3 John Gilmore, gnu@cygnus.com
5 This bug list is probably not up to date or accurate, but it reflects
6 some known bugs in gdb, if you are into bug-hunting.
9 Update the TODO list with all the lists of gdb bugs lying around on paper.
11 "share" command should not need to be manually run. It should be run
12 as soon as possible, automatically, both on "run" and on core files.
14 It should be possible to use symbols from shared libraries before we know
15 exactly where the libraries will be loaded. E.g. "b perror" before running
16 the program. This could maybe be done as an extension of the "breakpoint
17 re-evaluation" after new symbols are loaded.
19 Make single_step() insert and remove breakpoints in one operation.
21 Speed up single stepping by avoiding extraneous ptrace calls.
23 Speed up single stepping by not inserting and removing breakpoints
24 each time the inferior starts and stops.
26 Speed up watchpoints by not single-stepping them, but do something
27 faster like single-line execution. Speed them up tremendously on
28 machines that have watchpoint registers.
30 Update gdb.texinfo to include doc on the directory structure and
31 the various tricks of building gdb.
33 Do a tutorial in gdb.texinfo on how to do simple things in gdb.
34 E.g. how to set a breakpoint that just prints something and continues.
35 How to break on aborts. Etc.
37 Provide "voodoo" debugging of core files. This creates a zombie
38 process as a child of the debugger, and loads it up with the data,
39 stack, and regs of the core file. This allows you to call functions
40 in the executable, to manipulate the data in the core file.
42 GDB reopens the source file on every line, as you "next" through it.
44 Referencing the vtbl member of a struct doesn't work. It prints OK
45 if you print the struct, but it gets 0 if you try to deref it.
47 Persistent command history: A feature where you could save off a list
48 of the commands you did, so you can edit it into something that will bring
49 the target to the same place every time you source it. Sun wants it.
50 This would also be useful for automated fast watchpointing; if you go
51 past the place where it watchpoints, you just start it over again and
54 Deal with the Sun ptrace bug that loses the registers if the stack is
57 Finish the C++ exception handling stub routines. Lint points them out
58 as unused statics functions.
60 "i source" only shows you info about files that it can read. When it
61 can't read a file and complains, you can't see any info about it, like
62 where it was compiled. Perhaps "i source" should take an argument
65 See if coredep.c's fetch_core_registers can be used on more machines.
66 E.g. MIPS (mips-xdep.c).
68 coredep.c is completely broken. Needs work just to compile, it uses
69 "u" and doesn't declare it, etc.
71 unpack_double() does not handle IEEE float on the target unless the host
72 is also IEEE. Death on a vax.
74 Test cross-debugging Unix-to-Unix.
76 Check the RAPP remote protocol. What is it? It's in Makefile.dist
77 and one ChangeLog entry.
79 Set up interface between GDB and INFO so that you can hop into interactive
80 INFO and back out again. When running under Emacs, should use Emacs
81 info, else fork the info program. Installation of GDB should install
82 its texinfo files into the info tree automagically, including the readline
85 Improve backtrace output to avoid line wraps. Prettify it.
87 "help address" ought to find the "help set addressprint" entry.
89 Remove the VTBL internal guts from printouts of C++ structs, unless
92 Remove "at 0xnnnn" from the "b foo" response, if !addressprint and if
93 it matches the source line indicated.
95 The prompt at end of screen should accept space as well as CR.
97 "List" should put you into a pseudo-"more" where you can hit space
98 to get more, forever to eof.
100 Check STORE_RETURN_VALUE on all architectures. Check near it in tm-sparc.h
101 for other bogosities.
103 Check for storage leaks in GDB, I'm sure there are a lot!
105 vtblprint of a vtbl should demangle the names it's printing.
107 Backtrace should point out what the currently selected frame is, in its
108 display, perhaps showing ">3 foo (bar, ...)" rather than "#3 foo (bar, ...)".
110 "i program" should work for core files, and display more info, like what
111 actually caused it to die.
113 Hitting ^Z to an inferior doesn't work right, it takes several continues
114 to make it actually go.
116 "i fun" doesn't show misc function vector symbols.
118 "x/10i" should shorten the long name, if any, on subsequent lines.
120 Check through the code for FIXME comments and fix them. dbxread.c,
121 blockframe.c, and plenty more.
123 "next" over a function that longjumps, never stops until next time you happen
124 to get to that spot by accident. E.g. "n" over execute_command which has
127 Watchpoints seem not entirely reliable.
129 "set zeroprint off", don't bother printing members of structs which are entirely
130 zero. Useful for those big structs with few useful members.
132 GDB does four ioctl's for every command, probably switching terminal modes
133 to/from inferior or for readline or something.
135 terminal_ours versus terminal_inferior: cache state. Switch should be a noop
136 if the state is the same, too.
140 Clean up invalid_float handling so gdb doesn't coredump when it tries to
141 access a NaN. While this might work on SPARC, other machines are not
144 "b value_at ; commands ; continue ; end" stops EVERY OTHER TIME!
145 Then once you enter a command, it does the command, runs two more
146 times, and then stops again! Bizarre... (This behaviour has been
147 modified, but it is not yet 100% predictable when e.g. the commands
148 call functions in the child, and while there, the child is interrupted
149 with a signal, or hits a breakpoint.)
151 Symbol completion with TAB does not unmangle names!
153 help completion, help history should work.
155 Symbol completion doesn't handle e.g. W::f. (symtab.c,
156 make_symbol_completion_list).
158 Check that we can handle stack trace through varargs AND alloca in same
161 wait_for_inferior loops forever if wait() gives it an error.
163 "i frame" arg formatting sucks. Should wrap lines.
164 "bt" arg formatting needs the same treatment .
166 "i frame" shows wrong "arglist at" location, doesn't show where the args
167 should be found, only their actual values.
169 Symbolic display of addrs, (& disassembly prefixes), don't show static
170 fns, e.g. enable_command in gdb.
172 'ptype yylval' ==> "union YYSTYPE { ..... }". However, it is not a
173 union YYSTYPE, but is simply a YYSTYPE, which is a typedef for an
176 "show all" should work.
178 There should be a way for "set" commands to validate the new setting
179 before it takes effect.
181 The "display" command should become the "always" command, e.g.
186 "always x/i $pc", etc.
188 A mess of floating point opcodes are missing from sparc-opcode.h.
189 Also, a little program should test the table for bits that are
190 overspecified or underspecified. E.g. if the must-be-ones bits
191 and the must-be-zeroes bits leave some fields unexamined, and the format
192 string leaves them unprinted, then point this out. If multiple
193 non-alias patterns match, point this out too. Finally, there should
194 be a sparc-optest.s file that tries each pattern out. This file
195 should end up coming back the same (modulo transformation comments)
196 if fed to "gas" then the .o is fed to gdb for disassembly.
198 Merge the xxx-opcode.h files with gas again...
200 Eliminate all the core_file_command's in all the xdep files.
201 Eliminate separate declarations of registers[] everywhere.
203 "ena d" is ambiguous, why? "ena delete" seems to think it is a command!
205 Line numbers are off in some spots. In proceed() at 1st "oneproc = 1",
206 it seems to run that statement, but it doesn't actually.
208 Perhaps the tdep and xdep files, and the tm and xm files, into a config
209 subdirectory. If not, at least straighten out their names so that
210 they all start with the machine name.
212 inferior_status should include stop_print_frame. It won't need to be
213 reset in wait_for_inferior after bpstat_stop_status call, then.
215 i line VAR produces "Line number not known for symbol ``var''.". I
216 thought we were stashing that info now!
218 Make sure we can handle executables with no symbol info, e.g. /bin/csh.
220 We should be able to write to executables that aren't running.
222 We should be able to write to random files at hex offsets like adb.
224 Tiemann: It is very painful to look at fp registers that hold
225 double precision values. GDB is happy to show them to you as single
226 precision, but you cannot look at them as doubles. Perhaps casting
227 should be changed to make this work; or maybe a new "set" option that
228 sets the default fp precision to single, double, or quad. This is not
229 urgent, but would be nice to get into GDB 4.0.
231 Make "target xxx" command interruptible.
233 Handle add_file with separate text, data, and bss addresses. Maybe
234 handle separate addresses for each segment in the object file?
236 Handle free_named_symtab to cope with multiply-loaded object files
237 in a dynamic linking environment. Should remember the last copy loaded,
238 but not get too snowed if it finds references to the older copy.
240 Implement have_memory, have_stack, have_registers, have_execution.
241 Memory: core, exec, child, vxworks even without child.
242 stack: core, child, vxworks with child
243 registers: core, child, vxworks with child
244 execution: child, vxworks with child.
246 The original BFD core dump reading routine would itself coredump when fed
247 a garbage file as a core file. Does the current one?
249 Breakpoints should not be inserted and deleted all the time. Only the
250 one(s) there should be removed when we have to step over one. Support
251 breakpoints that don't have to be removed to step over them.
253 Generalize and Standardize the RPC interface to a target program,
254 improve it beyond the "ptrace" interface, and see if it can become a standard
255 for remote debugging. Is WRS interested in donating their target-end
258 Remove all references to:
265 now that we have BFD. All remaining are in machine dependent files.
267 When quitting with a running program, if a core file was previously
268 examined, you get "Couldn't read float regs from core file"...if
269 indeed it can't. generic_mourn_inferior...
271 Check signal argument to remote proceed's and error if set.
273 Sort help and info output.
275 Re-organize help categories into things that tend to fit on a screen
278 renote-nindy.c handles interrupts poorly; it error()s out of badly
279 chosen places, e.g. leaving current_frame zero, which causes core dumps
282 Add in commands like ADB's for searching for patterns, etc. We should
283 be able to examine and patch raw unsymboled binaries as well in gdb as
284 we can in adb. (E.g. increase the timeout in /bin/login without source).
286 Those xdep files that call register_addr without defining it are
287 probably simply broken. When reconfiguring this part of gdb, I could
288 only make guesses about how to redo some of those files, and I
289 probably guessed wrong, or left them "for later" when I have a
290 machine that can attempt to build them.
292 Use the complain() mechanism for handling all the error() calls in dbxread.c,
293 and in similar situations in coffread.c and mipsread.c.
295 When doing "step" or "next", if a few lines of source are skipped between
296 the previous line and the current one, print those lines, not just the
297 last line of a multiline statement.
299 When searching for C++ superclasses in value_cast in valops.c, we must
300 not search the "fields", only the "superclasses". There might be a
301 struct with a field name that matches the superclass name. This can
302 happen when the struct was defined before the superclass (before the
303 name became a typedef).
305 Handling of "&" address-of operator needs some serious overhaul
306 for ANSI C and consistency on arrays and functions.
307 For "float point[15];":
308 ptype &point[4] ==> Attempt to take address of non-lvalue.
309 p &point[4] ==> Dereferences point[4] rather than giving you point+4.
310 For "char *malloc();":
311 ptype malloc ==> "char *()"; should be same as
312 ptype &malloc ==> "char *(*)()"
313 call printf ("%x\n", malloc) ==> wierd value, should be same as
314 call printf ("%x\n", &malloc) ==> correct value
316 Fix dbxread.c symbol reading in the presence of interrupts. It currently
317 leaves a cleanup to blow away the entire symbol table when a QUIT occurs.
319 Mipsread.c reads include files depth-first, because the dependencies
320 in the psymtabs are way too inclusive (it seems to me). Figure out what
321 really depends on what, to avoid recursing 20 or 30 times while reading
324 value_add() should be subtracting the lower bound of arrays, if known,
325 and possibly checking against the upper bound for error reporting.
327 mipsread.c symbol table allocation and deallocation should be checked.
328 My suspicion is that it's full of memory leaks.
330 SunOS should have a target_lookup_symbol() for common'd things allocated
331 by the shared library linker ld.so.
333 When listing source lines, check for a preceding \n, to verify that
334 the file hasn't changed out from under us.
336 When listing source lines, eat leading whitespace corresponding to the
337 line-number prefix we print. This avoids long lines wrapping.
339 mipsread.c needs to check for old symtabs and psymtabs for the same
340 files, the way it happens for dbxread.c and coffread.c, for VxWorks
341 incremental symbol table reloading.
343 When attached to a non-child process, ^C or other signals are not
344 propagated to the child. Do this in the GDB signal handler, using
345 target_kill(). AMD version: ^C should do ^Ak to stop ebmon.