Commit | Line | Data |
---|---|---|
dd3b648e RP |
1 | |
2 | gdb bug list | |
3 | John Gilmore, gnu@cygnus.com | |
4 | ||
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. | |
7 | ||
8 | ||
9 | Update the TODO list with all the lists of gdb bugs lying around on paper. | |
10 | ||
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. | |
13 | ||
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. | |
18 | ||
19 | Make single_step() insert and remove breakpoints in one operation. | |
20 | ||
21 | Speed up single stepping by avoiding extraneous ptrace calls. | |
22 | ||
23 | Speed up single stepping by not inserting and removing breakpoints | |
24 | each time the inferior starts and stops. | |
25 | ||
26 | Speed up watchpoints by not single-stepping them, but do something | |
27 | faster like single-line execution. | |
28 | ||
29 | Update gdb.texinfo to include doc on the directory structure and | |
30 | the various tricks of building gdb. | |
31 | ||
32 | Do a tutorial in gdb.texinfo on how to do simple things in gdb. | |
33 | E.g. how to set a breakpoint that just prints something and continues. | |
34 | How to break on aborts. Etc. | |
35 | ||
36 | Do a "new features" section for release 4. | |
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. Sun wants 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 Sun ptrace bug that loses the registers if the stack is | |
56 | paged out. | |
57 | ||
58 | Finish the C++ exception handling stub routines. Lint points them out | |
59 | as unused statics functions. | |
60 | ||
61 | "i source" only shows you info about files that it can read. When it | |
62 | can't read a file and complains, you can't see any info about it, like | |
63 | where it was compiled. Perhaps "i source" should take an argument | |
64 | like that of "list". | |
65 | ||
66 | See if coredep.c's fetch_core_registers can be used on more machines. | |
67 | E.g. MIPS (mips-xdep.c). | |
68 | ||
69 | coredep.c is completely broken. Needs work just to compile, it uses | |
70 | "u" and doesn't declare it, etc. | |
71 | ||
72 | unpack_double() does not handle IEEE float on the target unless the host | |
73 | is also IEEE. Death on a vax. | |
74 | ||
75 | Test cross-debugging Unix-to-Unix. | |
76 | ||
77 | Check the RAPP remote protocol. What is it? It's in Makefile.dist | |
78 | and one ChangeLog entry. | |
79 | ||
80 | Set up interface between GDB and INFO so that you can hop into interactive | |
81 | INFO and back out again. When running under Emacs, should use Emacs | |
82 | info, else fork the info program. Installation of GDB should install | |
83 | its texinfo files into the info tree automagically, including the readline | |
84 | texinfo files.. | |
85 | ||
86 | Improve backtrace output to avoid line wraps. Prettify it. | |
87 | ||
88 | "help address" ought to find the "help set addressprint" entry. | |
89 | ||
90 | Remove the VTBL internal guts from printouts of C++ structs, unless | |
91 | vtblprint is set. | |
92 | ||
93 | Remove "at 0xnnnn" from the "b foo" response, if !addressprint and if | |
94 | it matches the source line indicated. | |
95 | ||
96 | The prompt at end of screen should accept space as well as CR. | |
97 | ||
98 | "List" should put you into a pseudo-"more" where you can hit space | |
99 | to get more, forever to eof. | |
100 | ||
101 | Check STORE_RETURN_VALUE on all architectures. Check near it in tm-sparc.h | |
102 | for other bogosities. | |
103 | ||
104 | Check for storage leaks in GDB, I'm sure there are a lot! | |
105 | ||
106 | vtblprint of a vtbl should demangle the names it's printing. | |
107 | ||
108 | Backtrace should point out what the currently selected frame is, in its | |
109 | display, perhaps showing ">3 foo (bar, ...)" rather than "#3 foo (bar, ...)". | |
110 | ||
111 | "i program" should work for core files, and display more info, like what | |
112 | actually caused it to die. | |
113 | ||
114 | Hitting ^Z to an inferior doesn't work right, it takes several continues | |
115 | to make it actually go. | |
116 | ||
117 | "i fun" doesn't show misc function vector symbols. | |
118 | ||
119 | "x/10i" should shorten the long name, if any, on subsequent lines. | |
120 | ||
121 | Check through the code for FIXME comments and fix them. dbxread.c, | |
122 | blockframe.c, and plenty more. | |
123 | ||
124 | "next" over a function that longjumps, never stops until next time you happen | |
125 | to get to that spot by accident. E.g. "n" over execute_command which has | |
126 | an error. | |
127 | ||
128 | Watchpoints seem not entirely reliable. | |
129 | ||
130 | "set zeroprint off", don't bother printing members of structs which are entirely | |
131 | zero. Useful for those big structs with few useful members. | |
132 | ||
133 | GDB does four ioctl's for every command, probably switching terminal modes | |
134 | to/from inferior or for readline or something. | |
135 | ||
136 | terminal_ours versus terminal_inferior: cache state. Switch should be a noop | |
137 | if the state is the same, too. | |
138 | ||
139 | ptype $i6 = void??! | |
140 | ||
141 | Clean up invalid_float handling so gdb doesn't coredump when it tries to | |
142 | access a NaN. While this might work on SPARC, other machines are not | |
143 | configured right. | |
144 | ||
145 | "b value_at ; commands ; continue ; end" stops EVERY OTHER TIME! | |
146 | Then once you enter a command, it does the command, runs two more | |
147 | times, and then stops again! Bizarre... (This behaviour has been | |
148 | modified, but it is not yet 100% predictable when e.g. the commands | |
149 | call functions in the child, and while there, the child is interrupted | |
150 | with a signal, or hits a breakpoint.) | |
151 | ||
152 | Symbol completion with TAB does not unmangle names! | |
153 | ||
154 | help completion, help history should work. | |
155 | ||
156 | Symbol completion doesn't handle e.g. W::f. (symtab.c, | |
157 | make_symbol_completion_list). | |
158 | ||
159 | AMD version: ^C should do ^Ak to stop ebmon. | |
160 | ||
161 | Check that we can handle stack trace through varargs AND alloca in same | |
162 | function, on 29K. | |
163 | ||
164 | wait_for_inferior loops forever if wait() gives it an error. | |
165 | ||
166 | "i frame" arg formatting sucks. Should wrap lines. | |
167 | "bt" arg formatting needs the same treatment . | |
168 | ||
169 | "i frame" shows wrong "arglist at" location, doesn't show where the args | |
170 | should be found, only their actual values. | |
171 | ||
172 | Symbolic display of addrs, (& disassembly prefixes), don't show static | |
173 | fns, e.g. enable_command in gdb. | |
174 | ||
175 | 'ptype yylval' ==> "union YYSTYPE { ..... }". However, it is not a | |
176 | union YYSTYPE, but is simply a YYSTYPE, which is a typedef for an | |
177 | unnamed union. | |
178 | ||
179 | "show all" should work. | |
180 | ||
181 | There should be a way for "set" commands to validate the new setting | |
182 | before it takes effect. | |
183 | ||
184 | The "display" command should become the "always" command, e.g. | |
185 | "always print XXX" | |
186 | "always p/xxx XXX" | |
187 | "always echo foo" | |
188 | "always call XXX" | |
189 | "always x/i $pc", etc. | |
190 | ||
191 | A mess of floating point opcodes are missing from sparc-opcode.h. | |
192 | Also, a little program should test the table for bits that are | |
193 | overspecified or underspecified. E.g. if the must-be-ones bits | |
194 | and the must-be-zeroes bits leave some fields unexamined, and the format | |
195 | string leaves them unprinted, then point this out. If multiple | |
196 | non-alias patterns match, point this out too. Finally, there should | |
197 | be a sparc-optest.s file that tries each pattern out. This file | |
198 | should end up coming back the same (modulo transformation comments) | |
199 | if fed to "gas" then the .o is fed to gdb for disassembly. | |
200 | ||
201 | Merge the xxx-opcode.h files with gas again... | |
202 | ||
203 | Eliminate all the core_file_command's in all the xdep files. | |
204 | Eliminate separate declarations of registers[] everywhere. | |
205 | ||
206 | "ena d" is ambiguous, why? "ena delete" seems to think it is a command! | |
207 | ||
208 | Line numbers are off in some spots. In proceed() at 1st "oneproc = 1", | |
209 | it seems to run that statement, but it doesn't actually. | |
210 | ||
211 | Perhaps the tdep and xdep files, and the tm and xm files, into a config | |
212 | subdirectory. If not, at least straighten out their names so that | |
213 | they all start with the machine name. | |
214 | ||
215 | inferior_status should include stop_print_frame. It won't need to be | |
216 | reset in wait_for_inferior after bpstat_stop_status call, then. | |
217 | ||
218 | i line VAR produces "Line number not known for symbol ``var''.". I | |
219 | thought we were stashing that info now! | |
220 | ||
221 | Make sure we can handle executables with no symbol info, e.g. /bin/csh. | |
222 | ||
223 | We should be able to write to executables that aren't running. | |
224 | ||
225 | We should be able to write to random files at hex offsets like adb. | |
226 | ||
227 | Tiemann: It is very painful to look at fp registers that hold | |
228 | double precision values. GDB is happy to show them to you as single | |
229 | precision, but you cannot look at them as doubles. Perhaps casting | |
230 | should be changed to make this work; or maybe a new "set" option that | |
231 | sets the default fp precision to single, double, or quad. This is not | |
232 | urgent, but would be nice to get into GDB 4.0. | |
233 | ||
234 | Make "target xxx" command interruptible. | |
235 | ||
236 | Handle add_file with separate text, data, and bss addresses. Maybe | |
237 | handle separate addresses for each segment in the object file? | |
238 | ||
239 | Handle free_named_symtab to cope with multiply-loaded object files | |
240 | in a dynamic linking environment. Should remember the last copy loaded, | |
241 | but not get too snowed if it finds references to the older copy. | |
242 | ||
243 | Implement have_memory, have_stack, have_registers, have_execution. | |
244 | Memory: core, exec, child, vxworks even without child. | |
245 | stack: core, child, vxworks with child | |
246 | registers: core, child, vxworks with child | |
247 | execution: child, vxworks with child. | |
248 | ||
249 | The original BFD core dump reading routine would itself coredump when fed | |
250 | a garbage file as a core file. Does the current one? | |
251 | ||
252 | Breakpoints should not be inserted and deleted all the time. Only the | |
253 | one(s) there should be removed when we have to step over one. Support | |
254 | breakpoints that don't have to be removed to step over them. | |
255 | ||
256 | Stop reading stop_registers! | |
257 | ||
258 | Generalize and Standardize the RPC interface to a target program, | |
259 | improve it beyond the "ptrace" interface, and see if it can become a standard | |
260 | for remote debugging. Is WRS interested in donating their target-end | |
261 | code? | |
262 | ||
263 | Remove all references to: | |
264 | text_offset | |
265 | data_offset | |
266 | text_data_start | |
267 | text_end | |
268 | exec_data_offset | |
269 | ... | |
270 | now that we have BFD. All remaining are in machine dependent files. | |
271 | ||
272 | When quitting with a running program, if a core file was previously | |
273 | examined, you get "Couldn't read float regs from core file"...if | |
274 | indeed it can't. generic_mourn_inferior... | |
275 | ||
276 | ... | |
277 | ||
278 | Check signal argument to remote proceed's and error if set. | |
279 | ||
280 | Handle floating point registers in core files under BFD. Currently | |
281 | they are punted. | |
282 | ||
283 | Sort help and info output. | |
284 | ||
285 | Re-organize help categories into things that tend to fit on a screen | |
286 | and hang together. | |
287 | ||
288 | When trying to print source lines but you can't find the file, | |
289 | print the file name and line number, and leave it selected anyway | |
290 | so "i source" will show it. | |
291 | ||
292 | renote-nindy.c handles interrupts poorly; it error()s out of badly | |
293 | chosen places, e.g. leaving current_frame zero, which causes core dumps | |
294 | on the next command. | |
295 | ||
296 | Add in commands like ADB's for searching for patterns, etc. We should | |
297 | be able to examine and patch raw unsymboled binaries as well in gdb as | |
298 | we can in adb. (E.g. increase the timeout in /bin/login without source). | |
299 | ||
300 | Those xdep files that call register_addr without defining it are | |
301 | probably simply broken. When reconfiguring this part of gdb, I could | |
302 | only make guesses about how to redo some of those files, and I | |
303 | probably guessed wrong, or left them "for later" when I have a | |
304 | machine that can attempt to build them. | |
305 | ||
306 | Use the complain() mechanism for handling all the error() calls in dbxread.c, | |
307 | and in similar situations in coffread.c and mipsread.c. | |
308 | ||
309 | When doing "step" or "next", if a few lines of source are skipped between | |
310 | the previous line and the current one, print those lines, not just the | |
311 | last line of a multiline statement. | |
312 | ||
313 | When searching for C++ superclasses in value_cast in valops.c, we must | |
314 | not search the "fields", only the "superclasses". There might be a | |
315 | struct with a field name that matches the superclass name. This can | |
316 | happen when the struct was defined before the superclass (before the | |
317 | name became a typedef). | |
318 | ||
36b9d39c JG |
319 | Handling of "&" address-of operator needs some serious overhaul |
320 | for ANSI C and consistency on arrays and functions. | |
321 | For "float point[15];": | |
dd3b648e RP |
322 | ptype &point[4] ==> Attempt to take address of non-lvalue. |
323 | p &point[4] ==> Dereferences point[4] rather than giving you point+4. | |
36b9d39c JG |
324 | For "char *malloc();": |
325 | ptype malloc ==> "char *()"; should be same as | |
326 | ptype &malloc ==> "char *(*)()" | |
327 | call printf ("%x\n", malloc) ==> wierd value, should be same as | |
328 | call printf ("%x\n", &malloc) ==> correct value | |
dd3b648e RP |
329 | |
330 | Fix symbol reading in the presence of interrupts. It currently leaves a | |
331 | cleanup to blow away the entire symbol table when a QUIT occurs. | |
332 |