eventually make
 311                 this redundant.
 312
 313 --
 314
 315 Obsolete the targets.
 316
 317 arm*-wince-pe
 318 mips*-*-pe
 319 sh*-*-pe
 320
 321 Obsolete the protocols:
 322
 323 RDB
 324
 325 ``As of version 5.3, WindRiver has removed the RDB server (RDB
 326 protocol support is built into gdb).''  -- Till.
 327
 328 --
 329
 330 Restructure gdb directory tree so that it avoids any 8.3 and 14
 331 filename problems.
 332
 333 --
 334
 335 Convert GDB build process to AUTOMAKE.
 336
 337 See also sub-directory configure below.
 338
 339 The current convention is (kind of) to use $(<header>_h) in all
 340 dependency lists.  It isn't done in a consistent way.
 341
 342 --
 343
 344                 GDB 5.2 - Known Problems
 345                 ========================
 346
 347 --
 348
 349                 Code Cleanups: General
 350                 ======================
 351
 352 The following are more general cleanups and fixes.  They are not tied
 353 to any specific release.
 354
 355 --
 356
 357 Investigate changing --target=a29k-amd-udi to a29k-*-coff* and
 358 rationalize *.mt files.  The got-ya is in remote-eb.c - it has its own
 359 custom tty manipulation - it should be using the serial object.
 360
 361 --
 362
 363 Rename read_register{,_pid}() to read_unsigned_register{,_pid}().
 364
 365 --
 366
 367 Can't build IRIX -> arm GDB.
 368 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
 369
 370 David Whedon writes:
 371 > Now I'm building for an embedded arm target.  If there is a way of turning
 372 > remote-rdi off, I couldn't find it. It looks like it gets built by default
 373 > in gdb/configure.tgt(line 58) Anyway, the build dies in
 374 > gdb/rdi-share/unixcomm.c.  SERPORT1 et. al. never get defined because we
 375 > aren't one of the architectures supported.
 376
 377 --
 378
 379 Problem with weak functions
 380 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
 381
 382 Dan Nicolaescu writes:
 383 > It seems that gdb-4.95.1  does not display correctly the function when
 384 > stoping in weak functions.
 385 >
 386 > It stops in a function that is defined as weak, not in the function
 387 > that is actually run...
 388
 389 --
 390
 391 Follow through `make check' with --enable-shared.
 392
 393 When the srcware tree is configured with --enable-shared, the `expect'
 394 program won't run properly.  Jim Wilson found out gdb has a local hack
 395 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
 396 similarly.
 397
 398 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
 399
 400 --
 401
 402 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
 403
 404 Patches in the database.
 405
 406 --
 407
 408 printcmd.c (print_address_numeric):
 409
 410 NOTE: This assumes that the significant address information is kept in
 411 the least significant bits of ADDR - the upper bits were either zero
 412 or sign extended.  Should ADDRESS_TO_POINTER() or some
 413 ADDRESS_TO_PRINTABLE() be used to do the conversion?
 414
 415 --
 416
 417 The BFD directory requires bug-fixed AUTOMAKE et.al.
 418
 419 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable.  It
 420 contained the full path to texinfo.tex when it should have only
 421 contained the directory.  The bug has been fixed in the current
 422 AUTOMAKE sources.  Automake snapshots can be found in:
 423         ftp://sourceware.cygnus.com/pub/gdb/snapshots
 424 and     ftp://sourceware.cygnus.com/pub/binutils
 425
 426 --
 427
 428 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
 429 determine the default isa/byte-order.
 430
 431 --
 432
 433 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependent
 434 BIG_ENDIAN and LITTLE_ENDIAN.
 435
 436 --
 437
 438 Eliminate more compiler warnings.
 439
 440 Of course there also needs to be the usual debate over which warnings
 441 are valid and how to best go about this.
 442
 443 One method: choose a single option; get agreement that it is
 444 reasonable; try it out to see if there isn't anything silly about it
 445 (-Wunused-parameters is an example of that) then incrementally hack
 446 away.
 447
 448 The other method is to enable all warnings and eliminate them from one
 449 file at a time.
 450
 451 --
 452
 453 Elimination of ``(catch_errors_ftype *) func''.
 454
 455 Like make_cleanup_func it isn't portable.
 456 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
 457 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
 458
 459 --
 460
 461 Nuke #define CONST_PTR.
 462
 463 --
 464
 465 Nuke USG define.
 466
 467 --
 468
 469 [PATCH/5] src/intl/Makefile.in:distclean additions
 470 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
 471
 472 Do not forget to merge the patch back into the trunk.
 473
 474 --
 475
 476 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
 477
 478 At present defs.h includes <endian.h> (which is linux specific) yet
 479 almost nothing depends on it.  Suggest "gdb_endian.h" which can also
 480 handle <machine/endian.h> and only include that where it is really
 481 needed.
 482
 483 --
 484
 485 Replace savestring() with something from libiberty.
 486
 487 An xstrldup()? but that would have different semantics.
 488
 489 --
 490
 491 Rationalize use of floatformat_unknown in GDB sources.
 492
 493 Instead of defaulting to floatformat_unknown, should hosts/targets
 494 specify the value explicitly?
 495
 496 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
 497
 498 --
 499
 500 Add a ``name'' member to include/floatformat.h:struct floatformat.
 501 Print that name in gdbarch.c.
 502
 503 --
 504
 505 Sort out the harris mess in include/floatformat.h (it hardwires two
 506 different floating point formats).
 507
 508 --
 509
 510 See of the GDB local floatformat_do_doublest() and libiberty's
 511 floatformat_to_double (which was once GDB's ...) can be merged some
 512 how.
 513
 514 --
 515
 516 Eliminate mmalloc(), mstrsave() et.al. from GDB.
 517
 518 Also eliminate it from defs.h.
 519
 520 --
 521
 522 Eliminate PTR.  ISO-C allows ``void *''.
 523
 524 --
 525
 526 Eliminate abort ().
 527
 528 GDB should never abort.  GDB should either throw ``error ()'' or
 529 ``internal_error ()''.  Better still GDB should naturally unwind with
 530 an error status.
 531
 532 --
 533
 534 GDB probably doesn't build on FreeBSD pre 2.2.x
 535 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
 536
 537 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
 538 suffer bit rot.
 539
 540 --
 541
 542 Deprecate "fg".  Apparently ``fg'' is actually continue.
 543
 544 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
 545
 546 --
 547
 548 Deprecate current use of ``floatformat_unknown''.
 549
 550 Require all targets to explicitly provide their float format instead
 551 of defaulting to floatformat unknown.  Doing the latter leads to nasty
 552 bugs.
 553
 554 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
 555
 556 --
 557
 558 Rationalize floatformat_to_double() vs floatformat_to_doublest().
 559
 560 Looks like GDB migrated floatformat_to_double() to libiberty but then
 561 turned around and created a ..._to_doublest() the latter containing
 562 several bug fixes.
 563
 564 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
 565
 566 --
 567
 568 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
 569
 570 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
 571
 572 --
 573
 574 The ``maintenance deprecate set endian big'' command doesn't notice
 575 that it is deprecating ``set endian'' and not ``set endian big'' (big
 576 is implemented using an enum).  Is anyone going to notice this?
 577
 578 --
 579
 580 When tab expanding something like ``set arch<tab>'' ignore the
 581 deprecated ``set archdebug'' and expand to ``set architecture''.
 582
 583 --
 584
 585 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
 586 like from arm-tdep.c.
 587
 588 --
 589
 590 Fix uses of ->function.cfunc = set_function().
 591
 592 The command.c code calls sfunc() when a set command.  Rather than
 593 change it suggest fixing the callback function so that it is more
 594 useful.  See:
 595
 596 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
 597
 598 See also ``Fix implementation of  ``target xxx''.'' below.
 599
 600 --
 601
 602 IRIX 3.x support is probably broken.
 603
 604 --
 605
 606 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
 607 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
 608
 609 Apart from the d30v, are there any sim/common simulators that make use
 610 of this?
 611
 612 A brief summary of what happened is that sim/common/sim-break.c was
 613 created as a good idea.  It turned out a better idea was to use
 614 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
 615
 616 --
 617
 618 Move remote_remove_hw_breakpoint, remote_insert_hw_breakpoint,
 619 remote_remove_watchpoint, remote_insert_watchpoint into target vector.
 620
 621 --
 622
 623 Eliminate ``extern'' from C files.
 624
 625 --
 626
 627 Replace ``STREQ()'' et.al. with ``strcmp() == 0'' et.al.
 628
 629 Extreme care is recommeded - perhaps only modify tests that are
 630 exercised by the testsuite (as determined using some type of code
 631 coverage analysis).
 632
 633 --
 634
 635 Replace the file gdb/CONTRIBUTE with a file that is generated from the
 636 gdb/doc/*.texinfo directory.
 637
 638 --
 639
 640 Rewrite/break up sparcl-tdep.c so that it uses ser*.c as the mechanism
 641 for accessing either the serial or UDP port.
 642
 643 --
 644
 645                         New Features and Fixes
 646                         ======================
 647
 648 These are harder than cleanups but easier than work involving
 649 fundamental architectural change.
 650
 651 --
 652
 653 Hardware watchpoint problems on x86 OSes, including Linux:
 654
 655 1. Delete/disable hardware watchpoints should free hardware debug
 656 registers.
 657 2. Watch for different values on a viariable with one hardware debug
 658 register.
 659
 660 According to Eli Zaretskii <eliz@delorie.com>:
 661
 662 These are not GDB/ia32 issues per se: the above features are all
 663 implemented in the DJGPP port of GDB and work in v5.0.  Every
 664 x86-based target should be able to lift the relevant parts of
 665 go32-nat.c and use them almost verbatim.  You get debug register
 666 sharing through reference counts, and the ability to watch large
 667 regions (up to 16 bytes) using multiple registers.  (The required
 668 infrastructure in high-level GDB application code, mostly in
 669 breakpoint.c, is also working since v5.0.)
 670
 671 --
 672
 673 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
 674 so that you can see how the GDB was created.
 675
 676 --
 677
 678 Add an "info bfd" command that displays supported object formats,
 679 similarly to objdump -i.
 680
 681 Is there a command already?
 682
 683 --
 684
 685 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
 686
 687 This requires internationalization.
 688
 689 --
 690
 691 Add support for:
 692
 693 (gdb) p fwprintf(stdout,L"%S\n", f)
 694 No symbol "L" in current context.
 695
 696 --
 697
 698 Cleanup configury support for optional sub-directories.
 699
 700 Check how GCC handles multiple front ends for an example of how things
 701 could work.  A tentative first step is to rationalize things so that
 702 all sub directories are handled in a fashion similar to gdb/mi.
 703
 704 See also automake above.
 705
 706 --
 707
 708 Add a transcript mechanism to GDB.
 709
 710 Such a mechanism might log all gdb input and output to a file in a
 711 form that would allow it to be replayed.  It could involve ``gdb
 712 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
 713
 714 --
 715
 716 Can the xdep files be replaced by autoconf?
 717
 718 --
 719
 720 Document trace machinery
 721
 722 --
 723
 724 Document ui-out and ui-file.
 725
 726 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
 727
 728 --
 729
 730 Update texinfo.tex to latest?
 731
 732 --
 733
 734 Incorporate agentexpr.texi into gdb.texinfo
 735
 736 agentexpr.texi mostly describes the details of the byte code used for
 737 tracepoints, not the internals of the support for this in GDB.  So it
 738 looks like gdb.texinfo is a better place for this information.
 739
 740 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
 741
 742 --
 743
 744 Document overlay machinery.
 745
 746 --
 747
 748 ``(gdb) catch signal SIGNAL''
 749
 750 Overlaps with ``handle SIGNAL'' but the implied behavior is different.
 751 You can attach commands to a catch but not a handle.  A handle has a
 752 limited number of hardwired actions.
 753
 754 --
 755
 756 Fix TUI
 757
 758      o  readline/*.h bitrot
 759
 760         The TUI isn't up-to-date with
 761         respect to the readline currently
 762         bundled with GDB.  Importing a
 763         new readline is on the 5.1 wish
 764         list so this can only get worse.
 765
 766         Grep for things like term_cursor_move.
 767
 768         (To be honest, I don't see anyone
 769         importing a new readline before 5.1 is
 770         out)
 771
 772      o  tui.c:va_catch_errors() bitrot
 773
 774         This nasty piece of work used knowledge
 775         of the internals of GDBs error functions :-(
 776         Ever since those internals were cleaned
 777         up this code has been broken. :-(
 778
 779      o  tuiWin.c:c_makeVisibleWithNewHeight() broken
 780         tuiLayout.c:_extractDisplayStartAddr() broken
 781
 782         Both these function call find_line_pc()
 783         incorrectly (wrong args, wrong return value).
 784
 785         I suspect this bug has always been there!
 786         It had been hidden because those files
 787         didn't include the necessary header files
 788         from gdb proper :-(
 789
 790      o  tuiRegs() host dependant
 791
 792         Not suprisingly, this isn't a very portable
 793         section of code.  However, I'm sure people
 794         could live with no regs in the short to
 795         medium term.
 796
 797      o  defs.h: #include "tui.h" et.al.
 798
 799         I'm not sure where this came from.
 800         It was a really bad idea.
 801
 802         To get things to compile I did a nasty
 803         hack (Just declare what was needed and
 804         replace any expressions like xx->y.z()
 805         in GDB proper with function calls).  I
 806         could commit it slightly cleaned up if
 807         you like.
 808
 809         Medium Term. the #ifdef TUI and TuiDo()
 810         should be changed to hooks (like GDBTK).
 811         The gdb-events.[hc] is there for that
 812         purpose (1)
 813
 814      o  tui.c:_tuiReset() host dependant
 815
 816         tui.c contains a lump of termio[s]
 817         I suspect an equivalent block of
 818         code can be lifted from readline.
 819         An equivalent readline function may
 820         even be available.
 821
 822      o  curses.h vs ncurses.h.
 823
 824         Simple portability problem.
 825
 826      o  subsetCompare()
 827
 828         This function is a mystery - where is it?
 829
 830      o  tui-file.[hc] cleanup
 831
 832         This can be significantly simplified.
 833
 834      o  The code should be pacified. (-Werror -W...)
 835
 836         There are plenty of #includes,
 837         duplicate #includes, missing function decls
 838         and the like.
 839
 840         Some of the problems I found were through
 841         fixing a few of the warnings.
 842
 843      o  The code should be GNUtified.
 844
 845         It would be very nice to have this code
 846         look like the rest of GDB.  That way people
 847         would be more accepting of it as a true
 848         gdb component.
 849
 850         Until it is GNUtified it is going to stick
 851         out like a sore thumb to the programmer.
 852
 853      o  The code should be clearly copyrighted
 854
 855         (FSF, with due credit to HP)
 856
 857 --
 858
 859 Add support for ``gdb --- PROGRAM ARGS ...''.
 860 Add support for ``gdb -cmd=...''
 861
 862 Along with many variations.  Check:
 863
 864 ????? for a full discussion.
 865
 866 for a discussion.
 867
 868 --
 869
 870 Implement ``(gdb) !ls''.
 871
 872 Which is very different from ``(gdb) ! ls''.  Implementing the latter
 873 is trivial.
 874
 875 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
 876
 877 --
 878
 879 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
 880 be passed.
 881
 882 --
 883
 884 When tab expanding something like ``set arch<tab>'' ignore the
 885 deprecated ``set archdebug'' and expand to ``set architecture''.
 886
 887 --
 888
 889 Replace the code that uses the host FPU with an emulator of the target
 890 FPU.
 891
 892 --
 893
 894 The "ocd reset" command needs to flush the dcache, which requires breaking
 895 the abstraction layer between the target independent and target code.  One
 896 way to address this is provide a generic "reset" command and target vector.
 897
 898 http://sources.redhat.com/ml/gdb-patches/2000-10/msg00011.html
 899
 900 --
 901
 902                         Thread Support
 903                         ==============
 904
 905 --
 906
 907 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
 908 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
 909
 910 The thread_db assisted debugging code doesn't handle exiting threads
 911 properly, at least in combination with glibc 2.1.3 (the framework is
 912 there, just not the actual code).  There are at least two problems
 913 that prevent this from working.
 914
 915 As an additional reference point, the pre thread_db code did not work
 916 either.
 917
 918 --
 919
 920 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
 921 Solaris/x86).
 922 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
 923
 924 Christopher Blizzard writes:
 925
 926 So, I've done some more digging into this and it looks like Jim
 927 Kingdon has reported this problem in the past:
 928
 929 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
 930
 931 I can reproduce this problem both with and without Tom's patch.  Has
 932 anyone seen this before?  Maybe have a solution for it hanging around?
 933 :)
 934
 935 There's a test case for this documented at:
 936
 937 when debugging threaded applications you get extra SIGTRAPs
 938 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
 939
 940 [There should be a GDB testcase - cagney]
 941
 942 --
 943
 944 GDB5 TOT on unixware 7
 945 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
 946
 947 Robert Lipe writes:
 948 > I just spun the top of tree of the GDB5 branch on UnixWare 7.  As a
 949 > practical matter, the current thread support is somewhat more annoying
 950 > than when GDB was thread-unaware.
 951
 952 --
 953
 954                         Language Support
 955                         ================
 956
 957 New languages come onto the scene all the time.
 958
 959 --
 960
 961 Re: Various C++ things
 962
 963 value_headof/value_from_vtable_info are worthless, and should be
 964 removed.  The one place in printcmd.c that uses it should use the RTTI
 965 functions.
 966
 967 RTTI for g++ should be using the typeinfo functions rather than the
 968 vtables.  The typeinfo functions are always at offset 4 from the
 969 beginning of the vtable, and are always right. The vtables will have
 970 weird names like E::VB sometimes.  The typeinfo function will always
 971 be "E type_info function", or somesuch.
 972
 973 value_virtual_fn_field needs to be fixed so there are no failures for
 974 virtual functions for C++ using g++.
 975
 976 Testsuite cases are the major priority right now for C++ support,
 977 since i have to make a lot of changes that could potentially break
 978 each other.
 979
 980 --
 981
 982 Add support for Modula3
 983
 984 Get DEC/Compaq to contribute their Modula-3 support.
 985
 986 --
 987
 988                 Remote Protocol Support
 989                 =======================
 990
 991 --
 992
 993 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
 994 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
 995
 996 (Broken) support for GDB's remote protocol across UDP is to be
 997 included in the follow-on release.
 998
 999 It should be noted that UDP can only work when the [Gg] packet fits in
1000 a single UDP packet.
1001
1002 There is also much debate over the merit of this.
1003
1004 --
1005
1006 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
1007
1008 Add support for packet enable/disable commands with these thread
1009 packets.  General cleanup.
1010
1011 [PATCH] Document the ThreadInfo remote protocol queries
1012 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
1013
1014 [PATCH] "info threads" queries for remote.c
1015 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
1016
1017 --
1018
1019 Remote protocol doco feedback.
1020
1021 Too much feedback to mention needs to be merged in (901660).  Search
1022 for the word ``remote''.
1023
1024
1025 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
1026 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
1027 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
1028
1029 --
1030
1031 GDB doesn't recover gracefully from remote protocol errors.
1032
1033 GDB wasn't checking for NAKs from the remote target.  Instead a NAK is
1034 ignored and a timeout is required before GDB retries.  A pre-cursor to
1035 fixing this this is making GDB's remote protocol packet more robust.
1036
1037 While downloading to a remote protocol target, gdb ignores packet
1038 errors in so far as it will continue to download with chunk N+1 even
1039 if chunk N was not correctly sent.  This causes gdb.base/remote.exp to
1040 take a painfully long time to run.  As a PS that test needs to be
1041 fixed so that it builds on 16 bit machines.
1042
1043 --
1044
1045 Fix the ``!'' packet.
1046
1047 JT reported that the existing targets do, in fact return ``OK'' so it
1048 is possible to merge remote and extended-remote targets.
1049
1050 --
1051
1052 Drop ``<address>'' from the [SsCc] packets.
1053
1054 I don't think that GDB generates them so having it in the protocol is
1055 silly.
1056
1057 --
1058
1059 Fix doco on the ``q'' packet.
1060
1061 It has evolved into a generic RPC.  The notes should reflect this and,
1062 perhaps, the ``Q'' packet can be deprecated.
1063
1064 The doco should mention that ``OK'' is a valid packet response.
1065
1066 The doco should explain why ``OK'' needs to be a valid packet
1067 response.
1068
1069 --
1070
1071 Add the cycle step command.
1072
1073 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
1074
1075 --
1076
1077 Resolve how to scale things to support very large packets.
1078
1079 --
1080
1081 Resolve how to handle a target that changes things like its endianess
1082 on the fly - should it be returned in the ``T'' packet?
1083
1084 Underlying problem is that the register file is target endian.  If the
1085 target endianess changes gdb doesn't know.
1086
1087 Suggest encoding registers as NN!VALUE.
1088
1089 --
1090
1091 GDB should allow incomming packets to be larger than outgoing ones.  A
1092 fully loaded T packet (containing all registers) can be very large -
1093 definitly larger than a corresponding Gg packet.
1094
1095 --
1096
1097                         Symbol Support
1098                         ==============
1099
1100 If / when GDB starts to support the debugging of multi-processor
1101 (rather than multi-thread) applications the symtab code will need to
1102 be updated a little so that several independent symbol tables are
1103 active at a given time.
1104
1105 The other interesting change is a clarification of the exact meaning
1106 of CORE_ADDR and that has had consequences for a few targets (that
1107 were abusing that data type).
1108
1109 --
1110
1111 Investiagate ways of reducing memory.
1112
1113 --
1114
1115 Investigate ways of improving load time.
1116
1117 --
1118
1119 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
1120
1121 Consequence of recent symtab clarification.  No marks for figuring out
1122 who maintains the d10v.
1123
1124 --
1125
1126 Get the MIPS to correctly sign extend all address <-> pointer
1127 conversions.
1128
1129 Consequence of recent symtab clarification.  No marks for figuring out
1130 who maintains the MIPS.
1131
1132 --
1133
1134 GDB truncates 64 bit enums.
1135
1136 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
1137
1138 --
1139
1140                         Testsuite Support
1141                         =================
1142
1143 There are never to many testcases.
1144
1145 --
1146
1147 Better thread testsuite.
1148
1149 --
1150
1151 Better C++ testsuite.
1152
1153 --
1154
1155 Look at adding a GDB specific testsuite directory so that white box
1156 tests of key internals can be added (eg ui_file).
1157
1158 --
1159
1160 Separate out tests that involve the floating point (FP).
1161
1162 (Something for people brining up new targets).  FP and non-fp tests
1163 are combined.  I think there should be set of basic tests that
1164 exercise pure integer support and then a more expanded set that
1165 exercise FP and FP/integer interactions.
1166
1167 As an example, the MIPS, for n32 as problems with passing FP's and
1168 structs.  Since most inferior call tests include FP it is difficult to
1169 determine of the integer tests are ok.
1170
1171 --
1172
1173                 Architectural Changes: General
1174                 ==============================
1175
1176 These are harder than simple cleanups / fixes and, consequently
1177 involve more work.  Typically an Architectural Change will be broken
1178 down into a more digestible set of cleanups and fixes.
1179
1180 --
1181
1182 Cleanup software single step.
1183
1184 At present many targets implement software single step by directly
1185 blatting memory (see rs6000-tdep.c).  Those targets should register
1186 the applicable breakpoints using the breakpoint framework.  Perhaphs a
1187 new internal breakpoint class ``step'' is needed.
1188
1189 --
1190
1191 Replace READ_FP() with FRAME_HANDLE().
1192
1193 READ_FP() is a hangover from the days of the vax when the ABI really
1194 did have a frame pointer register.  Modern architectures typically
1195 construct a virtual frame-handle from the stack pointer and various
1196 other bits of string.
1197
1198 Unfortunately GDB still treats this synthetic FP register as though it
1199 is real.  That in turn really confuses users (arm and ``print $fp'' VS
1200 ``info registers fp'').  The synthetic FP should be separated out of
1201 the true register set presented to the user.
1202
1203 --
1204
1205 Register Cache Cleanup (below from Andrew Cagney)
1206
1207 I would depict the current register architecture as something like:
1208
1209         High GDB  -->   Low GDB
1210             |              |
1211            \|/            \|/
1212             --- REG NR -----
1213                    |
1214     register + REGISTER_BYTE(reg_nr)
1215                    |
1216                   \|/
1217         -------------------------
1218         |  extern register[]    |
1219         -------------------------
1220
1221 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
1222 really clear on what mechanisms they should be using to manipulate that
1223 buffer.   Further, much code assumes, dangerously, that registers are
1224 contigious.  Having got mips-tdep.c to support multiple ABIs, believe
1225 me, that is a bad assumption.  Finally, that register cache layout is
1226 determined by the current remote/local target and _not_ the less
1227 specific target ISA.  In fact, in many cases it is determined by the
1228 somewhat arbitrary layout of the [gG] packets!
1229
1230
1231 How I would like the register file to work is more like:
1232
1233
1234           High GDB
1235              |
1236             \|/
1237        pseudo reg-nr
1238              |
1239         map pseudo <->
1240         random cache
1241         bytes
1242              |
1243             \|/
1244         ------------
1245         | register |
1246         |  cache   |
1247         ------------
1248             /|\
1249              |
1250         map random cache
1251         bytes to target
1252         dependent i-face
1253             /|\
1254              |
1255        target dependent
1256         such as [gG] packet
1257         or ptrace buffer
1258
1259 The main objectives being:
1260
1261         o       a clear separation between the low
1262                 level target and the high level GDB
1263
1264         o       a mechanism that solves the general
1265                 problem of register aliases, overlaps
1266                 etc instead of treating them as optional
1267                 extras that can be wedged in as an after
1268                 thought (that is a reasonable description
1269                 of the current code).
1270
1271                 Identify then solve the hard case and the
1272                 rest just falls out.  GDB solved the easy
1273                 case and then tried to ignore the real
1274                 world :-)
1275
1276         o       a removal of the assumption that the
1277                 mapping between the register cache
1278                 and virtual registers is largely static.
1279                 If you flip the USR/SSR stack register
1280                 select bit in the status-register then
1281                 the corresponding stack registers should
1282                 reflect the change.
1283
1284         o       a mechanism that clearly separates the
1285                 gdb internal register cache from any
1286                 target (not architecture) dependent
1287                 specifics such as [gG] packets.
1288
1289 Of course, like anything, it sounds good in theory.  In reality, it
1290 would have to contend with many<->many relationships at both the
1291 virt<->cache and cache<->target level.  For instance:
1292
1293         virt<->cache
1294                 Modifying an mmx register may involve
1295                 scattering values across both FP and
1296                 mmpx specific parts of a buffer
1297
1298         cache<->target
1299                 When writing back a SP it may need to
1300                 both be written to both SP and USP.
1301
1302
1303 Hmm,
1304
1305 Rather than let this like the last time it was discussed, just slip, I'm
1306 first going to add this e-mail (+ references) to TODO.  I'd then like to
1307 sketch out a broad strategy I think could get us there.
1308
1309
1310 First thing I'd suggest is separating out the ``extern registers[]''
1311 code so that we can at least identify what is using it.  At present
1312 things are scattered across many files.  That way we can at least
1313 pretend that there is a cache instead of a global array :-)
1314
1315 I'd then suggest someone putting up a proposal for the pseudo-reg /
1316 high-level side interface so that code can be adopted to it.  For old
1317 code, initially a blanket rename of write_register_bytes() to
1318 deprecated_write_register_bytes() would help.
1319
1320 Following that would, finaly be the corresponding changes to the target.
1321
1322 --
1323
1324 Check that GDB can handle all BFD architectures (Andrew Cagney)
1325
1326 There should be a test that checks that BFD/GDB are in sync with
1327 regard to architecture changes.  Something like a test that first
1328 queries GDB for all supported architectures and then feeds each back
1329 to GDB..  Anyone interested in learning how to write tests?  :-)
1330
1331 --
1332
1333                 Architectural Change: Multi-arch et al.
1334                 =======================================
1335
1336 The long term objective is to remove all assumptions that there is a
1337 single target with a single address space with a single instruction
1338 set architecture and single application binary interface.
1339
1340 This is an ongoing effort.  The first milestone is to enable
1341 ``multi-arch'' where by all architectural decisions are made at
1342 runtime.
1343
1344 It should be noted that ``gdbarch'' is really ``gdbabi'' and
1345 ``gdbisa''.  Once things are multi-arched breaking that down correctly
1346 will become much easier.
1347
1348 --
1349
1350 GDBARCH cleanup (Andrew Cagney)
1351
1352 The non-generated parts of gdbarch.{sh,h,c} should be separated out
1353 into arch-utils.[hc].
1354
1355 Document that gdbarch_init_ftype could easily fail because it didn't
1356 identify an architecture.
1357
1358 --
1359
1360 Fix BELIEVE_PPC_PROMOTION.  Change it to BELIEVE_PPC_PROMOTION_P?
1361
1362 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
1363 symtab file.
1364
1365 --
1366
1367 Fix target_signal_from_host() etc.
1368
1369 The name is wrong for starters.  ``target_signal'' should probably be
1370 ``gdb_signal''.  ``from_host'' should be ``from_target_signal''.
1371 After that it needs to be multi-arched and made independent of any
1372 host signal numbering.
1373
1374 Once this is done, the signal enum can probably be moved to
1375 include/gdb so that it is available to embedded stubs.
1376
1377 --
1378
1379 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
1380 EXTRA_FRAME_INFO.
1381
1382 This is a barrier to replacing mips_extra_func_info with something
1383 that works with multi-arch.
1384
1385 --
1386
1387 Multi-arch  mips_extra_func_info.
1388
1389 This first needs the alpha to be updated so that it uses ``struct
1390 frame_extra_info''.
1391
1392 --
1393
1394 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
1395
1396 Surely one of them is redundant.
1397
1398 --
1399
1400 Convert ALL architectures to MULTI-ARCH.
1401
1402 --
1403
1404 Select the initial multi-arch ISA / ABI based on --target or similar.
1405
1406 At present the default is based on what ever is first in the BFD
1407 archures table.  It should be determined based on the ``--target=...''
1408 name.
1409
1410 --
1411
1412 Make MIPS pure multi-arch.
1413
1414 It is only at the multi-arch enabled stage.
1415
1416 --
1417
1418 Truly multi-arch.
1419
1420 Enable the code to recognize --enable-targets=.... like BINUTILS does.
1421
1422 Can the tm.h and nm.h files be eliminated by multi-arch.
1423
1424 --
1425
1426         Architectural Change: MI, LIBGDB and scripting languages
1427         ========================================================
1428
1429 See also architectural changes related to the event loop.  LIBGDB
1430 can't be finished until there is a generic event loop being used by
1431 all targets.
1432
1433 The long term objective is it to be possible to integrate GDB into
1434 scripting languages.
1435
1436 --
1437
1438 Implement generic ``(gdb) commmand > file''
1439
1440 Once everything is going through ui_file it should be come fairly
1441 easy.
1442
1443 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1444
1445 --
1446
1447 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1448
1449 gdb_stdtarg is easily confused with gdb_stdarg.
1450
1451 --
1452
1453 Extra ui_file methods - dump.
1454
1455 Very useful for whitebox testing.
1456
1457 --
1458
1459 Eliminate error_begin().
1460
1461 With ui_file, there is no need for the statefull error_begin ()
1462 function.
1463
1464 --
1465
1466 Send normal output to gdb_stdout.
1467 Send error messages to gdb_stderror.
1468 Send debug and log output log gdb_stdlog.
1469
1470 GDB still contains many cases where (f)printf or printf_filtered () is
1471 used when it should be sending the messages to gdb_stderror or
1472 gdb_stdlog.  The thought of #defining printf to something has crossed
1473 peoples minds ;-)
1474
1475 --
1476
1477 Re-do GDB's output pager.
1478
1479 GDB's output pager still relies on people correctly using *_filtered
1480 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1481 Hopefully, with all normal output going to gdb_stdout, the pager can
1482 just look at the ui_file that the output is on and then use that to
1483 decide what to do about paging.  Sounds good in theory.
1484
1485 --
1486
1487 Check/cleanup MI documentation.
1488
1489 The list of commands specified in the documentation needs to be
1490 checked against the mi-cmds.c table in a mechanical way (so that they
1491 two can be kept up-to-date).
1492
1493 --
1494
1495 Convert MI into libgdb
1496
1497 MI provides a text interface into what should be many of the libgdb
1498 functions.  The implementation of those functions should be separated
1499 into the MI interface and the functions proper.  Those functions being
1500 moved to gdb/lib say.
1501
1502 --
1503
1504 Create libgdb.h
1505
1506 The first part can already be found in defs.h.
1507
1508 --
1509
1510 MI's input does not use buffering.
1511
1512 At present the MI interface reads raw characters of from an unbuffered
1513 FD.  This is to avoid several nasty buffer/race conditions.  That code
1514 should be changed so that it registers its self with the event loop
1515 (on the input FD) and then push commands up to MI as they arrive.
1516
1517 The serial code already does this.
1518
1519 --
1520
1521 Make MI interface accessible from existing CLI.
1522
1523 --
1524
1525 Add a breakpoint-edit command to MI.
1526
1527 It would be similar to MI's breakpoint create but would apply to an
1528 existing breakpoint.  It saves the need to delete/create breakpoints
1529 when ever they are changed.
1530
1531 --
1532
1533 Add directory path to MI breakpoint.
1534
1535 That way the GUI's task of finding the file within which the
1536 breakpoint was set is simplified.
1537
1538 --
1539
1540 Add a mechanism to reject certain expression classes to MI
1541
1542 There are situtations where you don't want GDB's expression
1543 parser/evaluator to perform inferior function calls or variable
1544 assignments.  A way of restricting the expression parser so that such
1545 operations are not accepted would be very helpful.
1546
1547 --
1548
1549 Remove sideffects from libgdb breakpoint create function.
1550
1551 The user can use the CLI to create a breakpoint with partial
1552 information - no file (gdb would use the file from the last
1553 breakpoint).
1554
1555 The libgdb interface currently affects that environment which can lead
1556 to confusion when a user is setting breakpoints via both the MI and
1557 the CLI.
1558
1559 This is also a good example of how getting the CLI ``right'' will be
1560 hard.
1561
1562 --
1563
1564 Move gdb_lasterr to ui_out?
1565
1566 The way GDB throws errors and records them needs a re-think.  ui_out
1567 handles the correct output well.  It doesn't resolve what to do with
1568 output / error-messages when things go wrong.
1569
1570 --
1571
1572 do_setshow_command contains a 1024 byte buffer.
1573
1574 The function assumes that there will never be any more than 1024 bytes
1575 of enum.  It should use mem_file.
1576
1577 --
1578
1579 Should struct cmd_list_element . completer take the command as an
1580 argument?
1581
1582 --
1583
1584 Should the bulk of top.c:line_completion_function() be moved to
1585 command.[hc]?  complete_on_cmdlist() and complete_on_enums() could
1586 then be made private.
1587
1588 --
1589
1590 top.c (execute_command): Should a command being valid when the target
1591 is running be made an attribute (predicate) to the command rather than
1592 an explicit set of tests.
1593
1594 --
1595
1596 top.c (execute_command): Should the bulk of this function be moved
1597 into command.[hc] so that top.c doesn't grub around in the command
1598 internals?
1599
1600 --
1601
1602                 Architectural Change: Async
1603                 ===========================
1604
1605 While GDB uses an event loop when prompting the user for input.  That
1606 event loop is not exploited by targets when they allow the target
1607 program to continue.  Typically targets still block in (target_wait())
1608 until the program again halts.
1609
1610 The closest a target comes to supporting full asynchronous mode are
1611 the remote targets ``async'' and ``extended-async''.
1612
1613 --
1614
1615 Asynchronous expression evaluator
1616
1617 Inferior function calls hang GDB.
1618
1619 --
1620
1621 Fix implementation of  ``target xxx''.
1622
1623 At present when the user specifies ``target xxxx'', the CLI maps that
1624 directly onto a target open method.  It is then assumed that the
1625 target open method should do all sorts of complicated things as this
1626 is the only chance it has.  Check how the various remote targets
1627 duplicate the target operations.  Check also how the various targets
1628 behave differently for purely arbitrary reasons.
1629
1630 What should happen is that ``target xxxx'' should call a generic
1631 ``target'' function and that should then co-ordinate the opening of
1632 ``xxxx''.  This becomes especially important when you're trying to
1633 open an asynchronous target that may need to perform background tasks
1634 as part of the ``attach'' phase.
1635
1636 Unfortunately, due to limitations in the old/creaking command.h
1637 interface, that isn't possible.  The function being called isn't told
1638 of the ``xxx'' or any other context information.
1639
1640 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1641 CLI code so that it passes to the callback function (attatched to a
1642 command) useful information such as the actual command and a context
1643 for that command.  Other changes such as making ``struct command''
1644 opaque may also help.
1645
1646 See also:
1647 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1648
1649 --
1650
1651 Make "target xxx" command interruptible.
1652
1653 As things become async this becomes possible.  A target would start
1654 the connect and then return control to the event loop.  A cntrl-c
1655 would notify the target that the operation is to be abandoned and the
1656 target code could respond.
1657
1658 --
1659
1660 Add a "suspend" subcommand of the "continue" command to suspend gdb
1661 while continuing  execution of the subprocess.  Useful when you are
1662 debugging servers and you want to dodge out and initiate a connection
1663 to a server running under gdb.
1664
1665 [hey async!!]
1666
1667 --
1668
1669                         TODO FAQ
1670                         ========
1671
1672 Frequently requested but not approved requests.
1673
1674 --
1675
1676 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1677
1678 The benefits on this one are thought to be marginal - GDBs design
1679 means that unused parameters are very common.  GCC 3.0 will also
1680 include the option -Wno-unused-parameter which means that ``-Wall
1681 -Wno-unused-parameters -Werror'' can be specified.
1682
1683 --
1684
1685
1686
1687                         Legacy Wish List
1688                         ================
1689
1690 This list is not up to date, and opinions vary about the importance or
1691 even desirability of some of the items.  If you do fix something, it
1692 always pays to check the below.
1693
1694 --
1695
1696 @c This does not work (yet if ever).  FIXME.
1697 @c @item --parse=@var{lang} @dots{}
1698 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1699 @c @samp{all} configures @value{GDBN} for all supported languages.  To get a
1700 @c list of all supported languages, omit the argument.  Without this
1701 @c option, @value{GDBN} is configured to parse all supported languages.
1702
1703 --
1704
1705 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1706 is its default value.  Clean this up.
1707
1708 --
1709
1710 It should be possible to use symbols from shared libraries before we know
1711 exactly where the libraries will be loaded.  E.g. "b perror" before running
1712 the program.  This could maybe be done as an extension of the "breakpoint
1713 re-evaluation" after new symbols are loaded.
1714
1715 --
1716
1717 Make single_step() insert and remove breakpoints in one operation.
1718
1719 [If this is talking about having single_step() insert the breakpoints,
1720 run the target then pull the breakpoints then it is wrong.  The
1721 function has to return as control has to eventually be passed back to
1722 the main event loop.]
1723
1724 --
1725
1726 Speed up single stepping by avoiding extraneous ptrace calls.
1727
1728 --
1729
1730 Speed up single stepping by not inserting and removing breakpoints
1731 each time the inferior starts and stops.
1732
1733 Breakpoints should not be inserted and deleted all the time.  Only the
1734 one(s) there should be removed when we have to step over one.  Support
1735 breakpoints that don't have to be removed to step over them.
1736
1737 [this has resulted in numerous debates.  The issue isn't clear cut]
1738
1739 --
1740
1741 Provide "voodoo" debugging of core files.  This creates a zombie
1742 process as a child of the debugger, and loads it up with the data,
1743 stack, and regs of the core file.  This allows you to call functions
1744 in the executable, to manipulate the data in the core file.
1745
1746 [you wish]
1747
1748 --
1749
1750 GDB reopens the source file on every line, as you "next" through it.
1751
1752 [still true? I've a memory of this being fixed]
1753
1754 --
1755
1756 Perhaps "i source" should take an argument like that of "list".
1757
1758 --
1759
1760 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1761 it matches the source line indicated.
1762
1763 --
1764
1765 The prompt at end of screen should accept space as well as CR.
1766
1767 --
1768
1769 Backtrace should point out what the currently selected frame is, in
1770 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1771 ...)" rather than "#3 foo (bar, ...)".
1772
1773 --
1774
1775 "i program" should work for core files, and display more info, like what
1776 actually caused it to die.
1777
1778 --
1779
1780 "x/10i" should shorten the long name, if any, on subsequent lines.
1781
1782 --
1783
1784 "next" over a function that longjumps, never stops until next time you happen
1785 to get to that spot by accident.  E.g. "n" over execute_command which has
1786 an error.
1787
1788 --
1789
1790 "set zeroprint off", don't bother printing members of structs which
1791 are entirely zero.  Useful for those big structs with few useful
1792 members.
1793
1794 --
1795
1796 GDB does four ioctl's for every command, probably switching terminal modes
1797 to/from inferior or for readline or something.
1798
1799 --
1800
1801 terminal_ours versus terminal_inferior: cache state.  Switch should be a noop
1802 if the state is the same, too.
1803
1804 --
1805
1806 "i frame" shows wrong "arglist at" location, doesn't show where the args
1807 should be found, only their actual values.
1808
1809 --
1810
1811 There should be a way for "set" commands to validate the new setting
1812 before it takes effect.
1813
1814 --
1815
1816 "ena d" is ambiguous, why?  "ena delete" seems to think it is a command!
1817
1818 --
1819
1820 i line VAR produces "Line number not known for symbol ``var''.".  I
1821 thought we were stashing that info now!
1822
1823 --
1824
1825 We should be able to write to random files at hex offsets like adb.
1826
1827 --
1828
1829 [elena - delete this]
1830
1831 Handle add_file with separate text, data, and bss addresses.  Maybe
1832 handle separate addresses for each segment in the object file?
1833
1834 --
1835
1836 [Jimb/Elena delete this one]
1837
1838 Handle free_named_symtab to cope with multiply-loaded object files
1839 in a dynamic linking environment.  Should remember the last copy loaded,
1840 but not get too snowed if it finds references to the older copy.
1841
1842 --
1843
1844 [elena delete this also]
1845
1846 Remove all references to:
1847         text_offset
1848         data_offset
1849         text_data_start
1850         text_end
1851         exec_data_offset
1852         ...
1853 now that we have BFD.  All remaining are in machine dependent files.
1854
1855 --
1856
1857 Re-organize help categories into things that tend to fit on a screen
1858 and hang together.
1859
1860 --
1861
1862 Add in commands like ADB's for searching for patterns, etc.  We should
1863 be able to examine and patch raw unsymboled binaries as well in gdb as
1864 we can in adb.  (E.g. increase the timeout in /bin/login without source).
1865
1866 [actually, add ADB interface :-]
1867
1868 --
1869
1870 When doing "step" or "next", if a few lines of source are skipped between
1871 the previous line and the current one, print those lines, not just the
1872 last line of a multiline statement.
1873
1874 --
1875
1876 Handling of "&" address-of operator needs some serious overhaul
1877 for ANSI C and consistency on arrays and functions.
1878   For "float point[15];":
1879 ptype &point[4]   ==> Attempt to take address of non-lvalue.
1880   For "char *malloc();":
1881 ptype malloc      ==> "char *()";  should be same as
1882 ptype &malloc     ==> "char *(*)()"
1883 call printf ("%x\n", malloc)   ==> weird value, should be same as
1884 call printf ("%x\n", &malloc)  ==> correct value
1885
1886 --
1887
1888 Fix dbxread.c symbol reading in the presence of interrupts.  It
1889 currently leaves a cleanup to blow away the entire symbol table when a
1890 QUIT occurs.  (What's wrong with that? -kingdon, 28 Oct 1993).
1891
1892 [I suspect that the grype was that, on a slow system, you might want
1893 to cntrl-c and get just half the symbols and then load the rest later
1894 - scary to be honest]
1895
1896 --
1897
1898 Mipsread.c reads include files depth-first, because the dependencies
1899 in the psymtabs are way too inclusive (it seems to me).  Figure out what
1900 really depends on what, to avoid recursing 20 or 30 times while reading
1901 real symtabs.
1902
1903 --
1904
1905 value_add() should be subtracting the lower bound of arrays, if known,
1906 and possibly checking against the upper bound for error reporting.
1907
1908 --
1909
1910 When listing source lines, check for a preceding \n, to verify that
1911 the file hasn't changed out from under us.
1912
1913 [fixed by some other means I think.  That hack wouldn't actually work
1914 reliably - the file might move such that another \n appears. ]
1915
1916 --
1917
1918 Get all the remote systems (where the protocol allows it) to be able to
1919 stop the remote system when the GDB user types ^C (like remote.c
1920 does).  For ebmon, use ^Ak.
1921
1922 --
1923
1924 Possible feature: A version of the "disassemble" command which shows
1925 both source and assembly code ("set symbol-filename on" is a partial
1926 solution).
1927
1928 [has this been done?  It was certainly done for MI and GDBtk]
1929
1930 --
1931
1932 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1933 using a 0 address for bad purposes internally).
1934
1935 --
1936
1937 Make "info path" and path_command work again (but independent of the
1938 environment either of gdb or that we'll pass to the inferior).
1939
1940 --
1941
1942 Make GDB understand the GCC feature for putting octal constants in
1943 enums.  Make it so overflow on an enum constant does not error_type
1944 the whole type.  Allow arbitrarily large enums with type attributes.
1945 Put all this stuff in the testsuite.
1946
1947 --
1948
1949 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1950 the value in hex; process type attributes).  Add this to the
1951 testsuite.  This way future compilers can add new types and old
1952 versions of GDB can do something halfway reasonable.
1953
1954 --
1955
1956 Fix mdebugread.c:parse_type to do fundamental types right (see
1957 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1958 that the debug format fixes the sizes of these things and it shouldn't
1959 depend on stuff like TARGET_PTR_BIT and so on.  For mdebug, there seem
1960 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1961 be aware of that).  Also use a switch statement for clarity and speed.
1962
1963 --
1964
1965 Investigate adding symbols in target_load--some targets do, some
1966 don't.
1967
1968 --
1969
1970 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1971 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1972 bar.c).
1973
1974 --
1975
1976 Merge xcoffread.c and coffread.c.  Use breakpoint_re_set instead of
1977 fixup_breakpoints.
1978
1979 --
1980
1981 Make a watchpoint which contains a function call an error (it is
1982 broken now, making it work is probably not worth the effort).
1983
1984 --
1985
1986 New test case based on weird.exp but in which type numbers are not
1987 renumbered (thus multiply defining a type).  This currently causes an
1988 infinite loop on "p v_comb".
1989
1990 --
1991
1992 [Hey! Hint Hint Delete Delete!!!]
1993
1994 Fix 386 floating point so that floating point registers are real
1995 registers (but code can deal at run-time if they are missing, like
1996 mips and 68k).  This would clean up "info float" and related stuff.
1997
1998 --
1999
2000 gcc -g -c enummask.c then gdb enummask.o, then "p v".  GDB complains
2001 about not being able to access memory location 0.
2002
2003 -------------------- enummask.c
2004 enum mask
2005 {
2006   ANIMAL = 0,
2007   VEGETABLE = 1,
2008   MINERAL = 2,
2009   BASIC_CATEGORY = 3,
2010
2011   WHITE = 0,
2012   BLUE = 4,
2013   GREEN = 8,
2014   BLACK = 0xc,
2015   COLOR = 0xc,
2016
2017   ALIVE = 0x10,
2018
2019   LARGE = 0x20
2020 } v;
2021
2022 --
2023
2024 If try to modify value in file with "set write off" should give
2025 appropriate error not "cannot access memory at address 0x65e0".
2026
2027 --
2028
2029 Allow core file without exec file on RS/6000.
2030
2031 --
2032
2033 Make sure "shell" with no arguments works right on DOS.
2034
2035 --
2036
2037 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
2038 the same directory can be NFS-mounted on unix or DOS, and work the
2039 same way.
2040
2041 --
2042
2043 [Is this another delete???]
2044
2045 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
2046 get RS/6000 to work right, might not be immediately relevant).
2047
2048 --
2049
2050 Work out some kind of way to allow running the inferior to be done as
2051 a sub-execution of, eg. breakpoint command lists.  Currently running
2052 the inferior interupts any command list execution.  This would require
2053 some rewriting of wait_for_inferior & friends, and hence should
2054 probably be done in concert with the above.
2055
2056 --
2057
2058 Add function arguments to gdb user defined functions.
2059
2060 --
2061
2062 Add convenience variables that refer to exec file, symbol file,
2063 selected frame source file, selected frame function, selected frame
2064 line number, etc.
2065
2066 --
2067
2068 Modify the handling of symbols grouped through BINCL/EINCL stabs to
2069 allocate a partial symtab for each BINCL/EINCL grouping.  This will
2070 seriously decrease the size of inter-psymtab dependencies and hence
2071 lessen the amount that needs to be read in when a new source file is
2072 accessed.
2073
2074 --
2075
2076 Add a command for searching memory, a la adb.  It specifies size,
2077 mask, value, start address.  ADB searches until it finds it or hits
2078 an error (or is interrupted).
2079
2080 --
2081
2082 Remove the range and type checking code and documentation, if not
2083 going to implement.
2084
2085 # Local Variables:
2086 # mode: text
2087 # End: