| 1 | If you find inaccuracies in this list, please send mail to |
| 2 | bug-gdb@prep.ai.mit.edu. If you would like to work on any of these, |
| 3 | you should consider sending mail to the same address, to find out |
| 4 | whether anyone else is working on it. |
| 5 | |
| 6 | General To Do List |
| 7 | ------------------ |
| 8 | |
| 9 | This list is probably not up to date, and opinions vary about the |
| 10 | importance or even desirability of some of the items. |
| 11 | |
| 12 | START_INFERIOR_TRAPS_EXPECTED need never be defined to 2, since that |
| 13 | is its default value. Clean this up. |
| 14 | |
| 15 | It should be possible to use symbols from shared libraries before we know |
| 16 | exactly where the libraries will be loaded. E.g. "b perror" before running |
| 17 | the program. This could maybe be done as an extension of the "breakpoint |
| 18 | re-evaluation" after new symbols are loaded. |
| 19 | |
| 20 | Make single_step() insert and remove breakpoints in one operation. |
| 21 | |
| 22 | Speed up single stepping by avoiding extraneous ptrace calls. |
| 23 | |
| 24 | Speed up single stepping by not inserting and removing breakpoints |
| 25 | each time the inferior starts and stops. |
| 26 | |
| 27 | Breakpoints should not be inserted and deleted all the time. Only the |
| 28 | one(s) there should be removed when we have to step over one. Support |
| 29 | breakpoints that don't have to be removed to step over them. |
| 30 | |
| 31 | Update gdbint.texinfo to include doc on the directory structure and |
| 32 | the various tricks of building gdb. |
| 33 | |
| 34 | Do a tutorial in gdb.texinfo on how to do simple things in gdb. |
| 35 | E.g. how to set a breakpoint that just prints something and continues. |
| 36 | How to break on aborts. Etc. |
| 37 | |
| 38 | Provide "voodoo" debugging of core files. This creates a zombie |
| 39 | process as a child of the debugger, and loads it up with the data, |
| 40 | stack, and regs of the core file. This allows you to call functions |
| 41 | in the executable, to manipulate the data in the core file. |
| 42 | |
| 43 | GDB reopens the source file on every line, as you "next" through it. |
| 44 | |
| 45 | Referencing the vtbl member of a struct doesn't work. It prints OK |
| 46 | if you print the struct, but it gets 0 if you try to deref it. |
| 47 | |
| 48 | Persistent command history: A feature where you could save off a list |
| 49 | of the commands you did, so you can edit it into something that will bring |
| 50 | the target to the same place every time you source it. |
| 51 | This would also be useful for automated fast watchpointing; if you go |
| 52 | past the place where it watchpoints, you just start it over again and |
| 53 | do it more carefully. |
| 54 | |
| 55 | Deal with the SunOS 4.0 and 4.1.1 ptrace bug that loses the registers if |
| 56 | the stack is paged out. |
| 57 | |
| 58 | Finish the C++ exception handling stub routines. Lint points them out |
| 59 | as unused statics functions. |
| 60 | |
| 61 | Perhaps "i source" should take an argument like that of "list". |
| 62 | |
| 63 | See if coredep.c's fetch_core_registers can be used on more machines. |
| 64 | E.g. MIPS (mips-xdep.c). |
| 65 | |
| 66 | unpack_double() does not handle IEEE float on the target unless the host |
| 67 | is also IEEE. Death on a vax. |
| 68 | |
| 69 | Set up interface between GDB and INFO so that you can hop into interactive |
| 70 | INFO and back out again. When running under Emacs, should use Emacs |
| 71 | info, else fork the info program. Installation of GDB should install |
| 72 | its texinfo files into the info tree automagically, including the readline |
| 73 | texinfo files. |
| 74 | |
| 75 | "help address" ought to find the "help set print address" entry. |
| 76 | |
| 77 | Remove the VTBL internal guts from printouts of C++ structs, unless |
| 78 | vtblprint is set. |
| 79 | |
| 80 | Remove "at 0xnnnn" from the "b foo" response, if `print address off' and if |
| 81 | it matches the source line indicated. |
| 82 | |
| 83 | The prompt at end of screen should accept space as well as CR. |
| 84 | |
| 85 | Check STORE_RETURN_VALUE on all architectures. Check near it in tm-sparc.h |
| 86 | for other bogosities. |
| 87 | |
| 88 | Check for storage leaks in GDB, I'm sure there are a lot! |
| 89 | |
| 90 | vtblprint of a vtbl should demangle the names it's printing. |
| 91 | |
| 92 | Backtrace should point out what the currently selected frame is, in |
| 93 | its display, perhaps showing "@3 foo (bar, ...)" or ">3 foo (bar, |
| 94 | ...)" rather than "#3 foo (bar, ...)". |
| 95 | |
| 96 | "i program" should work for core files, and display more info, like what |
| 97 | actually caused it to die. |
| 98 | |
| 99 | "x/10i" should shorten the long name, if any, on subsequent lines. |
| 100 | |
| 101 | Check through the code for FIXME comments and fix them. dbxread.c, |
| 102 | blockframe.c, and plenty more. (I count 634 as of 940621 - sts) |
| 103 | |
| 104 | "next" over a function that longjumps, never stops until next time you happen |
| 105 | to get to that spot by accident. E.g. "n" over execute_command which has |
| 106 | an error. |
| 107 | |
| 108 | "set zeroprint off", don't bother printing members of structs which |
| 109 | are entirely zero. Useful for those big structs with few useful |
| 110 | members. |
| 111 | |
| 112 | GDB does four ioctl's for every command, probably switching terminal modes |
| 113 | to/from inferior or for readline or something. |
| 114 | |
| 115 | terminal_ours versus terminal_inferior: cache state. Switch should be a noop |
| 116 | if the state is the same, too. |
| 117 | |
| 118 | ptype $i6 = void??! |
| 119 | |
| 120 | Clean up invalid_float handling so gdb doesn't coredump when it tries to |
| 121 | access a NaN. While this might work on SPARC, other machines are not |
| 122 | configured right. |
| 123 | |
| 124 | "b value_at ; commands ; continue ; end" stops EVERY OTHER TIME! |
| 125 | Then once you enter a command, it does the command, runs two more |
| 126 | times, and then stops again! Bizarre... (This behaviour has been |
| 127 | modified, but it is not yet 100% predictable when e.g. the commands |
| 128 | call functions in the child, and while there, the child is interrupted |
| 129 | with a signal, or hits a breakpoint.) |
| 130 | |
| 131 | help completion, help history should work. |
| 132 | |
| 133 | Check that we can handle stack trace through varargs AND alloca in same |
| 134 | function, on 29K. |
| 135 | |
| 136 | wait_for_inferior loops forever if wait() gives it an error. |
| 137 | |
| 138 | "i frame" shows wrong "arglist at" location, doesn't show where the args |
| 139 | should be found, only their actual values. |
| 140 | |
| 141 | There should be a way for "set" commands to validate the new setting |
| 142 | before it takes effect. |
| 143 | |
| 144 | A mess of floating point opcodes are missing from sparc-opcode.h. |
| 145 | Also, a little program should test the table for bits that are |
| 146 | overspecified or underspecified. E.g. if the must-be-ones bits |
| 147 | and the must-be-zeroes bits leave some fields unexamined, and the format |
| 148 | string leaves them unprinted, then point this out. If multiple |
| 149 | non-alias patterns match, point this out too. Finally, there should |
| 150 | be a sparc-optest.s file that tries each pattern out. This file |
| 151 | should end up coming back the same (modulo transformation comments) |
| 152 | if fed to "gas" then the .o is fed to gdb for disassembly. |
| 153 | |
| 154 | Eliminate all the core_file_command's in all the xdep files. |
| 155 | Eliminate separate declarations of registers[] everywhere. |
| 156 | |
| 157 | "ena d" is ambiguous, why? "ena delete" seems to think it is a command! |
| 158 | |
| 159 | Perhaps move the tdep, xdep, and nat files, into the config |
| 160 | subdirectories. If not, at least straighten out their names so that |
| 161 | they all start with the machine name. |
| 162 | |
| 163 | inferior_status should include stop_print_frame. It won't need to be |
| 164 | reset in wait_for_inferior after bpstat_stop_status call, then. |
| 165 | |
| 166 | i line VAR produces "Line number not known for symbol ``var''.". I |
| 167 | thought we were stashing that info now! |
| 168 | |
| 169 | We should be able to write to random files at hex offsets like adb. |
| 170 | |
| 171 | Make "target xxx" command interruptible. |
| 172 | |
| 173 | Handle add_file with separate text, data, and bss addresses. Maybe |
| 174 | handle separate addresses for each segment in the object file? |
| 175 | |
| 176 | Handle free_named_symtab to cope with multiply-loaded object files |
| 177 | in a dynamic linking environment. Should remember the last copy loaded, |
| 178 | but not get too snowed if it finds references to the older copy. |
| 179 | |
| 180 | Generalize and Standardize the RPC interface to a target program, |
| 181 | improve it beyond the "ptrace" interface, and see if it can become a |
| 182 | standard for remote debugging. (This is talking about the vxworks |
| 183 | interface. Seems unlikely to me that there will be "a standard" for |
| 184 | remote debugging anytime soon --kingdon, 8 Nov 1994). |
| 185 | |
| 186 | Remove all references to: |
| 187 | text_offset |
| 188 | data_offset |
| 189 | text_data_start |
| 190 | text_end |
| 191 | exec_data_offset |
| 192 | ... |
| 193 | now that we have BFD. All remaining are in machine dependent files. |
| 194 | |
| 195 | When quitting with a running program, if a core file was previously |
| 196 | examined, you get "Couldn't read float regs from core file"...if |
| 197 | indeed it can't. generic_mourn_inferior... |
| 198 | |
| 199 | Have remote targets give a warning on a signal argument to |
| 200 | target_resume. Or better yet, extend the protocols so that it works |
| 201 | like it does on the Unix-like systems. |
| 202 | |
| 203 | Sort help and info output. |
| 204 | |
| 205 | Re-organize help categories into things that tend to fit on a screen |
| 206 | and hang together. |
| 207 | |
| 208 | renote-nindy.c handles interrupts poorly; it error()s out of badly |
| 209 | chosen places, e.g. leaving current_frame zero, which causes core dumps |
| 210 | on the next command. |
| 211 | |
| 212 | Add in commands like ADB's for searching for patterns, etc. We should |
| 213 | be able to examine and patch raw unsymboled binaries as well in gdb as |
| 214 | we can in adb. (E.g. increase the timeout in /bin/login without source). |
| 215 | |
| 216 | Those xdep files that call register_addr without defining it are |
| 217 | probably simply broken. When reconfiguring this part of gdb, I could |
| 218 | only make guesses about how to redo some of those files, and I |
| 219 | probably guessed wrong, or left them "for later" when I have a |
| 220 | machine that can attempt to build them. |
| 221 | |
| 222 | When doing "step" or "next", if a few lines of source are skipped between |
| 223 | the previous line and the current one, print those lines, not just the |
| 224 | last line of a multiline statement. |
| 225 | |
| 226 | When searching for C++ superclasses in value_cast in valops.c, we must |
| 227 | not search the "fields", only the "superclasses". There might be a |
| 228 | struct with a field name that matches the superclass name. This can |
| 229 | happen when the struct was defined before the superclass (before the |
| 230 | name became a typedef). |
| 231 | |
| 232 | Handling of "&" address-of operator needs some serious overhaul |
| 233 | for ANSI C and consistency on arrays and functions. |
| 234 | For "float point[15];": |
| 235 | ptype &point[4] ==> Attempt to take address of non-lvalue. |
| 236 | For "char *malloc();": |
| 237 | ptype malloc ==> "char *()"; should be same as |
| 238 | ptype &malloc ==> "char *(*)()" |
| 239 | call printf ("%x\n", malloc) ==> weird value, should be same as |
| 240 | call printf ("%x\n", &malloc) ==> correct value |
| 241 | |
| 242 | Fix dbxread.c symbol reading in the presence of interrupts. It |
| 243 | currently leaves a cleanup to blow away the entire symbol table when a |
| 244 | QUIT occurs. (What's wrong with that? -kingdon, 28 Oct 1993). |
| 245 | |
| 246 | Mipsread.c reads include files depth-first, because the dependencies |
| 247 | in the psymtabs are way too inclusive (it seems to me). Figure out what |
| 248 | really depends on what, to avoid recursing 20 or 30 times while reading |
| 249 | real symtabs. |
| 250 | |
| 251 | value_add() should be subtracting the lower bound of arrays, if known, |
| 252 | and possibly checking against the upper bound for error reporting. |
| 253 | |
| 254 | mipsread.c symbol table allocation and deallocation should be checked. |
| 255 | My suspicion is that it's full of memory leaks. |
| 256 | |
| 257 | SunOS should have a target_lookup_symbol() for common'd things allocated |
| 258 | by the shared library linker ld.so. |
| 259 | |
| 260 | When listing source lines, check for a preceding \n, to verify that |
| 261 | the file hasn't changed out from under us. |
| 262 | |
| 263 | When listing source lines, eat leading whitespace corresponding to the |
| 264 | line-number prefix we print. This avoids long lines wrapping. |
| 265 | |
| 266 | mipsread.c needs to check for old symtabs and psymtabs for the same |
| 267 | files, the way it happens for dbxread.c and coffread.c, for VxWorks |
| 268 | incremental symbol table reloading. |
| 269 | |
| 270 | Get all the remote systems (where the protocol allows it) to be able to |
| 271 | stop the remote system when the GDB user types ^C (like remote.c |
| 272 | does). For ebmon, use ^Ak. |
| 273 | |
| 274 | Possible feature: A version of the "disassemble" command which shows |
| 275 | both source and assembly code ("set symbol-filename on" is a partial |
| 276 | solution). |
| 277 | |
| 278 | investigate "x/s 0" (right now stops early) (I think maybe GDB is |
| 279 | using a 0 address for bad purposes internally). |
| 280 | |
| 281 | Make "info path" and path_command work again (but independent of the |
| 282 | environment either of gdb or that we'll pass to the inferior). |
| 283 | |
| 284 | Make GDB understand the GCC feature for putting octal constants in |
| 285 | enums. Make it so overflow on an enum constant does not error_type |
| 286 | the whole type. Allow arbitrarily large enums with type attributes. |
| 287 | Put all this stuff in the testsuite. |
| 288 | |
| 289 | Make TYPE_CODE_ERROR with a non-zero TYPE_LENGTH more useful (print |
| 290 | the value in hex; process type attributes). Add this to the |
| 291 | testsuite. This way future compilers can add new types and old |
| 292 | versions of GDB can do something halfway reasonable. |
| 293 | |
| 294 | Clean up formatting of "info registers" on MIPS and 88k. See if it |
| 295 | is possible to do this generically across all target architectures. |
| 296 | |
| 297 | GDB gets bfd/core.c and gdb/core.c confused (this should be easy to |
| 298 | repeat even with something more recent than GDB 4.9). |
| 299 | |
| 300 | Check that unmatched RBRAC doesn't abort(). |
| 301 | |
| 302 | Fix mdebugread.c:parse_type to do fundamental types right (see |
| 303 | rs6000_builtin_type in stabsread.c for what "right" is--the point is |
| 304 | that the debug format fixes the sizes of these things and it shouldn't |
| 305 | depend on stuff like TARGET_PTR_BIT and so on. For mdebug, there seem |
| 306 | to be separate bt* codes for 64 bit and 32 bit things, and GDB should |
| 307 | be aware of that). Also use a switch statement for clarity and speed. |
| 308 | |
| 309 | Investigate adding symbols in target_load--some targets do, some |
| 310 | don't. |
| 311 | |
| 312 | Put dirname in psymtabs and change lookup*symtab to use dirname (so |
| 313 | /foo/bar.c works whether compiled by cc /foo/bar.c, or cd /foo; cc |
| 314 | bar.c). |
| 315 | |
| 316 | Merge xcoffread.c and coffread.c. Use breakpoint_re_set instead of |
| 317 | fixup_breakpoints. |
| 318 | |
| 319 | Fix byte order and int size sins in tm-a29k.h |
| 320 | (EXTRACT_RETURN_VALUE). Perhaps should reproduce bug and verify fix |
| 321 | (or perhaps should just fix it...). |
| 322 | |
| 323 | Make a watchpoint on a constant expression an error (or warning |
| 324 | perhaps) |
| 325 | |
| 326 | Make a watchpoint which contains a function call an error (it is |
| 327 | broken now, making it work is probably not worth the effort). |
| 328 | |
| 329 | Re-do calls to signal() in remote.c, and inflow.c (set_sigint_trap and |
| 330 | so on) to be independent of the debugging target, using target_stop to |
| 331 | stop the inferior. Probably the part which is now handled by |
| 332 | interrupt_query in remote.c can be done without any new features in |
| 333 | the debugging target. |
| 334 | |
| 335 | New test case based on weird.exp but in which type numbers are not |
| 336 | renumbered (thus multiply defining a type). This currently causes an |
| 337 | infinite loop on "p v_comb". |
| 338 | |
| 339 | Nuke baseclass_addr. |
| 340 | |
| 341 | Nuke USG define. |
| 342 | |
| 343 | "source file more recent" loses on re-read |
| 344 | |
| 345 | Fix 386 floating point so that floating point registers are real |
| 346 | registers (but code can deal at run-time if they are missing, like |
| 347 | mips and 68k). This would clean up "info float" and related stuff. |
| 348 | |
| 349 | Look at Solaris bug in interrupt.exp. Can get out of syscall with |
| 350 | PRSABORT (syscall will return EINTR) but merely doing that leads to a |
| 351 | "can't read memory" error. |
| 352 | |
| 353 | gcc -g -c enummask.c then gdb enummask.o, then "p v". GDB complains |
| 354 | about not being able to access memory location 0. |
| 355 | |
| 356 | -------------------- enummask.c |
| 357 | enum mask |
| 358 | { |
| 359 | ANIMAL = 0, |
| 360 | VEGETABLE = 1, |
| 361 | MINERAL = 2, |
| 362 | BASIC_CATEGORY = 3, |
| 363 | |
| 364 | WHITE = 0, |
| 365 | BLUE = 4, |
| 366 | GREEN = 8, |
| 367 | BLACK = 0xc, |
| 368 | COLOR = 0xc, |
| 369 | |
| 370 | ALIVE = 0x10, |
| 371 | |
| 372 | LARGE = 0x20 |
| 373 | } v; |
| 374 | |
| 375 | If try to modify value in file with "set write off" should give |
| 376 | appropriate error not "cannot access memory at address 0x65e0". |
| 377 | |
| 378 | Why do we allow a target to omit standard register names (NO_STD_REGS |
| 379 | in tm-z8k.h)? I thought the standard register names were supposed to |
| 380 | be just that, standard. |
| 381 | |
| 382 | Make DEBUG_EXPRESSIONS a maintenance command, dependent on |
| 383 | MAINTENANCE_COMMANDS. |
| 384 | |
| 385 | Allow core file without exec file on RS/6000. |
| 386 | |
| 387 | Make sure "shell" with no arguments works right on DOS. |
| 388 | |
| 389 | Make gdb.ini (as well as .gdbinit) be checked on all platforms, so |
| 390 | the same directory can be NFS-mounted on unix or DOS, and work the |
| 391 | same way. |
| 392 | |
| 393 | cd ~/tmp/<M-?> causes infinite loop (where ~/tmp is a directory). |
| 394 | |
| 395 | Get SECT_OFF_TEXT stuff out of objfile_relocate (might be needed to |
| 396 | get RS/6000 to work right, might not be immediately relevant). |
| 397 | |
| 398 | Clean up add_toc_to_loadinfo |
| 399 | |
| 400 | Think about attached processes and sharing terminal. |
| 401 | |
| 402 | John sez in reference to ignoring errors from tcsegpgrp if attach_flag: |
| 403 | set_tty_state should not have any trouble with attached processes. |
| 404 | Instead, the tty handling should leave the pgrp of the tty alone when |
| 405 | attaching to processes (perhaps pass terminal_init_inferior a flag |
| 406 | saying whether we're attaching). |
| 407 | |
| 408 | PAGE_SIZE redefined warnings on AIX. Probably should be using |
| 409 | BFD_PAGE_SIZE throughout BFD. |
| 410 | |
| 411 | Rewrite proceed, wait_for_inferior, and normal_stop to clean them up. |
| 412 | Suggestions: |
| 413 | |
| 414 | 1) Make each test in wait_for_inferior a seperate subroutine |
| 415 | call. |
| 416 | 2) Combine wait_for_inferior and normal_stop to clean up |
| 417 | communication via global variables. |
| 418 | 3) See if you can find some way to clean up the global |
| 419 | variables that are used; possibly group them by data flow |
| 420 | and information content? |
| 421 | |
| 422 | Work out some kind of way to allow running the inferior to be done as |
| 423 | a sub-execution of, eg. breakpoint command lists. Currently running |
| 424 | the inferior interupts any command list execution. This would require |
| 425 | some rewriting of wait_for_inferior & friends, and hence should |
| 426 | probably be done in concert with the above. |
| 427 | |
| 428 | Add function arguments to gdb user defined functions. |
| 429 | |
| 430 | Add convenience variables that refer to exec file, symbol file, |
| 431 | selected frame source file, selected frame function, selected frame |
| 432 | line number, etc. |
| 433 | |
| 434 | Add a "suspend" subcommand of the "continue" command to suspend gdb |
| 435 | while continuing execution of the subprocess. Useful when you are |
| 436 | debugging servers and you want to dodge out and initiate a connection |
| 437 | to a server running under gdb. |
| 438 | |
| 439 | Add stab information to allow reasonable debugging of inline functions |
| 440 | (possibly they should show up on a stack backtrace? With a note |
| 441 | indicating that they weren't "real"?). |
| 442 | |
| 443 | Modify the naked "until" command to step until past the current source |
| 444 | line, rather than past the current pc value. This is tricky simply |
| 445 | because the low level routines have no way of specifying a multi-line |
| 446 | step range, and there is no way of saying "don't print stuff when we |
| 447 | stop" from above (otherwise could just call step many times). |
| 448 | |
| 449 | Modify the handling of symbols grouped through BINCL/EINCL stabs to |
| 450 | allocate a partial symtab for each BINCL/EINCL grouping. This will |
| 451 | seriously decrease the size of inter-psymtab dependencies and hence |
| 452 | lessen the amount that needs to be read in when a new source file is |
| 453 | accessed. |
| 454 | |
| 455 | Do an "x/i $pc" after each stepi or nexti. |
| 456 | |
| 457 | Modify all of the disassemblers to use printf_filtered to get correct |
| 458 | more filtering. |
| 459 | |
| 460 | Modify gdb to work correctly with Pascal. |
| 461 | |
| 462 | Add a command for searching memory, a la adb. It specifies size, |
| 463 | mask, value, start address. ADB searches until it finds it or hits |
| 464 | an error (or is interrupted). |
| 465 | |
| 466 | The GDB_TARGET_IS_HPPA kludges in partial-stab.h should be zapped sometime |
| 467 | after GCC-2.7 has been released (whenever we decide corner case debugging |
| 468 | problems with GCC-2.6 compiled code are not important). |
| 469 | |
| 470 | # Local Variables: |
| 471 | # mode: text |
| 472 | # End: |