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