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                 Known problems in GDB 5.0
   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 a follow-on
  12 release.
  13
  14 --
  15
  16 The BFD directory requires bug-fixed AUTOMAKE et.al.
  17
  18 AUTOMAKE 1.4 incorrectly set the TEXINPUTS environment variable.  It
  19 contained the full path to texinfo.tex when it should have only
  20 contained the directory.  The bug has been fixed in the current
  21 AUTOMAKE sources.  Automake snapshots can be found in:
  22         ftp://sourceware.cygnus.com/pub/gdb/snapshots
  23 and     ftp://sourceware.cygnus.com/pub/binutils
  24
  25 --
  26
  27 RFD: infrun.c: No bpstat_stop_status call after proceed over break?
  28 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00665.html
  29
  30 GDB misses watchpoint triggers after proceeding over a breakpoint on
  31 x86 targets.
  32
  33 --
  34
  35 x86 linux GDB and SIGALRM (???)
  36 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00803.html
  37
  38 This problem has been fixed, but a regression test still needs to be
  39 added to the testsuite:
  40 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00309.html
  41
  42 Mark
  43
  44 --
  45
  46 Revised UDP support (was: Re: [Fwd: [patch] UDP transport support])
  47 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00000.html
  48
  49 (Broken) support for GDB's remote protocol across UDP is to be
  50 included in the follow-on release.
  51
  52 --
  53
  54 Can't build IRIX -> arm GDB.
  55 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00356.html
  56
  57 David Whedon writes:
  58 > Now I'm building for an embedded arm target.  If there is a way of turning
  59 > remote-rdi off, I couldn't find it. It looks like it gets built by default
  60 > in gdb/configure.tgt(line 58) Anyway, the build dies in
  61 > gdb/rdi-share/unixcomm.c.  SERPORT1 et. al. never get defined because we
  62 > aren't one of the architectures supported.
  63
  64 --
  65
  66 Problem with weak functions
  67 http://sourceware.cygnus.com/ml/gdb/2000-05/msg00060.html
  68
  69 Dan Nicolaescu writes:
  70 > It seems that gdb-4.95.1  does not display correctly the function when
  71 > stoping in weak functions.
  72 >
  73 > It stops in a function that is defined as weak, not in the function
  74 > that is actualy run...
  75
  76 --
  77
  78 GDB 5.0 doesn't work on Linux/SPARC
  79
  80 --
  81
  82                 Code Cleanups: Next Release
  83                 ===========================
  84
  85 The following are small cleanups that will hopefully be completed by
  86 the follow on to 5.0.
  87
  88 --
  89
  90 Delete macro TARGET_BYTE_ORDER_SELECTABLE.
  91
  92 Patches in the database.
  93
  94 --
  95
  96 Purge PARAMS.
  97
  98 Eliminate all uses of PARAMS in GDB's source code.
  99
 100 --
 101
 102 Fix copyright notices.
 103
 104 Turns out that ``1998-2000'' isn't considered valid :-(
 105
 106 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00467.html
 107
 108 --
 109
 110 Find something better than DEFAULT_BFD_ARCH, DEFAULT_BFD_VEC to
 111 determine the default isa/byte-order.
 112
 113 --
 114
 115 Rely on BFD_BIG_ENDIAN and BFD_LITTLE_ENDIAN instead of host dependant
 116 BIG_ENDIAN and LITTLE_ENDIAN.
 117
 118 --
 119
 120 printcmd.c (print_address_numeric):
 121
 122 NOTE: This assumes that the significant address information is kept in
 123 the least significant bits of ADDR - the upper bits were either zero
 124 or sign extended.  Should ADDRESS_TO_POINTER() or some
 125 ADDRESS_TO_PRINTABLE() be used to do the conversion?
 126
 127 --
 128
 129                 Code Cleanups: General
 130                 ======================
 131
 132 The following are more general cleanups and fixes.  They are not tied
 133 to any specific release.
 134
 135 --
 136
 137 Eliminate more compiler warnings.
 138
 139 Of course there also needs to be the usual debate over which warnings
 140 are valid and how to best go about this.
 141
 142 One method: choose a single option; get agreement that it is
 143 reasonable; try it out to see if there isn't anything silly about it
 144 (-Wunused-parameters is an example of that) then incrementally hack
 145 away.
 146
 147 The other method is to enable all warnings and eliminate them from one
 148 file at a time.
 149
 150 --
 151
 152 Elimination of ``(catch_errors_ftype *) func''.
 153
 154 Like make_cleanup_func it isn't portable.
 155 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00791.html
 156 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00814.html
 157
 158
 159 --
 160
 161 Nuke USG define.
 162
 163 --
 164
 165 [PATCH/5] src/intl/Makefile.in:distclean additions
 166 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00363.html
 167
 168 Do not forget to merge the patch back into the trunk.
 169
 170 --
 171
 172 Rationalize the host-endian code (grep for HOST_BYTE_ORDER).
 173
 174 At present defs.h includes <endian.h> (which is linux specific) yet
 175 almost nothing depends on it.  Suggest "gdb_endian.h" which can also
 176 handle <machine/endian.h> and only include that where it is really
 177 needed.
 178
 179 --
 180
 181 Replace asprintf() calls with xasprintf() calls.
 182
 183 As with things like strdup() most calls to asprintf() don't check the
 184 return value.
 185
 186 --
 187
 188 Replace strsave() + mstrsave() with libiberty:xstrdup().
 189
 190 --
 191
 192 Replace savestring() with something from libiberty.
 193
 194 An xstrldup()? but that would have different semantics.
 195
 196 --
 197
 198 Rationalize use of floatformat_unknown in GDB sources.
 199
 200 Instead of defaulting to floatformat_unknown, should hosts/targets
 201 specify the value explicitly?
 202
 203 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
 204
 205 --
 206
 207 Add a ``name'' member to include/floatformat.h:struct floatformat.
 208 Print that name in gdbarch.c.
 209
 210 --
 211
 212 Sort out the harris mess in include/floatformat.h (it hardwires two
 213 different floating point formats).
 214
 215 --
 216
 217 See of the GDB local floatformat_do_doublest() and libiberty's
 218 floatformat_to_double (which was once GDB's ...) can be merged some
 219 how.
 220
 221 --
 222
 223 Eliminate mmalloc() from GDB.
 224
 225 Also eliminate it from defs.h.
 226
 227 --
 228
 229 Eliminate PTR.  ISO-C allows ``void *''.
 230
 231 --
 232
 233 Eliminate abort ().
 234
 235 GDB should never abort.  GDB should either throw ``error ()'' or
 236 ``internal_error ()''.  Better still GDB should naturally unwind with
 237 an error status.
 238
 239 --
 240
 241 GDB probably doesn't build on FreeBSD pre 2.2.x
 242 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00378.html
 243
 244 Fixes to get FreeBSD working on 2.2.x, 3.x and 4.x caused the code to
 245 suffer bit rot.
 246
 247 --
 248
 249 Deprecate "fg".  Apparently ``fg'' is actually continue.
 250
 251 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00417.html
 252
 253 --
 254
 255 Deprecate current use of ``floatformat_unknown''.
 256
 257 Require all targets to explicitly provide their float format instead
 258 of defaulting to floatformat unknown.  Doing the latter leads to nasty
 259 bugs.
 260
 261 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00447.html
 262
 263 --
 264
 265 Rationalize floatformat_to_double() vs floatformat_to_doublest().
 266
 267 Looks like GDB migrated floatformat_to_double() to libiberty but then
 268 turned around and created a ..._to_doublest() the latter containing
 269 several bug fixes.
 270
 271 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00472.html
 272
 273 --
 274
 275 Move floatformat_ia64_ext to libiberty/include floatformat.[ch].
 276
 277 http://sourceware.cygnus.com/ml/gdb-patches/2000-05/msg00466.html
 278
 279 --
 280
 281 Follow through `make check' with --enable-shared.
 282
 283 When the srcware tree is configured with --enable-shared, the `expect'
 284 program won't run properly.  Jim Wilson found out gdb has a local hack
 285 to set LD_LIBRARY_PATH, but, AFAIK, no other project has been hacked
 286 similarly.
 287
 288 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00845.html
 289
 290 --
 291
 292 The ``maintenance deprecate set endian big'' command doesn't notice
 293 that it is deprecating ``set endian'' and not ``set endian big'' (big
 294 is implemented using an enum).  Is anyone going to notice this?
 295
 296 --
 297
 298 When tab expanding something like ``set arch<tab>'' ignore the
 299 deprecated ``set archdebug'' and expand to ``set architecture''.
 300
 301 --
 302
 303 Eliminate ``arm_register_names[j] = (char *) regnames[j]'' and the
 304 like from arm-tdep.c.
 305
 306 --
 307
 308 Fix uses of ->function.cfunc = set_function().
 309
 310 The command.c code calls sfunc() when a set command.  Rather than
 311 change it suggest fixing the callback function so that it is more
 312 useful.  See:
 313
 314 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
 315
 316 See also ``Fix implementation of  ``target xxx''.'' below.
 317
 318 --
 319
 320 IRIX 3.x support is probably broken.
 321
 322 --
 323
 324 Delete sim/SIM_HAVE_BREAKPOINTS and gdb/SIM_HAS_BREAKPOINTS.
 325 http://sourceware.cygnus.com/ml/gdb-patches/2000-07/msg00042.html
 326
 327 Apart from the d30v, are there any sim/common simulators that make use
 328 of this?
 329
 330 A brief summary of what happended is that sim/common/sim-break.c was
 331 created as a good idea.  It turned out a better idea was to use
 332 SIM_SIGBREAK and have GDB pass back sim_resume (..., SIGBREAK).
 333
 334 --
 335
 336 parse.c:build_parse() has a buffer overrun.
 337
 338 --
 339
 340                         New Features and Fixes
 341                         ======================
 342
 343 These are harder than cleanups but easier than work involving
 344 fundamental architectural change.
 345
 346 --
 347
 348 Add built-by, build-date, tm, xm, nm and anything else into gdb binary
 349 so that you can see how the GDB was created.
 350
 351 Some of these (*m.h) would be added to the generated config.h.  That
 352 in turn would fix a long standing bug where by the build process many
 353 not notice a changed tm.h file.  Since everything depends on config.h,
 354 a change to *m.h forces a change to config.h and, consequently forces
 355 a rebuild.
 356
 357 --
 358
 359 Add an "info bfd" command that displays supported object formats,
 360 similarly to objdump -i.
 361
 362 Is there a command already?
 363
 364 --
 365
 366 Fix ``I'm sorry, Dave, I can't do that.'' from symfile.c.
 367
 368 This requires internationalization.
 369
 370 --
 371
 372 Convert GDB build process to AUTOMAKE.
 373
 374 See also sub-directory configure below.
 375
 376 The current convention is (kind of) to use $(<header>_h) in all
 377 dependency lists.  It isn't done in a consistent way.
 378
 379 --
 380
 381 Cleanup configury support for optional sub-directories.
 382
 383 Check how GCC handles multiple front ends for an example of how things
 384 could work.  A tentative first step is to rationalize things so that
 385 all sub directories are handled in a fashion similar to gdb/mi.
 386
 387 See also automake above.
 388
 389 --
 390
 391 Restructure gdb directory tree so that it avoids any 8.3 and 14
 392 filename problems.
 393
 394 --
 395
 396 Add a transcript mechanism to GDB.
 397
 398 Such a mechanism might log all gdb input and output to a file in a
 399 form that would allow it to be replayed.  It could involve ``gdb
 400 --transcript=FILE'' or it could involve ``(gdb) transcript file''.
 401
 402 --
 403
 404 Can the xdep files be replaced by autoconf?
 405
 406 --
 407
 408 Document trace machinery
 409
 410 --
 411
 412 Document ui-out and ui-file.
 413
 414 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00121.html
 415
 416 --
 417
 418 Update texinfo.tex to latest?
 419
 420
 421
 422 --
 423
 424 Incorporate agentexpr.texi into gdb.texinfo
 425
 426 agentexpr.texi mostly describes the details of the byte code used for
 427 tracepoints, not the internals of the support for this in GDB.  So it
 428 looks like gdb.texinfo is a better place for this information.
 429
 430 http://sourceware.cygnus.com/ml/gdb-patches/2000-04/msg00566.html
 431
 432 --
 433
 434 Document overlay machinery.
 435
 436 --
 437
 438 ``(gdb) catch signal SIGNAL''
 439
 440 Overlaps with ``handle SIGNAL'' but the implied behavour is different.
 441 You can attach commands to a catch but not a handle.  A handle has a
 442 limited number of hardwired actions.
 443
 444 --
 445
 446 Get the TUI working on all platforms.
 447
 448 --
 449
 450 Add support for ``gdb --- PROGRAM ARGS ...''.
 451 Add support for ``gdb -cmd=...''
 452
 453 Along with many variations.  Check:
 454
 455 ????? for a full discussion.
 456
 457 for a discussion.
 458
 459 --
 460
 461 Implement ``(gdb) !ls''.
 462
 463 Which is very different from ``(gdb) ! ls''.  Implementing the latter
 464 is trivial.
 465
 466 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00034.html
 467
 468 --
 469
 470 Change the (char *list[]) to (const char (*)[]) so that dynamic lists can
 471 be passed.
 472
 473 --
 474
 475 When tab expanding something like ``set arch<tab>'' ignore the
 476 deprecated ``set archdebug'' and expand to ``set architecture''.
 477
 478 --
 479
 480 Replace the code that uses the host FPU with an emulator of the target
 481 FPU.
 482
 483 --
 484
 485                         Thread Support
 486                         ==============
 487
 488 --
 489
 490 Generic: lin-thread cannot handle thread exit (Mark Kettenis, Michael
 491 Snyder) http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00525.html
 492
 493 The thread_db assisted debugging code doesn't handle exiting threads
 494 properly, at least in combination with glibc 2.1.3 (the framework is
 495 there, just not the actual code).  There are at least two problems
 496 that prevent this from working.
 497
 498 As an additional reference point, the pre thread_db code did not work
 499 either.
 500
 501 --
 502
 503 GNU/Linux/x86 and random thread signals (and Solaris/SPARC but not
 504 Solaris/x86).
 505 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00336.html
 506
 507 Christopher Blizzard writes:
 508
 509 So, I've done some more digging into this and it looks like Jim
 510 Kingdon has reported this problem in the past:
 511
 512 http://sourceware.cygnus.com/ml/bug-gdb/1999-10/msg00058.html
 513
 514 I can reproduce this problem both with and without Tom's patch.  Has
 515 anyone seen this before?  Maybe have a solution for it hanging around?
 516 :)
 517
 518 There's a test case for this documented at:
 519
 520 when debugging threaded applications you get extra SIGTRAPs
 521 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9565
 522
 523 [There should be a GDB testcase - cagney]
 524
 525 --
 526
 527 GDB5 TOT on unixware 7
 528 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00119.html
 529
 530 Robert Lipe writes:
 531 > I just spun the top of tree of the GDB5 branch on UnixWare 7.  As a
 532 > practical matter, the current thread support is somewhat more annoying
 533 > than when GDB was thread-unaware.
 534
 535 --
 536
 537 Migrate qfThreadInfo packet -> qThreadInfo. (Andrew Cagney)
 538
 539 Add support for packet enable/disable commands with these thread
 540 packets.  General cleanup.
 541
 542 [PATCH] Document the ThreadInfo remote protocol queries
 543 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00832.html
 544
 545 [PATCH] "info threads" queries for remote.c
 546 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00831.html
 547
 548 --
 549
 550                         Language Support
 551                         ================
 552
 553 New languages come onto the scene all the time.
 554
 555 --
 556
 557 Pascal (Pierre Muller, David Taylor)
 558
 559 Pierre Muller has contributed patches for adding Pascal Language
 560 support to GDB.
 561
 562 2 pascal language patches inserted in database
 563 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00521.html
 564
 565 Indent -gnu ?
 566 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00496.html
 567
 568 --
 569
 570 Java (Anthony Green, David Taylor)
 571
 572 Anthony Green has a number of Java patches that did not make it into
 573 the 5.0 release.  The first two are in cvs now, but the third needs
 574 some fixing up before it can go in.
 575
 576 Patch: java tests
 577 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00512.html
 578
 579 Patch: java booleans
 580 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00515.html
 581
 582 Patch: handle N_MAIN stab
 583 http://sourceware.cygnus.com/ml/gdb-patches/2000-q1/msg00527.html
 584
 585 --
 586
 587 [Comming...]
 588
 589 Modify gdb to work correctly with Pascal.
 590
 591 --
 592
 593 Re: Various C++ things
 594
 595 value_headof/value_from_vtable_info are worthless, and should be
 596 removed.  The one place in printcmd.c that uses it should use the RTTI
 597 functions.
 598
 599 RTTI for g++ should be using the typeinfo functions rather than the
 600 vtables.  The typeinfo functions are always at offset 4 from the
 601 beginning of the vtable, and are always right. The vtables will have
 602 weird names like E::VB sometimes.  The typeinfo function will always
 603 be "E type_info function", or somesuch.
 604
 605 value_virtual_fn_field needs to be fixed so there are no failures for
 606 virtual functions for C++ using g++.
 607
 608 Testsuite cases are the major priority right now for C++ support,
 609 since i have to make a lot of changes that could potentially break
 610 each other.
 611
 612 --
 613
 614 Add support for Modula3
 615
 616 Get DEC/Compaq to contribute their Modula-3 support.
 617
 618 --
 619
 620                 Remote Protocol Support
 621                 =======================
 622
 623 --
 624
 625 set/show remote X-packet ...
 626
 627 ``(gdb) help set remote X-packet'' doesn't list the applicable
 628 responses.  The help message needs to be expanded.
 629
 630 --
 631
 632 Remote protocol doco feedback.
 633
 634 Too much feedback to mention needs to be merged in (901660).  Search
 635 for the word ``remote''.
 636
 637
 638 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00023.html
 639 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00056.html
 640 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00382.html
 641
 642 --
 643
 644 GDB doesn't recover gracefully from remote protocol errors.
 645
 646 GDB wasn't checking for NAKs from the remote target.  Instead a NAK is
 647 ignored and a timeout is required before GDB retries.  A pre-cursor to
 648 fixing this this is making GDB's remote protocol packet more robust.
 649
 650 While downloading to a remote protocol target, gdb ignores packet
 651 errors in so far as it will continue to edownload with chunk N+1 even
 652 if chunk N was not correctly sent.  This causes gdb.base/remote.exp to
 653 take a painfully long time to run.  As a PS that test needs to be
 654 fixed so that it builds on 16 bit machines.
 655
 656 --
 657
 658 Add the cycle step command.
 659
 660 http://sourceware.cygnus.com/ml/gdb/2000-q1/msg00237.html
 661
 662 --
 663
 664 Resolve how to scale things to support very large packets.
 665
 666 --
 667
 668 Resolve how to handle a target that changes things like its endianess
 669 on the fly - should it be returned in the ``T'' packet?
 670
 671 Underlying problem is that the register file is target endian.  If the
 672 target endianess changes gdb doesn't know.
 673
 674 --
 675
 676                         Symbol Support
 677                         ==============
 678
 679 If / when GDB starts to support the debugging of multi-processor
 680 (rather than multi-thread) applications the symtab code will need to
 681 be updated a little so that several independant symbol tables are
 682 active at a given time.
 683
 684 The other interesting change is a clarification of the exact meaning
 685 of CORE_ADDR and that has had consequences for a few targets (that
 686 were abusing that data type).
 687
 688 --
 689
 690 Investiagate ways of reducing memory.
 691
 692 --
 693
 694 Investigate ways of improving load time.
 695
 696 --
 697
 698 Get the d10v to use POINTER_TO_ADDRESS and ADDRESS_TO_POINTER.
 699
 700 Consequence of recent symtab clarification.  No marks for figuring out
 701 who maintains the d10v.
 702
 703 --
 704
 705 Get the MIPS to correctly sign extend all address <-> pointer
 706 conversions.
 707
 708 Consequence of recent symtab clarification.  No marks for figuring out
 709 who maintains the MIPS.
 710
 711 --
 712
 713 GDB truncates 64 bit enums.
 714
 715 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00290.html
 716
 717 --
 718
 719                         Testsuite Support
 720                         =================
 721
 722 There are never to many testcases.
 723
 724 --
 725
 726 Better thread testsuite.
 727
 728 --
 729
 730 Better C++ testsuite.
 731
 732 --
 733
 734 Look at adding a GDB specific testsuite directory so that white box
 735 tests of key internals can be added (eg ui_file).
 736
 737 --
 738
 739 Separate out tests that involve the floating point (FP).
 740
 741 (Something for people brining up new targets).  FP and non-fp tests
 742 are combined.  I think there should be set of basic tests that
 743 exercise pure integer support and then a more expanded set that
 744 exercise FP and FP/integer interactions.
 745
 746 As an example, the MIPS, for n32 as problems with passing FP's and
 747 structs.  Since most inferior call tests include FP it is difficult to
 748 determine of the integer tests are ok.
 749
 750 --
 751
 752                 Architectural Changes: General
 753                 ==============================
 754
 755 These are harder than simple cleanups / fixes and, consequently
 756 involve more work.  Typically an Architectural Change will be broken
 757 down into a more digestible set of cleanups and fixes.
 758
 759 --
 760
 761 Cleanup software single step.
 762
 763 At present many targets implement software single step by directly
 764 blatting memory (see rs6000-tdep.c).  Those targets should register
 765 the applicable breakpoints using the breakpoint framework.  Perhaphs a
 766 new internal breakpoint class ``step'' is needed.
 767
 768 --
 769
 770 Replace READ_FP() with FRAME_HANDLE().
 771
 772 READ_FP() is a hangover from the days of the vax when the ABI really
 773 did have a frame pointer register.  Modern architectures typically
 774 construct a virtual frame-handle from the stack pointer and various
 775 other bits of string.
 776
 777 Unfortunatly GDB still treats this synthetic FP register as though it
 778 is real.  That in turn really confuses users (arm and ``print $fp'' VS
 779 ``info registers fp'').  The synthetic FP should be separated out of
 780 the true register set presented to the user.
 781
 782 --
 783
 784 Register Cache Cleanup (below from Andrew Cagney)
 785
 786 I would depict the current register architecture as something like:
 787
 788         High GDB  -->   Low GDB
 789             |              |
 790            \|/            \|/
 791             --- REG NR -----
 792                    |
 793     register + REGISTER_BYTE(reg_nr)
 794                    |
 795                   \|/
 796         -------------------------
 797         |  extern register[]    |
 798         -------------------------
 799
 800 where neither the high (valops.c et.al.) or low gdb (*-tdep.c) are
 801 really clear on what mechanisms they should be using to manipulate that
 802 buffer.   Further, much code assumes, dangerously, that registers are
 803 contigious.  Having got mips-tdep.c to support multiple ABIs, believe
 804 me, that is a bad assumption.  Finally, that register cache layout is
 805 determined by the current remote/local target and _not_ the less
 806 specific target ISA.  In fact, in many cases it is determined by the
 807 somewhat arbitrary layout of the [gG] packets!
 808
 809
 810 How I would like the register file to work is more like:
 811
 812
 813           High GDB
 814              |
 815             \|/
 816        pseudo reg-nr
 817              |
 818         map pseudo <->
 819         random cache
 820         bytes
 821              |
 822             \|/
 823         ------------
 824         | register |
 825         |  cache   |
 826         ------------
 827             /|\
 828              |
 829         map random cache
 830         bytes to target
 831         dependant i-face
 832             /|\
 833              |
 834        target dependant
 835         such as [gG] packet
 836         or ptrace buffer
 837
 838 The main objectives being:
 839
 840         o       a clear separation between the low
 841                 level target and the high level GDB
 842
 843         o       a mechanism that solves the general
 844                 problem of register aliases, overlaps
 845                 etc instead of treating them as optional
 846                 extras that can be wedged in as an after
 847                 thought (that is a reasonable description
 848                 of the current code).
 849
 850                 Identify then solve the hard case and the
 851                 rest just falls out.  GDB solved the easy
 852                 case and then tried to ignore the real
 853                 world :-)
 854
 855         o       a removal of the assumption that the
 856                 mapping between the register cache
 857                 and virtual registers is largely static.
 858                 If you flip the USR/SSR stack register
 859                 select bit in the status-register then
 860                 the corresponding stack registers should
 861                 reflect the change.
 862
 863         o       a mechanism that clearly separates the
 864                 gdb internal register cache from any
 865                 target (not architecture) dependant
 866                 specifics such as [gG] packets.
 867
 868 Of course, like anything, it sounds good in theory.  In reality, it
 869 would have to contend with many<->many relationships at both the
 870 virt<->cache and cache<->target level.  For instance:
 871
 872         virt<->cache
 873                 Modifying an mmx register may involve
 874                 scattering values across both FP and
 875                 mmpx specific parts of a buffer
 876
 877         cache<->target
 878                 When writing back a SP it may need to
 879                 both be written to both SP and USP.
 880
 881
 882 Hmm,
 883
 884 Rather than let this like the last time it was discussed, just slip, I'm
 885 first going to add this e-mail (+ references) to TODO.  I'd then like to
 886 sketch out a broad strategy I think could get us there.
 887
 888
 889 First thing I'd suggest is separating out the ``extern registers[]''
 890 code so that we can at least identify what is using it.  At present
 891 things are scattered across many files.  That way we can at least
 892 pretend that there is a cache instead of a global array :-)
 893
 894 I'd then suggest someone putting up a proposal for the pseudo-reg /
 895 high-level side interface so that code can be adopted to it.  For old
 896 code, initially a blanket rename of write_register_bytes() to
 897 deprecated_write_register_bytes() would help.
 898
 899 Following that would, finaly be the corresponding changes to the target.
 900
 901 --
 902
 903 Check that GDB can handle all BFD architectures (Andrew Cagney)
 904
 905 There should be a test that checks that BFD/GDB are in sync with
 906 regard to architecture changes.  Something like a test that first
 907 queries GDB for all supported architectures and then feeds each back
 908 to GDB..  Anyone interested in learning how to write tests?  :-)
 909
 910 --
 911
 912                 Architectural Change: Multi-arch et al.
 913                 =======================================
 914
 915 The long term objective is to remove all assumptions that there is a
 916 single target with a single address space with a single instruction
 917 set architecture and single application binary interface.
 918
 919 This is an ongoing effort.  The first milestone is to enable
 920 ``multi-arch'' where by all architectural decisions are made at
 921 runtime.
 922
 923 It should be noted that ``gdbarch'' is really ``gdbabi'' and
 924 ``gdbisa''.  Once things are multi-arched breaking that down correctly
 925 will become much easier.
 926
 927 --
 928
 929 GDBARCH cleanup (Andrew Cagney)
 930
 931 The non-generated parts of gdbarch.{sh,h,c} should be separated out
 932 into arch-utils.[hc].
 933
 934 Document that gdbarch_init_ftype could easily fail because it didn't
 935 identify an architecture.
 936
 937 --
 938
 939 Fix BELIEVE_PPC_PROMOTION.  Change it to BELIEVE_PPC_PROMOTION_P?
 940
 941 At present there is still #ifdef BELIEVE_PPC_PROMOTION code in the
 942 symtab file.
 943
 944 --
 945
 946 Fix target_signal_from_host() etc.
 947
 948 The name is wrong for starters.  ``target_signal'' should probably be
 949 ``gdb_signal''.  ``from_host'' should be ``from_target_signal''.
 950 After that it needs to be multi-arched and made independant of any
 951 host signal numbering.
 952
 953 --
 954
 955 Update ALPHA so that it uses ``struct frame_extra_info'' instead of
 956 EXTRA_FRAME_INFO.
 957
 958 This is a barrier to replacing mips_extra_func_info with something
 959 that works with multi-arch.
 960
 961 --
 962
 963 Multi-arch  mips_extra_func_info.
 964
 965 This first needs the alpha to be updated so that it uses ``struct
 966 frame_extra_info''.
 967
 968 --
 969
 970 Rationalize TARGET_SINGLE_FORMAT and TARGET_SINGLE_BIT et al.
 971
 972 Surely one of them is redundant.
 973
 974 --
 975
 976 Convert ALL architectures to MULTI-ARCH.
 977
 978 --
 979
 980 Select the initial multi-arch ISA / ABI based on --target or similar.
 981
 982 At present the default is based on what ever is first in the BFD
 983 archures table.  It should be determined based on the ``--target=...''
 984 name.
 985
 986 --
 987
 988 Make MIPS pure multi-arch.
 989
 990 It is only at the multi-arch enabled stage.
 991
 992 --
 993
 994 Truly multi-arch.
 995
 996 Enable the code to recognize --enable-targets=.... like BINUTILS does.
 997
 998 Can the tm.h and nm.h files be eliminated by multi-arch.
 999
1000 --
1001
1002         Architectural Change: MI, LIBGDB and scripting languages
1003         ========================================================
1004
1005 See also architectural changes related to the event loop.  LIBGDB
1006 can't be finished until there is a generic event loop being used by
1007 all targets.
1008
1009 The long term objective is it to be possible to integrate GDB into
1010 scripting languages.
1011
1012 --
1013
1014 Implement generic ``(gdb) commmand > file''
1015
1016 Once everything is going through ui_file it should be come fairly
1017 easy.
1018
1019 http://sourceware.cygnus.com/ml/gdb/2000-04/msg00104.html
1020
1021 --
1022
1023 Replace gdb_stdtarg with gdb_targout (and possibly gdb_targerr).
1024
1025 gdb_stdtarg is easily confused with gdb_stdarg.
1026
1027 --
1028
1029 Extra ui_file methods - dump.
1030
1031 Very useful for whitebox testing.
1032
1033 --
1034
1035 Eliminate error_begin().
1036
1037 With ui_file, there is no need for the statefull error_begin ()
1038 function.
1039
1040 --
1041
1042 Send normal output to gdb_stdout.
1043 Send error messages to gdb_stderror.
1044 Send debug and log output log gdb_stdlog.
1045
1046 GDB still contains many cases where (f)printf or printf_filtered () is
1047 used when it should be sending the messages to gdb_stderror or
1048 gdb_stdlog.  The thought of #defining printf to something has crossed
1049 peoples minds ;-)
1050
1051 --
1052
1053 Re-do GDB's output pager.
1054
1055 GDB's output pager still relies on people correctly using *_filtered
1056 for gdb_stdout and *_unfiltered for gdb_stdlog / gdb_stderr.
1057 Hopefully, with all normal output going to gdb_stdout, the pager can
1058 just look at the ui_file that the output is on and then use that to
1059 decide what to do about paging.  Sounds good in theory.
1060
1061 --
1062
1063 Check/cleanup MI documentation.
1064
1065 The list of commands specified in the documentation needs to be
1066 checked against the mi-cmds.c table in a mechanical way (so that they
1067 two can be kept up-to-date).
1068
1069 --
1070
1071 Convert MI into libgdb
1072
1073 MI provides a text interface into what should be many of the libgdb
1074 functions.  The implementation of those functions should be separated
1075 into the MI interface and the functions proper.  Those functions being
1076 moved to gdb/lib say.
1077
1078 --
1079
1080 Create libgdb.h
1081
1082 The first part can already be found in defs.h.
1083
1084 --
1085
1086 MI's input does not use buffering.
1087
1088 At present the MI interface reads raw characters of from an unbuffered
1089 FD.  This is to avoid several nasty buffer/race conditions.  That code
1090 should be changed so that it registers its self with the event loop
1091 (on the input FD) and then push commands up to MI as they arrive.
1092
1093 The serial code already does this.
1094
1095 --
1096
1097 Make MI interface accessable from existing CLI.
1098
1099 --
1100
1101 Add a breakpoint-edit command to MI.
1102
1103 It would be similar to MI's breakpoint create but would apply to an
1104 existing breakpoint.  It saves the need to delete/create breakpoints
1105 when ever they are changed.
1106
1107 --
1108
1109 Add directory path to MI breakpoint.
1110
1111 That way the GUI's task of finding the file within which the
1112 breakpoint was set is simplified.
1113
1114 --
1115
1116 Add a mechanism to reject certain expression classes to MI
1117
1118 There are situtations where you don't want GDB's expression
1119 parser/evaluator to perform inferior function calls or variable
1120 assignments.  A way of restricting the expression parser so that such
1121 operations are not accepted would be very helpful.
1122
1123 --
1124
1125 Remove sideffects from libgdb breakpoint create function.
1126
1127 The user can use the CLI to create a breakpoint with partial
1128 information - no file (gdb would use the file from the last
1129 breakpoint).
1130
1131 The libgdb interface currently affects that environment which can lead
1132 to confusion when a user is setting breakpoints via both the MI and
1133 the CLI.
1134
1135 This is also a good example of how getting the CLI ``right'' will be
1136 hard.
1137
1138 --
1139
1140 Move gdb_lasterr to ui_out?
1141
1142 The way GDB throws errors and records them needs a re-think.  ui_out
1143 handles the correct output well.  It doesn't resolve what to do with
1144 output / error-messages when things go wrong.
1145
1146 --
1147
1148 do_setshow_command contains a 1024 byte buffer.
1149
1150 The function assumes that there will never be any more than 1024 bytes
1151 of enum.  It should use mem_file.
1152
1153 --
1154
1155 Should struct cmd_list_element . completer take the command as an
1156 argument?
1157
1158 --
1159
1160 Should the bulk of top.c:line_completion_function() be moved to
1161 command.[hc]?  complete_on_cmdlist() and complete_on_enums() could
1162 then be made private.
1163
1164 --
1165
1166 top.c (execute_command): Should a command being valid when the target
1167 is running be made an attribute (predicate) to the command rather than
1168 an explicit set of tests.
1169
1170 --
1171
1172 top.c (execute_command): Should the bulk of this function be moved
1173 into command.[hc] so that top.c doesn't grub around in the command
1174 internals?
1175
1176 --
1177
1178                 Architectural Change: Async
1179                 ===========================
1180
1181 While GDB uses an event loop when prompting the user for input.  That
1182 event loop is not exploited by targets when they allow the target
1183 program to continue.  Typically targets still block in (target_wait())
1184 until the program again halts.
1185
1186 The closest a target comes to supporting full asynchronous mode are
1187 the remote targets ``async'' and ``extended-async''.
1188
1189 --
1190
1191 Asynchronous expression evaluator
1192
1193 Inferior function calls hang GDB.
1194
1195 --
1196
1197 Fix implementation of  ``target xxx''.
1198
1199 At present when the user specifies ``target xxxx'', the CLI maps that
1200 directly onto a target open method.  It is then assumed that the
1201 target open method should do all sorts of complicated things as this
1202 is the only chance it has.  Check how the various remote targets
1203 duplicate the target operations.  Check also how the various targets
1204 behave differently for purely arbitrary reasons.
1205
1206 What should happen is that ``target xxxx'' should call a generic
1207 ``target'' function and that should then co-ordinate the opening of
1208 ``xxxx''.  This becomes especially important when you're trying to
1209 open an asynchronous target that may need to perform background tasks
1210 as part of the ``attach'' phase.
1211
1212 Unfortunatly, due to limitations in the old/creaking command.h
1213 interface, that isn't possible.  The function being called isn't told
1214 of the ``xxx'' or any other context information.
1215
1216 Consequently a precursor to fixing ``target xxxx'' is to clean up the
1217 CLI code so that it passes to the callback function (attatched to a
1218 command) useful information such as the actual command and a context
1219 for that command.  Other changes such as making ``struct command''
1220 opaque may also help.
1221
1222 See also:
1223 http://sourceware.cygnus.com/ml/gdb-patches/2000-06/msg00062.html
1224
1225 --
1226
1227 Make "target xxx" command interruptible.
1228
1229 As things become async this becomes possible.  A target would start
1230 the connect and then return control to the event loop.  A cntrl-c
1231 would notify the target that the operation is to be abandoned and the
1232 target code could respond.
1233
1234 --
1235
1236 Add a "suspend" subcommand of the "continue" command to suspend gdb
1237 while continuing  execution of the subprocess.  Useful when you are
1238 debugging servers and you want to dodge out and initiate a connection
1239 to a server running under gdb.
1240
1241 [hey async!!]
1242
1243 --
1244
1245                         TODO FAQ
1246                         ========
1247
1248 Frequently requested but not approved requests.
1249
1250 --
1251
1252 Eliminate unused argument warnings using ATTRIBUTE_UNUSED.
1253
1254 The benefits on this one are thought to be marginal - GDBs design
1255 means that unused parameters are very common.  GCC 3.0 will also
1256 include the option -Wno-unused-parameter which means that ``-Wall
1257 -Wno-unused-parameters -Werror'' can be specified.
1258
1259 --
1260
1261
1262
1263                         Legacy Wish List
1264                         ================
1265
1266 This list is not up to date, and opinions vary about the importance or
1267 even desirability of some of the items.  If you do fix something, it
1268 always pays to check the below.
1269
1270 --
1271
1272 @c This does not work (yet if ever).  FIXME.
1273 @c @item --parse=@var{lang} @dots{}
1274 @c Configure the @value{GDBN} expression parser to parse the listed languages.
1275 @c @samp{all} configures @value{GDBN} for all supported languages.  To get a
1276 @c list of all supported languages, omit the argument.  Without this
1277 @c option, @value{GDBN} is configured to parse all supported languages.
1278
1279 --
1280
1281 START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that
1282 is its default value.  Clean this up.
1283
1284 --
1285
1286 It should be possible to use symbols from shared libraries before we know
1287 exactly where the libraries will be loaded.  E.g. "b perror" before running
1288 the program.  This could maybe be done as an extension of the "breakpoint
1289 re-evaluation" after new symbols are loaded.
1290
1291 --
1292
1293 Make single_step() insert and remove breakpoints in one operation.
1294
1295 [If this is talking about having single_step() insert the breakpoints,
1296 run the target then pull the breakpoints then it is wrong.  The
1297 function has to return as control has to eventually be passed back to
1298 the main event loop.]
1299
1300 --
1301
1302 Speed up single stepping by avoiding extraneous ptrace calls.
1303
1304 --
1305
1306 Speed up single stepping by not inserting and removing breakpoints
1307 each time the inferior starts and stops.
1308
1309 Breakpoints should not be inserted and deleted all the time.  Only the
1310 one(s) there should be removed when we have to step over one.  Support
1311 breakpoints that don't have to be removed to step over them.
1312
1313 [this has resulted in numerous debates.  The issue isn't clear cut]
1314
1315 --
1316
1317 Provide "voodoo" debugging of core files.  This creates a zombie
1318 process as a child of the debugger, and loads it up with the data,
1319 stack, and regs of the core file.  This allows you to call functions
1320 in the executable, to manipulate the data in the core file.
1321
1322 [you wish]
1323
1324 --
1325
1326 GDB reopens the source file on every line, as you "next" through it.
1327
1328 [still true? I've a memory of this being fixed]
1329
1330 --
1331
1332 Perhaps "i source" should take an argument like that of "list".
1333
1334 --
1335
1336 Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if
1337 it matches the source line indicated.
1338
1339 --
1340
1341 The prompt at end of screen should accept space as well as CR.
1342
1343 --
1344
1345 Backtrace should point out what the currently selected frame is, in
1346 its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar,
1347 ...)" rather than "#3 foo (bar, ...)".
1348
1349 --
1350
1351 "i program" should work for core files, and display more info, like what
1352 actually caused it to die.
1353
1354 --
1355
1356 "x/10i" should shorten the long name, if any, on subsequent lines.
1357
1358 --
1359
1360 "next" over a function that longjumps, never stops until next time you happen
1361 to get to that spot by accident.  E.g. "n" over execute_command which has
1362 an error.
1363
1364 --
1365
1366 "set zeroprint off", don't bother printing members of structs which
1367 are entirely zero.  Useful for those big structs with few useful
1368 members.
1369
1370 --
1371
1372 GDB does four ioctl's for every command, probably switching terminal modes
1373 to/from inferior or for readline or something.
1374
1375 --
1376
1377 terminal_ours versus terminal_inferior: cache state.  Switch should be a noop
1378 if the state is the same, too.
1379
1380 --
1381
1382 "i frame" shows wrong "arglist at" location, doesn't show where the args
1383 should be found, only their actual values.
1384
1385 --
1386
1387 There should be a way for "set" commands to validate the new setting
1388 before it takes effect.
1389
1390 --
1391
1392 "ena d" is ambiguous, why?  "ena delete" seems to think it is a command!
1393
1394 --
1395
1396 i line VAR produces "Line number not known for symbol ``var''.".  I
1397 thought we were stashing that info now!
1398
1399 --
1400
1401 We should be able to write to random files at hex offsets like adb.
1402
1403 --
1404
1405 [elena - delete this]
1406
1407 Handle add_file with separate text, data, and bss addresses.  Maybe
1408 handle separate addresses for each segment in the object file?
1409
1410 --
1411
1412 [Jimb/Elena delete this one]
1413
1414 Handle free_named_symtab to cope with multiply-loaded object files
1415 in a dynamic linking environment.  Should remember the last copy loaded,
1416 but not get too snowed if it finds references to the older copy.
1417
1418 --
1419
1420 [elena delete this also]
1421
1422 Remove all references to:
1423         text_offset
1424         data_offset
1425         text_data_start
1426         text_end
1427         exec_data_offset
1428         ...
1429 now that we have BFD.  All remaining are in machine dependent files.
1430
1431 --
1432
1433 Re-organize help categories into things that tend to fit on a screen
1434 and hang together.
1435
1436 --
1437
1438 Add in commands like ADB's for searching for patterns, etc.  We should
1439 be able to examine and patch raw unsymboled binaries as well in gdb as
1440 we can in adb.  (E.g. increase the timeout in /bin/login without source).
1441
1442 [actually, add ADB interface :-]
1443
1444 --
1445
1446 When doing "step" or "next", if a few lines of source are skipped between
1447 the previous line and the current one, print those lines, not just the
1448 last line of a multiline statement.
1449
1450 --
1451
1452 Handling of "&" address-of operator needs some serious overhaul
1453 for ANSI C and consistency on arrays and functions.
1454   For "float point[15];":
1455 ptype &point[4]   ==> Attempt to take address of non-lvalue.
1456   For "char *malloc();":
1457 ptype malloc      ==> "char *()";  should be same as
1458 ptype &malloc     ==> "char *(*)()"
1459 call printf ("%x\n", malloc)   ==> weird value, should be same as
1460 call printf ("%x\n", &malloc)  ==> correct value
1461
1462 --
1463
1464 Fix dbxread.c symbol reading in the presence of interrupts.  It
1465 currently leaves a cleanup to blow away the entire symbol table when a
1466 QUIT occurs.  (What's wrong with that? -kingdon, 28 Oct 1993).
1467
1468 [I suspect that the grype was that, on a slow system, you might want
1469 to cntrl-c and get just half the symbols and then load the rest later
1470 - scary to be honest]
1471
1472 --
1473
1474 Mipsread.c reads include files depth-first, because the dependencies
1475 in the psymtabs are way too inclusive (it seems to me).  Figure out what
1476 really depends on what, to avoid recursing 20 or 30 times while reading
1477 real symtabs.
1478
1479 --
1480
1481 value_add() should be subtracting the lower bound of arrays, if known,
1482 and possibly checking against the upper bound for error reporting.
1483
1484 --
1485
1486 When listing source lines, check for a preceding \n, to verify that
1487 the file hasn't changed out from under us.
1488
1489 [fixed by some other means I think.  That hack wouldn't actually work
1490 reliably - the file might move such that another \n appears. ]
1491
1492 --
1493
1494 Get all the remote systems (where the protocol allows it) to be able to
1495 stop the remote system when the GDB user types ^C (like remote.c
1496 does).  For ebmon, use ^Ak.
1497
1498 --
1499
1500 Possible feature: A version of the "disassemble" command which shows
1501 both source and assembly code ("set symbol-filename on" is a partial
1502 solution).
1503
1504 [has this been done?  It was certainly done for MI and GDBtk]
1505
1506 --
1507
1508 investigate "x/s 0" (right now stops early) (I think maybe GDB is
1509 using a 0 address for bad purposes internally).
1510
1511 --
1512
1513 Make "info path" and path_command work again (but independent of the
1514 environment either of gdb or that we'll pass to the inferior).
1515
1516 --
1517
1518 Make GDB understand the GCC feature for putting octal constants in
1519 enums.  Make it so overflow on an enum constant does not error_type
1520 the whole type.  Allow arbitrarily large enums with type attributes.
1521 Put all this stuff in the testsuite.
1522
1523 --
1524
1525 Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print
1526 the value in hex; process type attributes).  Add this to the
1527 testsuite.  This way future compilers can add new types and old
1528 versions of GDB can do something halfway reasonable.
1529
1530 --
1531
1532 Fix mdebugread.c:parse_type to do fundamental types right (see
1533 rs6000_builtin_type in stabsread.c for what "right" is--the point is
1534 that the debug format fixes the sizes of these things and it shouldn't
1535 depend on stuff like TARGET_PTR_BIT and so on.  For mdebug, there seem
1536 to be separate bt* codes for 64 bit and 32 bit things, and GDB should
1537 be aware of that).  Also use a switch statement for clarity and speed.
1538
1539 --
1540
1541 Investigate adding symbols in target_load--some targets do, some
1542 don't.
1543
1544 --
1545
1546 Put dirname in psymtabs and change lookup*symtab to use dirname (so
1547 /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc
1548 bar.c).
1549
1550 --
1551
1552 Merge xcoffread.c and coffread.c.  Use breakpoint_re_set instead of
1553 fixup_breakpoints.
1554
1555 --
1556
1557 Make a watchpoint which contains a function call an error (it is
1558 broken now, making it work is probably not worth the effort).
1559
1560 --
1561
1562 New test case based on weird.exp but in which type numbers are not
1563 renumbered (thus multiply defining a type).  This currently causes an
1564 infinite loop on "p v_comb".
1565
1566 --
1567
1568 [Hey! Hint Hint Delete Delete!!!]
1569
1570 Fix 386 floating point so that floating point registers are real
1571 registers (but code can deal at run-time if they are missing, like
1572 mips and 68k).  This would clean up "info float" and related stuff.
1573
1574 --
1575
1576 gcc -g -c enummask.c then gdb enummask.o, then "p v".  GDB complains
1577 about not being able to access memory location 0.
1578
1579 -------------------- enummask.c
1580 enum mask
1581 {
1582   ANIMAL = 0,
1583   VEGETABLE = 1,
1584   MINERAL = 2,
1585   BASIC_CATEGORY = 3,
1586
1587   WHITE = 0,
1588   BLUE = 4,
1589   GREEN = 8,
1590   BLACK = 0xc,
1591   COLOR = 0xc,
1592
1593   ALIVE = 0x10,
1594
1595   LARGE = 0x20
1596 } v;
1597
1598 --
1599
1600 If try to modify value in file with "set write off" should give
1601 appropriate error not "cannot access memory at address 0x65e0".
1602
1603 --
1604
1605 Allow core file without exec file on RS/6000.
1606
1607 --
1608
1609 Make sure "shell" with no arguments works right on DOS.
1610
1611 --
1612
1613 Make gdb.ini (as well as .gdbinit) be checked on all platforms, so
1614 the same directory can be NFS-mounted on unix or DOS, and work the
1615 same way.
1616
1617 --
1618
1619 [Is this another delete???]
1620
1621 Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to
1622 get RS/6000 to work right, might not be immediately relevant).
1623
1624 --
1625
1626 Work out some kind of way to allow running the inferior to be done as
1627 a sub-execution of, eg. breakpoint command lists.  Currently running
1628 the inferior interupts any command list execution.  This would require
1629 some rewriting of wait_for_inferior & friends, and hence should
1630 probably be done in concert with the above.
1631
1632 --
1633
1634 Add function arguments to gdb user defined functions.
1635
1636 --
1637
1638 Add convenience variables that refer to exec file, symbol file,
1639 selected frame source file, selected frame function, selected frame
1640 line number, etc.
1641
1642 --
1643
1644 Modify the handling of symbols grouped through BINCL/EINCL stabs to
1645 allocate a partial symtab for each BINCL/EINCL grouping.  This will
1646 seriously decrease the size of inter-psymtab dependencies and hence
1647 lessen the amount that needs to be read in when a new source file is
1648 accessed.
1649
1650 --
1651
1652 Add a command for searching memory, a la adb.  It specifies size,
1653 mask, value, start address.  ADB searches until it finds it or hits
1654 an error (or is interrupted).
1655
1656 --
1657
1658 Remove the range and type checking code and documentation, if not
1659 going to implement.
1660
1661 # Local Variables:
1662 # mode: text
1663 # End: