Commit | Line | Data |
---|---|---|
a933dad1 | 1 | Debugging GNU Emacs |
70d9a9cd | 2 | Copyright (c) 1985, 2000, 2001, 2005 Free Software Foundation, Inc. |
a933dad1 DL |
3 | |
4 | Permission is granted to anyone to make or distribute verbatim copies | |
5 | of this document as received, in any medium, provided that the | |
6 | copyright notice and permission notice are preserved, | |
7 | and that the distributor grants the recipient permission | |
8 | for further redistribution as permitted by this notice. | |
9 | ||
10 | Permission is granted to distribute modified versions | |
11 | of this document, or of portions of it, | |
12 | under the above conditions, provided also that they | |
13 | carry prominent notices stating who last changed them. | |
14 | ||
437368fe EZ |
15 | [People who debug Emacs on Windows using native Windows debuggers |
16 | should read the Windows-specific section near the end of this | |
17 | document.] | |
18 | ||
42a3c627 | 19 | ** When you debug Emacs with GDB, you should start it in the directory |
7483daa1 RS |
20 | where the executable was made. That directory has a .gdbinit file |
21 | that defines various "user-defined" commands for debugging Emacs. | |
42a3c627 | 22 | |
70d9a9cd DK |
23 | ** When you are trying to analyze failed assertions, it will be |
24 | essential to compile Emacs either completely without optimizations or | |
25 | at least (when using GCC) with the -fno-crossjumping option. Failure | |
26 | to do so may make the compiler recycle the same abort call for all | |
27 | assertions in a given function, rendering the stack backtrace useless | |
28 | for identifying the specific failed assertion. | |
29 | ||
42a3c627 | 30 | ** It is a good idea to run Emacs under GDB (or some other suitable |
437368fe EZ |
31 | debugger) *all the time*. Then, when Emacs crashes, you will be able |
32 | to debug the live process, not just a core dump. (This is especially | |
33 | important on systems which don't support core files, and instead print | |
34 | just the registers and some stack addresses.) | |
35 | ||
42a3c627 | 36 | ** If Emacs hangs, or seems to be stuck in some infinite loop, typing |
437368fe EZ |
37 | "kill -TSTP PID", where PID is the Emacs process ID, will cause GDB to |
38 | kick in, provided that you run under GDB. | |
39 | ||
40 | ** Getting control to the debugger | |
a933dad1 | 41 | |
3102e429 | 42 | `Fsignal' is a very useful place to put a breakpoint in. |
a933dad1 DL |
43 | All Lisp errors go through there. |
44 | ||
3102e429 | 45 | It is useful, when debugging, to have a guaranteed way to return to |
eb55f651 | 46 | the debugger at any time. When using X, this is easy: type C-z at the |
3102e429 EZ |
47 | window where Emacs is running under GDB, and it will stop Emacs just |
48 | as it would stop any ordinary program. When Emacs is running in a | |
49 | terminal, things are not so easy. | |
50 | ||
51 | The src/.gdbinit file in the Emacs distribution arranges for SIGINT | |
52 | (C-g in Emacs) to be passed to Emacs and not give control back to GDB. | |
53 | On modern POSIX systems, you can override that with this command: | |
54 | ||
7718638c | 55 | handle SIGINT stop nopass |
3102e429 EZ |
56 | |
57 | After this `handle' command, SIGINT will return control to GDB. If | |
58 | you want the C-g to cause a QUIT within Emacs as well, omit the | |
59 | `nopass'. | |
60 | ||
61 | A technique that can work when `handle SIGINT' does not is to store | |
62 | the code for some character into the variable stop_character. Thus, | |
a933dad1 DL |
63 | |
64 | set stop_character = 29 | |
65 | ||
66 | makes Control-] (decimal code 29) the stop character. | |
67 | Typing Control-] will cause immediate stop. You cannot | |
68 | use the set command until the inferior process has been started. | |
69 | Put a breakpoint early in `main', or suspend the Emacs, | |
70 | to get an opportunity to do the set command. | |
71 | ||
8ef597fe NR |
72 | When Emacs is running in a terminal, it is useful to use a separate terminal |
73 | for the debug session. This can be done by starting Emacs as usual, then | |
74 | attaching to it from gdb with the `attach' command which is explained in the | |
75 | node "Attach" of the GDB manual. | |
76 | ||
a933dad1 DL |
77 | ** Examining Lisp object values. |
78 | ||
79 | When you have a live process to debug, and it has not encountered a | |
80 | fatal error, you can use the GDB command `pr'. First print the value | |
81 | in the ordinary way, with the `p' command. Then type `pr' with no | |
82 | arguments. This calls a subroutine which uses the Lisp printer. | |
83 | ||
1ce9f40a KS |
84 | You can also use `pp value' to print the emacs value directly. |
85 | ||
86 | Note: It is not a good idea to try `pr' or `pp' if you know that Emacs | |
87 | is in deep trouble: its stack smashed (e.g., if it encountered SIGSEGV | |
88 | due to stack overflow), or crucial data structures, such as `obarray', | |
437368fe EZ |
89 | corrupted, etc. In such cases, the Emacs subroutine called by `pr' |
90 | might make more damage, like overwrite some data that is important for | |
91 | debugging the original problem. | |
92 | ||
3102e429 EZ |
93 | Also, on some systems it is impossible to use `pr' if you stopped |
94 | Emacs while it was inside `select'. This is in fact what happens if | |
95 | you stop Emacs while it is waiting. In such a situation, don't try to | |
96 | use `pr'. Instead, use `s' to step out of the system call. Then | |
97 | Emacs will be between instructions and capable of handling `pr'. | |
a933dad1 | 98 | |
3102e429 EZ |
99 | If you can't use `pr' command, for whatever reason, you can fall back |
100 | on lower-level commands. Use the `xtype' command to print out the | |
101 | data type of the last data value. Once you know the data type, use | |
102 | the command that corresponds to that type. Here are these commands: | |
a933dad1 DL |
103 | |
104 | xint xptr xwindow xmarker xoverlay xmiscfree xintfwd xboolfwd xobjfwd | |
105 | xbufobjfwd xkbobjfwd xbuflocal xbuffer xsymbol xstring xvector xframe | |
106 | xwinconfig xcompiled xcons xcar xcdr xsubr xprocess xfloat xscrollbar | |
107 | ||
108 | Each one of them applies to a certain type or class of types. | |
109 | (Some of these types are not visible in Lisp, because they exist only | |
110 | internally.) | |
111 | ||
112 | Each x... command prints some information about the value, and | |
113 | produces a GDB value (subsequently available in $) through which you | |
114 | can get at the rest of the contents. | |
115 | ||
437368fe | 116 | In general, most of the rest of the contents will be additional Lisp |
a933dad1 DL |
117 | objects which you can examine in turn with the x... commands. |
118 | ||
437368fe EZ |
119 | Even with a live process, these x... commands are useful for |
120 | examining the fields in a buffer, window, process, frame or marker. | |
121 | Here's an example using concepts explained in the node "Value History" | |
aa1f38cd NR |
122 | of the GDB manual to print values associated with the variable |
123 | called frame. First, use these commands: | |
437368fe EZ |
124 | |
125 | cd src | |
126 | gdb emacs | |
aa1f38cd | 127 | b set_frame_buffer_list |
177c0ea7 | 128 | r -q |
437368fe | 129 | |
8ef597fe | 130 | Then Emacs hits the breakpoint: |
437368fe EZ |
131 | |
132 | (gdb) p frame | |
aa1f38cd | 133 | $1 = 139854428 |
437368fe EZ |
134 | (gdb) xtype |
135 | Lisp_Vectorlike | |
136 | PVEC_FRAME | |
137 | (gdb) xframe | |
aa1f38cd | 138 | $2 = (struct frame *) 0x8560258 |
437368fe EZ |
139 | (gdb) p *$ |
140 | $3 = { | |
aa1f38cd NR |
141 | size = 1073742931, |
142 | next = 0x85dfe58, | |
143 | name = 140615219, | |
437368fe EZ |
144 | [...] |
145 | } | |
146 | (gdb) p $3->name | |
aa1f38cd | 147 | $4 = 140615219 |
437368fe EZ |
148 | |
149 | Now we can use `pr' to print the name of the frame: | |
150 | ||
151 | (gdb) pr | |
152 | "emacs@steenrod.math.nwu.edu" | |
153 | ||
154 | The Emacs C code heavily uses macros defined in lisp.h. So suppose | |
155 | we want the address of the l-value expression near the bottom of | |
156 | `add_command_key' from keyboard.c: | |
157 | ||
158 | XVECTOR (this_command_keys)->contents[this_command_key_count++] = key; | |
159 | ||
160 | XVECTOR is a macro, and therefore GDB does not know about it. | |
161 | GDB cannot evaluate "p XVECTOR (this_command_keys)". | |
162 | ||
163 | However, you can use the xvector command in GDB to get the same | |
164 | result. Here is how: | |
165 | ||
166 | (gdb) p this_command_keys | |
167 | $1 = 1078005760 | |
168 | (gdb) xvector | |
169 | $2 = (struct Lisp_Vector *) 0x411000 | |
170 | 0 | |
171 | (gdb) p $->contents[this_command_key_count] | |
172 | $3 = 1077872640 | |
173 | (gdb) p &$ | |
174 | $4 = (int *) 0x411008 | |
175 | ||
176 | Here's a related example of macros and the GDB `define' command. | |
177 | There are many Lisp vectors such as `recent_keys', which contains the | |
178 | last 100 keystrokes. We can print this Lisp vector | |
179 | ||
180 | p recent_keys | |
181 | pr | |
182 | ||
183 | But this may be inconvenient, since `recent_keys' is much more verbose | |
184 | than `C-h l'. We might want to print only the last 10 elements of | |
185 | this vector. `recent_keys' is updated in keyboard.c by the command | |
186 | ||
187 | XVECTOR (recent_keys)->contents[recent_keys_index] = c; | |
188 | ||
189 | So we define a GDB command `xvector-elts', so the last 10 keystrokes | |
177c0ea7 | 190 | are printed by |
437368fe EZ |
191 | |
192 | xvector-elts recent_keys recent_keys_index 10 | |
193 | ||
194 | where you can define xvector-elts as follows: | |
195 | ||
196 | define xvector-elts | |
197 | set $i = 0 | |
198 | p $arg0 | |
199 | xvector | |
200 | set $foo = $ | |
201 | while $i < $arg2 | |
177c0ea7 | 202 | p $foo->contents[$arg1-($i++)] |
437368fe EZ |
203 | pr |
204 | end | |
205 | document xvector-elts | |
206 | Prints a range of elements of a Lisp vector. | |
207 | xvector-elts v n i | |
208 | prints `i' elements of the vector `v' ending at the index `n'. | |
209 | end | |
210 | ||
211 | ** Getting Lisp-level backtrace information within GDB | |
212 | ||
3102e429 EZ |
213 | The most convenient way is to use the `xbacktrace' command. This |
214 | shows the names of the Lisp functions that are currently active. | |
437368fe EZ |
215 | |
216 | If that doesn't work (e.g., because the `backtrace_list' structure is | |
217 | corrupted), type "bt" at the GDB prompt, to produce the C-level | |
218 | backtrace, and look for stack frames that call Ffuncall. Select them | |
219 | one by one in GDB, by typing "up N", where N is the appropriate number | |
220 | of frames to go up, and in each frame that calls Ffuncall type this: | |
221 | ||
222 | p *args | |
223 | pr | |
224 | ||
225 | This will print the name of the Lisp function called by that level | |
226 | of function calling. | |
227 | ||
228 | By printing the remaining elements of args, you can see the argument | |
229 | values. Here's how to print the first argument: | |
177c0ea7 | 230 | |
437368fe EZ |
231 | p args[1] |
232 | pr | |
233 | ||
234 | If you do not have a live process, you can use xtype and the other | |
235 | x... commands such as xsymbol to get such information, albeit less | |
236 | conveniently. For example: | |
237 | ||
238 | p *args | |
239 | xtype | |
240 | ||
241 | and, assuming that "xtype" says that args[0] is a symbol: | |
242 | ||
177c0ea7 | 243 | xsymbol |
437368fe EZ |
244 | |
245 | ** Debugging what happens while preloading and dumping Emacs | |
246 | ||
247 | Type `gdb temacs' and start it with `r -batch -l loadup dump'. | |
248 | ||
249 | If temacs actually succeeds when running under GDB in this way, do not | |
250 | try to run the dumped Emacs, because it was dumped with the GDB | |
251 | breakpoints in it. | |
252 | ||
253 | ** Debugging `temacs' | |
254 | ||
255 | Debugging `temacs' is useful when you want to establish whether a | |
256 | problem happens in an undumped Emacs. To run `temacs' under a | |
257 | debugger, type "gdb temacs", then start it with `r -batch -l loadup'. | |
258 | ||
259 | ** If you encounter X protocol errors | |
260 | ||
261 | Try evaluating (x-synchronize t). That puts Emacs into synchronous | |
262 | mode, where each Xlib call checks for errors before it returns. This | |
263 | mode is much slower, but when you get an error, you will see exactly | |
264 | which call really caused the error. | |
265 | ||
ec4054f0 EZ |
266 | You can start Emacs in a synchronous mode by invoking it with the -xrm |
267 | option, like this: | |
268 | ||
9031cdf2 | 269 | emacs -xrm "emacs.synchronous: true" |
ec4054f0 EZ |
270 | |
271 | Setting a breakpoint in the function `x_error_quitter' and looking at | |
272 | the backtrace when Emacs stops inside that function will show what | |
273 | code causes the X protocol errors. | |
274 | ||
1d3adc26 EZ |
275 | Some bugs related to the X protocol disappear when Emacs runs in a |
276 | synchronous mode. To track down those bugs, we suggest the following | |
277 | procedure: | |
278 | ||
279 | - Run Emacs under a debugger and put a breakpoint inside the | |
280 | primitive function which, when called from Lisp, triggers the X | |
281 | protocol errors. For example, if the errors happen when you | |
282 | delete a frame, put a breakpoint inside `Fdelete_frame'. | |
283 | ||
284 | - When the breakpoint breaks, step through the code, looking for | |
285 | calls to X functions (the ones whose names begin with "X" or | |
286 | "Xt" or "Xm"). | |
287 | ||
288 | - Insert calls to `XSync' before and after each call to the X | |
289 | functions, like this: | |
290 | ||
291 | XSync (f->output_data.x->display_info->display, 0); | |
292 | ||
293 | where `f' is the pointer to the `struct frame' of the selected | |
294 | frame, normally available via XFRAME (selected_frame). (Most | |
295 | functions which call X already have some variable that holds the | |
296 | pointer to the frame, perhaps called `f' or `sf', so you shouldn't | |
297 | need to compute it.) | |
298 | ||
299 | If your debugger can call functions in the program being debugged, | |
300 | you should be able to issue the calls to `XSync' without recompiling | |
301 | Emacs. For example, with GDB, just type: | |
302 | ||
303 | call XSync (f->output_data.x->display_info->display, 0) | |
304 | ||
305 | before and immediately after the suspect X calls. If your | |
306 | debugger does not support this, you will need to add these pairs | |
307 | of calls in the source and rebuild Emacs. | |
308 | ||
309 | Either way, systematically step through the code and issue these | |
310 | calls until you find the first X function called by Emacs after | |
311 | which a call to `XSync' winds up in the function | |
312 | `x_error_quitter'. The first X function call for which this | |
313 | happens is the one that generated the X protocol error. | |
314 | ||
315 | - You should now look around this offending X call and try to figure | |
316 | out what is wrong with it. | |
317 | ||
46def989 JD |
318 | ** If Emacs causes errors or memory leaks in your X server |
319 | ||
320 | You can trace the traffic between Emacs and your X server with a tool | |
321 | like xmon, available at ftp://ftp.x.org/contrib/devel_tools/. | |
322 | ||
323 | Xmon can be used to see exactly what Emacs sends when X protocol errors | |
324 | happen. If Emacs causes the X server memory usage to increase you can | |
325 | use xmon to see what items Emacs creates in the server (windows, | |
326 | graphical contexts, pixmaps) and what items Emacs delete. If there | |
327 | are consistently more creations than deletions, the type of item | |
328 | and the activity you do when the items get created can give a hint where | |
329 | to start debugging. | |
330 | ||
437368fe EZ |
331 | ** If the symptom of the bug is that Emacs fails to respond |
332 | ||
333 | Don't assume Emacs is `hung'--it may instead be in an infinite loop. | |
334 | To find out which, make the problem happen under GDB and stop Emacs | |
335 | once it is not responding. (If Emacs is using X Windows directly, you | |
336 | can stop Emacs by typing C-z at the GDB job.) Then try stepping with | |
337 | `step'. If Emacs is hung, the `step' command won't return. If it is | |
338 | looping, `step' will return. | |
339 | ||
340 | If this shows Emacs is hung in a system call, stop it again and | |
341 | examine the arguments of the call. If you report the bug, it is very | |
342 | important to state exactly where in the source the system call is, and | |
343 | what the arguments are. | |
344 | ||
345 | If Emacs is in an infinite loop, try to determine where the loop | |
346 | starts and ends. The easiest way to do this is to use the GDB command | |
347 | `finish'. Each time you use it, Emacs resumes execution until it | |
348 | exits one stack frame. Keep typing `finish' until it doesn't | |
349 | return--that means the infinite loop is in the stack frame which you | |
350 | just tried to finish. | |
351 | ||
352 | Stop Emacs again, and use `finish' repeatedly again until you get back | |
353 | to that frame. Then use `next' to step through that frame. By | |
354 | stepping, you will see where the loop starts and ends. Also, examine | |
355 | the data being used in the loop and try to determine why the loop does | |
356 | not exit when it should. | |
357 | ||
358 | ** If certain operations in Emacs are slower than they used to be, here | |
359 | is some advice for how to find out why. | |
360 | ||
361 | Stop Emacs repeatedly during the slow operation, and make a backtrace | |
362 | each time. Compare the backtraces looking for a pattern--a specific | |
363 | function that shows up more often than you'd expect. | |
364 | ||
365 | If you don't see a pattern in the C backtraces, get some Lisp | |
366 | backtrace information by typing "xbacktrace" or by looking at Ffuncall | |
367 | frames (see above), and again look for a pattern. | |
368 | ||
369 | When using X, you can stop Emacs at any time by typing C-z at GDB. | |
370 | When not using X, you can do this with C-g. On non-Unix platforms, | |
371 | such as MS-DOS, you might need to press C-BREAK instead. | |
372 | ||
a933dad1 DL |
373 | ** If GDB does not run and your debuggers can't load Emacs. |
374 | ||
375 | On some systems, no debugger can load Emacs with a symbol table, | |
376 | perhaps because they all have fixed limits on the number of symbols | |
377 | and Emacs exceeds the limits. Here is a method that can be used | |
378 | in such an extremity. Do | |
379 | ||
380 | nm -n temacs > nmout | |
381 | strip temacs | |
382 | adb temacs | |
383 | 0xd:i | |
384 | 0xe:i | |
385 | 14:i | |
386 | 17:i | |
387 | :r -l loadup (or whatever) | |
388 | ||
389 | It is necessary to refer to the file `nmout' to convert | |
390 | numeric addresses into symbols and vice versa. | |
391 | ||
392 | It is useful to be running under a window system. | |
393 | Then, if Emacs becomes hopelessly wedged, you can create | |
394 | another window to do kill -9 in. kill -ILL is often | |
395 | useful too, since that may make Emacs dump core or return | |
396 | to adb. | |
397 | ||
398 | ||
399 | ** Debugging incorrect screen updating. | |
400 | ||
401 | To debug Emacs problems that update the screen wrong, it is useful | |
402 | to have a record of what input you typed and what Emacs sent to the | |
403 | screen. To make these records, do | |
404 | ||
405 | (open-dribble-file "~/.dribble") | |
406 | (open-termscript "~/.termscript") | |
407 | ||
408 | The dribble file contains all characters read by Emacs from the | |
409 | terminal, and the termscript file contains all characters it sent to | |
410 | the terminal. The use of the directory `~/' prevents interference | |
411 | with any other user. | |
412 | ||
413 | If you have irreproducible display problems, put those two expressions | |
414 | in your ~/.emacs file. When the problem happens, exit the Emacs that | |
415 | you were running, kill it, and rename the two files. Then you can start | |
416 | another Emacs without clobbering those files, and use it to examine them. | |
125f929e MB |
417 | |
418 | An easy way to see if too much text is being redrawn on a terminal is to | |
419 | evaluate `(setq inverse-video t)' before you try the operation you think | |
420 | will cause too much redrawing. This doesn't refresh the screen, so only | |
421 | newly drawn text is in inverse video. | |
437368fe | 422 | |
3f715e77 EZ |
423 | The Emacs display code includes special debugging code, but it is |
424 | normally disabled. You can enable it by building Emacs with the | |
425 | pre-processing symbol GLYPH_DEBUG defined. Here's one easy way, | |
426 | suitable for Unix and GNU systems, to build such a debugging version: | |
427 | ||
428 | MYCPPFLAGS='-DGLYPH_DEBUG=1' make | |
429 | ||
430 | Building Emacs like that activates many assertions which scrutinize | |
431 | display code operation more than Emacs does normally. (To see the | |
432 | code which tests these assertions, look for calls to the `xassert' | |
433 | macros.) Any assertion that is reported to fail should be | |
434 | investigated. | |
435 | ||
436 | Building with GLYPH_DEBUG defined also defines several helper | |
437 | functions which can help debugging display code. One such function is | |
438 | `dump_glyph_matrix'. If you run Emacs under GDB, you can print the | |
439 | contents of any glyph matrix by just calling that function with the | |
440 | matrix as its argument. For example, the following command will print | |
441 | the contents of the current matrix of the window whose pointer is in | |
442 | `w': | |
443 | ||
444 | (gdb) p dump_glyph_matrix (w->current_matrix, 2) | |
445 | ||
446 | (The second argument 2 tells dump_glyph_matrix to print the glyphs in | |
447 | a long form.) You can dump the selected window's current glyph matrix | |
448 | interactively with "M-x dump-glyph-matrix RET"; see the documentation | |
449 | of this function for more details. | |
450 | ||
451 | Several more functions for debugging display code are available in | |
71d6b459 EZ |
452 | Emacs compiled with GLYPH_DEBUG defined; type "C-h f dump- TAB" and |
453 | "C-h f trace- TAB" to see the full list. | |
3f715e77 | 454 | |
1ce9f40a KS |
455 | When you debug display problems running emacs under X, you can use |
456 | the `ff' command to flush all pending display updates to the screen. | |
457 | ||
437368fe EZ |
458 | |
459 | ** Debugging LessTif | |
460 | ||
461 | If you encounter bugs whereby Emacs built with LessTif grabs all mouse | |
462 | and keyboard events, or LessTif menus behave weirdly, it might be | |
463 | helpful to set the `DEBUGSOURCES' and `DEBUG_FILE' environment | |
464 | variables, so that one can see what LessTif was doing at this point. | |
465 | For instance | |
177c0ea7 | 466 | |
6806e867 | 467 | export DEBUGSOURCES="RowColumn.c:MenuShell.c:MenuUtil.c" |
437368fe | 468 | export DEBUG_FILE=/usr/tmp/LESSTIF_TRACE |
2aa25884 | 469 | emacs & |
437368fe EZ |
470 | |
471 | causes LessTif to print traces from the three named source files to a | |
2aa25884 EZ |
472 | file in `/usr/tmp' (that file can get pretty large). The above should |
473 | be typed at the shell prompt before invoking Emacs, as shown by the | |
474 | last line above. | |
437368fe EZ |
475 | |
476 | Running GDB from another terminal could also help with such problems. | |
477 | You can arrange for GDB to run on one machine, with the Emacs display | |
478 | appearing on another. Then, when the bug happens, you can go back to | |
479 | the machine where you started GDB and use the debugger from there. | |
480 | ||
481 | ||
437368fe EZ |
482 | ** Debugging problems which happen in GC |
483 | ||
f46cc673 EZ |
484 | The array `last_marked' (defined on alloc.c) can be used to display up |
485 | to 500 last objects marked by the garbage collection process. | |
62578de5 EZ |
486 | Whenever the garbage collector marks a Lisp object, it records the |
487 | pointer to that object in the `last_marked' array. The variable | |
f46cc673 EZ |
488 | `last_marked_index' holds the index into the `last_marked' array one |
489 | place beyond where the pointer to the very last marked object is | |
490 | stored. | |
437368fe EZ |
491 | |
492 | The single most important goal in debugging GC problems is to find the | |
493 | Lisp data structure that got corrupted. This is not easy since GC | |
494 | changes the tag bits and relocates strings which make it hard to look | |
495 | at Lisp objects with commands such as `pr'. It is sometimes necessary | |
496 | to convert Lisp_Object variables into pointers to C struct's manually. | |
497 | Use the `last_marked' array and the source to reconstruct the sequence | |
498 | that objects were marked. | |
499 | ||
500 | Once you discover the corrupted Lisp object or data structure, it is | |
8e92a96a EZ |
501 | useful to look at it in a fresh Emacs session and compare its contents |
502 | with a session that you are debugging. | |
437368fe | 503 | |
93699019 EZ |
504 | ** Debugging problems with non-ASCII characters |
505 | ||
506 | If you experience problems which seem to be related to non-ASCII | |
507 | characters, such as \201 characters appearing in the buffer or in your | |
508 | files, set the variable byte-debug-flag to t. This causes Emacs to do | |
509 | some extra checks, such as look for broken relations between byte and | |
510 | character positions in buffers and strings; the resulting diagnostics | |
511 | might pinpoint the cause of the problem. | |
512 | ||
036cb5a2 EZ |
513 | ** Debugging the TTY (non-windowed) version |
514 | ||
515 | The most convenient method of debugging the character-terminal display | |
516 | is to do that on a window system such as X. Begin by starting an | |
517 | xterm window, then type these commands inside that window: | |
518 | ||
519 | $ tty | |
520 | $ echo $TERM | |
521 | ||
522 | Let's say these commands print "/dev/ttyp4" and "xterm", respectively. | |
523 | ||
524 | Now start Emacs (the normal, windowed-display session, i.e. without | |
525 | the `-nw' option), and invoke "M-x gdb RET emacs RET" from there. Now | |
526 | type these commands at GDB's prompt: | |
527 | ||
528 | (gdb) set args -nw -t /dev/ttyp4 | |
529 | (gdb) set environment TERM xterm | |
530 | (gdb) run | |
531 | ||
532 | The debugged Emacs should now start in no-window mode with its display | |
533 | directed to the xterm window you opened above. | |
534 | ||
e039053d EZ |
535 | Similar arrangement is possible on a character terminal by using the |
536 | `screen' package. | |
537 | ||
19cf8f36 EZ |
538 | ** Running Emacs built with malloc debugging packages |
539 | ||
540 | If Emacs exhibits bugs that seem to be related to use of memory | |
541 | allocated off the heap, it might be useful to link Emacs with a | |
542 | special debugging library, such as Electric Fence (a.k.a. efence) or | |
543 | GNU Checker, which helps find such problems. | |
544 | ||
545 | Emacs compiled with such packages might not run without some hacking, | |
546 | because Emacs replaces the system's memory allocation functions with | |
547 | its own versions, and because the dumping process might be | |
548 | incompatible with the way these packages use to track allocated | |
549 | memory. Here are some of the changes you might find necessary | |
550 | (SYSTEM-NAME and MACHINE-NAME are the names of your OS- and | |
551 | CPU-specific headers in the subdirectories of `src'): | |
552 | ||
553 | - In src/s/SYSTEM-NAME.h add "#define SYSTEM_MALLOC". | |
554 | ||
555 | - In src/m/MACHINE-NAME.h add "#define CANNOT_DUMP" and | |
556 | "#define CANNOT_UNEXEC". | |
557 | ||
558 | - Configure with a different --prefix= option. If you use GCC, | |
559 | version 2.7.2 is preferred, as some malloc debugging packages | |
560 | work a lot better with it than with 2.95 or later versions. | |
561 | ||
562 | - Type "make" then "make -k install". | |
563 | ||
564 | - If required, invoke the package-specific command to prepare | |
565 | src/temacs for execution. | |
566 | ||
567 | - cd ..; src/temacs | |
568 | ||
569 | (Note that this runs `temacs' instead of the usual `emacs' executable. | |
570 | This avoids problems with dumping Emacs mentioned above.) | |
571 | ||
572 | Some malloc debugging libraries might print lots of false alarms for | |
573 | bitfields used by Emacs in some data structures. If you want to get | |
574 | rid of the false alarms, you will have to hack the definitions of | |
575 | these data structures on the respective headers to remove the `:N' | |
576 | bitfield definitions (which will cause each such field to use a full | |
577 | int). | |
578 | ||
437368fe EZ |
579 | ** Some suggestions for debugging on MS Windows: |
580 | ||
581 | (written by Marc Fleischeuers, Geoff Voelker and Andrew Innes) | |
582 | ||
3102e429 | 583 | To debug Emacs with Microsoft Visual C++, you either start emacs from |
961e2394 EZ |
584 | the debugger or attach the debugger to a running emacs process. |
585 | ||
586 | To start emacs from the debugger, you can use the file bin/debug.bat. | |
587 | The Microsoft Developer studio will start and under Project, Settings, | |
3102e429 | 588 | Debug, General you can set the command-line arguments and Emacs's |
437368fe EZ |
589 | startup directory. Set breakpoints (Edit, Breakpoints) at Fsignal and |
590 | other functions that you want to examine. Run the program (Build, | |
591 | Start debug). Emacs will start and the debugger will take control as | |
592 | soon as a breakpoint is hit. | |
593 | ||
3102e429 | 594 | You can also attach the debugger to an already running Emacs process. |
437368fe EZ |
595 | To do this, start up the Microsoft Developer studio and select Build, |
596 | Start debug, Attach to process. Choose the Emacs process from the | |
597 | list. Send a break to the running process (Debug, Break) and you will | |
598 | find that execution is halted somewhere in user32.dll. Open the stack | |
599 | trace window and go up the stack to w32_msg_pump. Now you can set | |
600 | breakpoints in Emacs (Edit, Breakpoints). Continue the running Emacs | |
601 | process (Debug, Step out) and control will return to Emacs, until a | |
602 | breakpoint is hit. | |
603 | ||
3102e429 | 604 | To examine the contents of a Lisp variable, you can use the function |
437368fe EZ |
605 | 'debug_print'. Right-click on a variable, select QuickWatch (it has |
606 | an eyeglass symbol on its button in the toolbar), and in the text | |
607 | field at the top of the window, place 'debug_print(' and ')' around | |
608 | the expression. Press 'Recalculate' and the output is sent to stderr, | |
609 | and to the debugger via the OutputDebugString routine. The output | |
610 | sent to stderr should be displayed in the console window that was | |
611 | opened when the emacs.exe executable was started. The output sent to | |
612 | the debugger should be displayed in the 'Debug' pane in the Output | |
613 | window. If Emacs was started from the debugger, a console window was | |
614 | opened at Emacs' startup; this console window also shows the output of | |
615 | 'debug_print'. | |
616 | ||
617 | For example, start and run Emacs in the debugger until it is waiting | |
618 | for user input. Then click on the `Break' button in the debugger to | |
619 | halt execution. Emacs should halt in `ZwUserGetMessage' waiting for | |
620 | an input event. Use the `Call Stack' window to select the procedure | |
621 | `w32_msp_pump' up the call stack (see below for why you have to do | |
622 | this). Open the QuickWatch window and enter | |
623 | "debug_print(Vexec_path)". Evaluating this expression will then print | |
3102e429 | 624 | out the contents of the Lisp variable `exec-path'. |
437368fe EZ |
625 | |
626 | If QuickWatch reports that the symbol is unknown, then check the call | |
627 | stack in the `Call Stack' window. If the selected frame in the call | |
628 | stack is not an Emacs procedure, then the debugger won't recognize | |
629 | Emacs symbols. Instead, select a frame that is inside an Emacs | |
630 | procedure and try using `debug_print' again. | |
631 | ||
632 | If QuickWatch invokes debug_print but nothing happens, then check the | |
633 | thread that is selected in the debugger. If the selected thread is | |
634 | not the last thread to run (the "current" thread), then it cannot be | |
635 | used to execute debug_print. Use the Debug menu to select the current | |
636 | thread and try using debug_print again. Note that the debugger halts | |
637 | execution (e.g., due to a breakpoint) in the context of the current | |
638 | thread, so this should only be a problem if you've explicitly switched | |
639 | threads. | |
640 | ||
3102e429 | 641 | It is also possible to keep appropriately masked and typecast Lisp |
437368fe EZ |
642 | symbols in the Watch window, this is more convenient when steeping |
643 | though the code. For instance, on entering apply_lambda, you can | |
644 | watch (struct Lisp_Symbol *) (0xfffffff & args[0]). | |
961e2394 EZ |
645 | |
646 | Optimizations often confuse the MS debugger. For example, the | |
647 | debugger will sometimes report wrong line numbers, e.g., when it | |
648 | prints the backtrace for a crash. It is usually best to look at the | |
649 | disassembly to determine exactly what code is being run--the | |
650 | disassembly will probably show several source lines followed by a | |
651 | block of assembler for those lines. The actual point where Emacs | |
652 | crashes will be one of those source lines, but not neccesarily the one | |
653 | that the debugger reports. | |
654 | ||
655 | Another problematic area with the MS debugger is with variables that | |
656 | are stored in registers: it will sometimes display wrong values for | |
657 | those variables. Usually you will not be able to see any value for a | |
658 | register variable, but if it is only being stored in a register | |
659 | temporarily, you will see an old value for it. Again, you need to | |
660 | look at the disassembly to determine which registers are being used, | |
661 | and look at those registers directly, to see the actual current values | |
662 | of these variables. | |
ab5796a9 MB |
663 | |
664 | ;;; arch-tag: fbf32980-e35d-481f-8e4c-a2eca2586e6b |