Commit | Line | Data |
---|---|---|
a933dad1 | 1 | Debugging GNU Emacs |
437368fe | 2 | Copyright (c) 1985, 2000, 2001 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 RS |
19 | ** When you debug Emacs with GDB, you should start it in the directory |
20 | where you built Emacs. That directory has a .gdbinit file that defines | |
21 | various "user-defined" commands for debugging Emacs. | |
22 | ||
23 | ** It is a good idea to run Emacs under GDB (or some other suitable | |
437368fe EZ |
24 | debugger) *all the time*. Then, when Emacs crashes, you will be able |
25 | to debug the live process, not just a core dump. (This is especially | |
26 | important on systems which don't support core files, and instead print | |
27 | just the registers and some stack addresses.) | |
28 | ||
42a3c627 | 29 | ** If Emacs hangs, or seems to be stuck in some infinite loop, typing |
437368fe EZ |
30 | "kill -TSTP PID", where PID is the Emacs process ID, will cause GDB to |
31 | kick in, provided that you run under GDB. | |
32 | ||
33 | ** Getting control to the debugger | |
a933dad1 | 34 | |
3102e429 | 35 | `Fsignal' is a very useful place to put a breakpoint in. |
a933dad1 DL |
36 | All Lisp errors go through there. |
37 | ||
3102e429 | 38 | It is useful, when debugging, to have a guaranteed way to return to |
eb55f651 | 39 | the debugger at any time. When using X, this is easy: type C-z at the |
3102e429 EZ |
40 | window where Emacs is running under GDB, and it will stop Emacs just |
41 | as it would stop any ordinary program. When Emacs is running in a | |
42 | terminal, things are not so easy. | |
43 | ||
44 | The src/.gdbinit file in the Emacs distribution arranges for SIGINT | |
45 | (C-g in Emacs) to be passed to Emacs and not give control back to GDB. | |
46 | On modern POSIX systems, you can override that with this command: | |
47 | ||
7718638c | 48 | handle SIGINT stop nopass |
3102e429 EZ |
49 | |
50 | After this `handle' command, SIGINT will return control to GDB. If | |
51 | you want the C-g to cause a QUIT within Emacs as well, omit the | |
52 | `nopass'. | |
53 | ||
54 | A technique that can work when `handle SIGINT' does not is to store | |
55 | the code for some character into the variable stop_character. Thus, | |
a933dad1 DL |
56 | |
57 | set stop_character = 29 | |
58 | ||
59 | makes Control-] (decimal code 29) the stop character. | |
60 | Typing Control-] will cause immediate stop. You cannot | |
61 | use the set command until the inferior process has been started. | |
62 | Put a breakpoint early in `main', or suspend the Emacs, | |
63 | to get an opportunity to do the set command. | |
64 | ||
a933dad1 DL |
65 | ** Examining Lisp object values. |
66 | ||
67 | When you have a live process to debug, and it has not encountered a | |
68 | fatal error, you can use the GDB command `pr'. First print the value | |
69 | in the ordinary way, with the `p' command. Then type `pr' with no | |
70 | arguments. This calls a subroutine which uses the Lisp printer. | |
71 | ||
437368fe EZ |
72 | Note: It is not a good idea to try `pr' if you know that Emacs is in |
73 | deep trouble: its stack smashed (e.g., if it encountered SIGSEGV due | |
74 | to stack overflow), or crucial data structures, such as `obarray', | |
75 | corrupted, etc. In such cases, the Emacs subroutine called by `pr' | |
76 | might make more damage, like overwrite some data that is important for | |
77 | debugging the original problem. | |
78 | ||
3102e429 EZ |
79 | Also, on some systems it is impossible to use `pr' if you stopped |
80 | Emacs while it was inside `select'. This is in fact what happens if | |
81 | you stop Emacs while it is waiting. In such a situation, don't try to | |
82 | use `pr'. Instead, use `s' to step out of the system call. Then | |
83 | Emacs will be between instructions and capable of handling `pr'. | |
a933dad1 | 84 | |
3102e429 EZ |
85 | If you can't use `pr' command, for whatever reason, you can fall back |
86 | on lower-level commands. Use the `xtype' command to print out the | |
87 | data type of the last data value. Once you know the data type, use | |
88 | the command that corresponds to that type. Here are these commands: | |
a933dad1 DL |
89 | |
90 | xint xptr xwindow xmarker xoverlay xmiscfree xintfwd xboolfwd xobjfwd | |
91 | xbufobjfwd xkbobjfwd xbuflocal xbuffer xsymbol xstring xvector xframe | |
92 | xwinconfig xcompiled xcons xcar xcdr xsubr xprocess xfloat xscrollbar | |
93 | ||
94 | Each one of them applies to a certain type or class of types. | |
95 | (Some of these types are not visible in Lisp, because they exist only | |
96 | internally.) | |
97 | ||
98 | Each x... command prints some information about the value, and | |
99 | produces a GDB value (subsequently available in $) through which you | |
100 | can get at the rest of the contents. | |
101 | ||
437368fe | 102 | In general, most of the rest of the contents will be additional Lisp |
a933dad1 DL |
103 | objects which you can examine in turn with the x... commands. |
104 | ||
437368fe EZ |
105 | Even with a live process, these x... commands are useful for |
106 | examining the fields in a buffer, window, process, frame or marker. | |
107 | Here's an example using concepts explained in the node "Value History" | |
108 | of the GDB manual to print the variable frame from this line in | |
109 | xmenu.c: | |
110 | ||
111 | buf.frame_or_window = frame; | |
112 | ||
113 | First, use these commands: | |
114 | ||
115 | cd src | |
116 | gdb emacs | |
117 | b xmenu.c:1296 | |
177c0ea7 | 118 | r -q |
437368fe EZ |
119 | |
120 | Then type C-x 5 2 to create a new frame, and it hits the breakpoint: | |
121 | ||
122 | (gdb) p frame | |
123 | $1 = 1077872640 | |
124 | (gdb) xtype | |
125 | Lisp_Vectorlike | |
126 | PVEC_FRAME | |
127 | (gdb) xframe | |
128 | $2 = (struct frame *) 0x3f0800 | |
129 | (gdb) p *$ | |
130 | $3 = { | |
177c0ea7 JB |
131 | size = 536871989, |
132 | next = 0x366240, | |
133 | name = 809661752, | |
437368fe EZ |
134 | [...] |
135 | } | |
136 | (gdb) p $3->name | |
137 | $4 = 809661752 | |
138 | ||
139 | Now we can use `pr' to print the name of the frame: | |
140 | ||
141 | (gdb) pr | |
142 | "emacs@steenrod.math.nwu.edu" | |
143 | ||
144 | The Emacs C code heavily uses macros defined in lisp.h. So suppose | |
145 | we want the address of the l-value expression near the bottom of | |
146 | `add_command_key' from keyboard.c: | |
147 | ||
148 | XVECTOR (this_command_keys)->contents[this_command_key_count++] = key; | |
149 | ||
150 | XVECTOR is a macro, and therefore GDB does not know about it. | |
151 | GDB cannot evaluate "p XVECTOR (this_command_keys)". | |
152 | ||
153 | However, you can use the xvector command in GDB to get the same | |
154 | result. Here is how: | |
155 | ||
156 | (gdb) p this_command_keys | |
157 | $1 = 1078005760 | |
158 | (gdb) xvector | |
159 | $2 = (struct Lisp_Vector *) 0x411000 | |
160 | 0 | |
161 | (gdb) p $->contents[this_command_key_count] | |
162 | $3 = 1077872640 | |
163 | (gdb) p &$ | |
164 | $4 = (int *) 0x411008 | |
165 | ||
166 | Here's a related example of macros and the GDB `define' command. | |
167 | There are many Lisp vectors such as `recent_keys', which contains the | |
168 | last 100 keystrokes. We can print this Lisp vector | |
169 | ||
170 | p recent_keys | |
171 | pr | |
172 | ||
173 | But this may be inconvenient, since `recent_keys' is much more verbose | |
174 | than `C-h l'. We might want to print only the last 10 elements of | |
175 | this vector. `recent_keys' is updated in keyboard.c by the command | |
176 | ||
177 | XVECTOR (recent_keys)->contents[recent_keys_index] = c; | |
178 | ||
179 | So we define a GDB command `xvector-elts', so the last 10 keystrokes | |
177c0ea7 | 180 | are printed by |
437368fe EZ |
181 | |
182 | xvector-elts recent_keys recent_keys_index 10 | |
183 | ||
184 | where you can define xvector-elts as follows: | |
185 | ||
186 | define xvector-elts | |
187 | set $i = 0 | |
188 | p $arg0 | |
189 | xvector | |
190 | set $foo = $ | |
191 | while $i < $arg2 | |
177c0ea7 | 192 | p $foo->contents[$arg1-($i++)] |
437368fe EZ |
193 | pr |
194 | end | |
195 | document xvector-elts | |
196 | Prints a range of elements of a Lisp vector. | |
197 | xvector-elts v n i | |
198 | prints `i' elements of the vector `v' ending at the index `n'. | |
199 | end | |
200 | ||
201 | ** Getting Lisp-level backtrace information within GDB | |
202 | ||
3102e429 EZ |
203 | The most convenient way is to use the `xbacktrace' command. This |
204 | shows the names of the Lisp functions that are currently active. | |
437368fe EZ |
205 | |
206 | If that doesn't work (e.g., because the `backtrace_list' structure is | |
207 | corrupted), type "bt" at the GDB prompt, to produce the C-level | |
208 | backtrace, and look for stack frames that call Ffuncall. Select them | |
209 | one by one in GDB, by typing "up N", where N is the appropriate number | |
210 | of frames to go up, and in each frame that calls Ffuncall type this: | |
211 | ||
212 | p *args | |
213 | pr | |
214 | ||
215 | This will print the name of the Lisp function called by that level | |
216 | of function calling. | |
217 | ||
218 | By printing the remaining elements of args, you can see the argument | |
219 | values. Here's how to print the first argument: | |
177c0ea7 | 220 | |
437368fe EZ |
221 | p args[1] |
222 | pr | |
223 | ||
224 | If you do not have a live process, you can use xtype and the other | |
225 | x... commands such as xsymbol to get such information, albeit less | |
226 | conveniently. For example: | |
227 | ||
228 | p *args | |
229 | xtype | |
230 | ||
231 | and, assuming that "xtype" says that args[0] is a symbol: | |
232 | ||
177c0ea7 | 233 | xsymbol |
437368fe EZ |
234 | |
235 | ** Debugging what happens while preloading and dumping Emacs | |
236 | ||
237 | Type `gdb temacs' and start it with `r -batch -l loadup dump'. | |
238 | ||
239 | If temacs actually succeeds when running under GDB in this way, do not | |
240 | try to run the dumped Emacs, because it was dumped with the GDB | |
241 | breakpoints in it. | |
242 | ||
243 | ** Debugging `temacs' | |
244 | ||
245 | Debugging `temacs' is useful when you want to establish whether a | |
246 | problem happens in an undumped Emacs. To run `temacs' under a | |
247 | debugger, type "gdb temacs", then start it with `r -batch -l loadup'. | |
248 | ||
249 | ** If you encounter X protocol errors | |
250 | ||
251 | Try evaluating (x-synchronize t). That puts Emacs into synchronous | |
252 | mode, where each Xlib call checks for errors before it returns. This | |
253 | mode is much slower, but when you get an error, you will see exactly | |
254 | which call really caused the error. | |
255 | ||
ec4054f0 EZ |
256 | You can start Emacs in a synchronous mode by invoking it with the -xrm |
257 | option, like this: | |
258 | ||
9031cdf2 | 259 | emacs -xrm "emacs.synchronous: true" |
ec4054f0 EZ |
260 | |
261 | Setting a breakpoint in the function `x_error_quitter' and looking at | |
262 | the backtrace when Emacs stops inside that function will show what | |
263 | code causes the X protocol errors. | |
264 | ||
1d3adc26 EZ |
265 | Some bugs related to the X protocol disappear when Emacs runs in a |
266 | synchronous mode. To track down those bugs, we suggest the following | |
267 | procedure: | |
268 | ||
269 | - Run Emacs under a debugger and put a breakpoint inside the | |
270 | primitive function which, when called from Lisp, triggers the X | |
271 | protocol errors. For example, if the errors happen when you | |
272 | delete a frame, put a breakpoint inside `Fdelete_frame'. | |
273 | ||
274 | - When the breakpoint breaks, step through the code, looking for | |
275 | calls to X functions (the ones whose names begin with "X" or | |
276 | "Xt" or "Xm"). | |
277 | ||
278 | - Insert calls to `XSync' before and after each call to the X | |
279 | functions, like this: | |
280 | ||
281 | XSync (f->output_data.x->display_info->display, 0); | |
282 | ||
283 | where `f' is the pointer to the `struct frame' of the selected | |
284 | frame, normally available via XFRAME (selected_frame). (Most | |
285 | functions which call X already have some variable that holds the | |
286 | pointer to the frame, perhaps called `f' or `sf', so you shouldn't | |
287 | need to compute it.) | |
288 | ||
289 | If your debugger can call functions in the program being debugged, | |
290 | you should be able to issue the calls to `XSync' without recompiling | |
291 | Emacs. For example, with GDB, just type: | |
292 | ||
293 | call XSync (f->output_data.x->display_info->display, 0) | |
294 | ||
295 | before and immediately after the suspect X calls. If your | |
296 | debugger does not support this, you will need to add these pairs | |
297 | of calls in the source and rebuild Emacs. | |
298 | ||
299 | Either way, systematically step through the code and issue these | |
300 | calls until you find the first X function called by Emacs after | |
301 | which a call to `XSync' winds up in the function | |
302 | `x_error_quitter'. The first X function call for which this | |
303 | happens is the one that generated the X protocol error. | |
304 | ||
305 | - You should now look around this offending X call and try to figure | |
306 | out what is wrong with it. | |
307 | ||
46def989 JD |
308 | ** If Emacs causes errors or memory leaks in your X server |
309 | ||
310 | You can trace the traffic between Emacs and your X server with a tool | |
311 | like xmon, available at ftp://ftp.x.org/contrib/devel_tools/. | |
312 | ||
313 | Xmon can be used to see exactly what Emacs sends when X protocol errors | |
314 | happen. If Emacs causes the X server memory usage to increase you can | |
315 | use xmon to see what items Emacs creates in the server (windows, | |
316 | graphical contexts, pixmaps) and what items Emacs delete. If there | |
317 | are consistently more creations than deletions, the type of item | |
318 | and the activity you do when the items get created can give a hint where | |
319 | to start debugging. | |
320 | ||
437368fe EZ |
321 | ** If the symptom of the bug is that Emacs fails to respond |
322 | ||
323 | Don't assume Emacs is `hung'--it may instead be in an infinite loop. | |
324 | To find out which, make the problem happen under GDB and stop Emacs | |
325 | once it is not responding. (If Emacs is using X Windows directly, you | |
326 | can stop Emacs by typing C-z at the GDB job.) Then try stepping with | |
327 | `step'. If Emacs is hung, the `step' command won't return. If it is | |
328 | looping, `step' will return. | |
329 | ||
330 | If this shows Emacs is hung in a system call, stop it again and | |
331 | examine the arguments of the call. If you report the bug, it is very | |
332 | important to state exactly where in the source the system call is, and | |
333 | what the arguments are. | |
334 | ||
335 | If Emacs is in an infinite loop, try to determine where the loop | |
336 | starts and ends. The easiest way to do this is to use the GDB command | |
337 | `finish'. Each time you use it, Emacs resumes execution until it | |
338 | exits one stack frame. Keep typing `finish' until it doesn't | |
339 | return--that means the infinite loop is in the stack frame which you | |
340 | just tried to finish. | |
341 | ||
342 | Stop Emacs again, and use `finish' repeatedly again until you get back | |
343 | to that frame. Then use `next' to step through that frame. By | |
344 | stepping, you will see where the loop starts and ends. Also, examine | |
345 | the data being used in the loop and try to determine why the loop does | |
346 | not exit when it should. | |
347 | ||
348 | ** If certain operations in Emacs are slower than they used to be, here | |
349 | is some advice for how to find out why. | |
350 | ||
351 | Stop Emacs repeatedly during the slow operation, and make a backtrace | |
352 | each time. Compare the backtraces looking for a pattern--a specific | |
353 | function that shows up more often than you'd expect. | |
354 | ||
355 | If you don't see a pattern in the C backtraces, get some Lisp | |
356 | backtrace information by typing "xbacktrace" or by looking at Ffuncall | |
357 | frames (see above), and again look for a pattern. | |
358 | ||
359 | When using X, you can stop Emacs at any time by typing C-z at GDB. | |
360 | When not using X, you can do this with C-g. On non-Unix platforms, | |
361 | such as MS-DOS, you might need to press C-BREAK instead. | |
362 | ||
a933dad1 DL |
363 | ** If GDB does not run and your debuggers can't load Emacs. |
364 | ||
365 | On some systems, no debugger can load Emacs with a symbol table, | |
366 | perhaps because they all have fixed limits on the number of symbols | |
367 | and Emacs exceeds the limits. Here is a method that can be used | |
368 | in such an extremity. Do | |
369 | ||
370 | nm -n temacs > nmout | |
371 | strip temacs | |
372 | adb temacs | |
373 | 0xd:i | |
374 | 0xe:i | |
375 | 14:i | |
376 | 17:i | |
377 | :r -l loadup (or whatever) | |
378 | ||
379 | It is necessary to refer to the file `nmout' to convert | |
380 | numeric addresses into symbols and vice versa. | |
381 | ||
382 | It is useful to be running under a window system. | |
383 | Then, if Emacs becomes hopelessly wedged, you can create | |
384 | another window to do kill -9 in. kill -ILL is often | |
385 | useful too, since that may make Emacs dump core or return | |
386 | to adb. | |
387 | ||
388 | ||
389 | ** Debugging incorrect screen updating. | |
390 | ||
391 | To debug Emacs problems that update the screen wrong, it is useful | |
392 | to have a record of what input you typed and what Emacs sent to the | |
393 | screen. To make these records, do | |
394 | ||
395 | (open-dribble-file "~/.dribble") | |
396 | (open-termscript "~/.termscript") | |
397 | ||
398 | The dribble file contains all characters read by Emacs from the | |
399 | terminal, and the termscript file contains all characters it sent to | |
400 | the terminal. The use of the directory `~/' prevents interference | |
401 | with any other user. | |
402 | ||
403 | If you have irreproducible display problems, put those two expressions | |
404 | in your ~/.emacs file. When the problem happens, exit the Emacs that | |
405 | you were running, kill it, and rename the two files. Then you can start | |
406 | another Emacs without clobbering those files, and use it to examine them. | |
125f929e MB |
407 | |
408 | An easy way to see if too much text is being redrawn on a terminal is to | |
409 | evaluate `(setq inverse-video t)' before you try the operation you think | |
410 | will cause too much redrawing. This doesn't refresh the screen, so only | |
411 | newly drawn text is in inverse video. | |
437368fe | 412 | |
3f715e77 EZ |
413 | The Emacs display code includes special debugging code, but it is |
414 | normally disabled. You can enable it by building Emacs with the | |
415 | pre-processing symbol GLYPH_DEBUG defined. Here's one easy way, | |
416 | suitable for Unix and GNU systems, to build such a debugging version: | |
417 | ||
418 | MYCPPFLAGS='-DGLYPH_DEBUG=1' make | |
419 | ||
420 | Building Emacs like that activates many assertions which scrutinize | |
421 | display code operation more than Emacs does normally. (To see the | |
422 | code which tests these assertions, look for calls to the `xassert' | |
423 | macros.) Any assertion that is reported to fail should be | |
424 | investigated. | |
425 | ||
426 | Building with GLYPH_DEBUG defined also defines several helper | |
427 | functions which can help debugging display code. One such function is | |
428 | `dump_glyph_matrix'. If you run Emacs under GDB, you can print the | |
429 | contents of any glyph matrix by just calling that function with the | |
430 | matrix as its argument. For example, the following command will print | |
431 | the contents of the current matrix of the window whose pointer is in | |
432 | `w': | |
433 | ||
434 | (gdb) p dump_glyph_matrix (w->current_matrix, 2) | |
435 | ||
436 | (The second argument 2 tells dump_glyph_matrix to print the glyphs in | |
437 | a long form.) You can dump the selected window's current glyph matrix | |
438 | interactively with "M-x dump-glyph-matrix RET"; see the documentation | |
439 | of this function for more details. | |
440 | ||
441 | Several more functions for debugging display code are available in | |
71d6b459 EZ |
442 | Emacs compiled with GLYPH_DEBUG defined; type "C-h f dump- TAB" and |
443 | "C-h f trace- TAB" to see the full list. | |
3f715e77 | 444 | |
437368fe EZ |
445 | |
446 | ** Debugging LessTif | |
447 | ||
448 | If you encounter bugs whereby Emacs built with LessTif grabs all mouse | |
449 | and keyboard events, or LessTif menus behave weirdly, it might be | |
450 | helpful to set the `DEBUGSOURCES' and `DEBUG_FILE' environment | |
451 | variables, so that one can see what LessTif was doing at this point. | |
452 | For instance | |
177c0ea7 | 453 | |
6806e867 | 454 | export DEBUGSOURCES="RowColumn.c:MenuShell.c:MenuUtil.c" |
437368fe | 455 | export DEBUG_FILE=/usr/tmp/LESSTIF_TRACE |
2aa25884 | 456 | emacs & |
437368fe EZ |
457 | |
458 | causes LessTif to print traces from the three named source files to a | |
2aa25884 EZ |
459 | file in `/usr/tmp' (that file can get pretty large). The above should |
460 | be typed at the shell prompt before invoking Emacs, as shown by the | |
461 | last line above. | |
437368fe EZ |
462 | |
463 | Running GDB from another terminal could also help with such problems. | |
464 | You can arrange for GDB to run on one machine, with the Emacs display | |
465 | appearing on another. Then, when the bug happens, you can go back to | |
466 | the machine where you started GDB and use the debugger from there. | |
467 | ||
468 | ||
437368fe EZ |
469 | ** Debugging problems which happen in GC |
470 | ||
f46cc673 EZ |
471 | The array `last_marked' (defined on alloc.c) can be used to display up |
472 | to 500 last objects marked by the garbage collection process. | |
62578de5 EZ |
473 | Whenever the garbage collector marks a Lisp object, it records the |
474 | pointer to that object in the `last_marked' array. The variable | |
f46cc673 EZ |
475 | `last_marked_index' holds the index into the `last_marked' array one |
476 | place beyond where the pointer to the very last marked object is | |
477 | stored. | |
437368fe EZ |
478 | |
479 | The single most important goal in debugging GC problems is to find the | |
480 | Lisp data structure that got corrupted. This is not easy since GC | |
481 | changes the tag bits and relocates strings which make it hard to look | |
482 | at Lisp objects with commands such as `pr'. It is sometimes necessary | |
483 | to convert Lisp_Object variables into pointers to C struct's manually. | |
484 | Use the `last_marked' array and the source to reconstruct the sequence | |
485 | that objects were marked. | |
486 | ||
487 | Once you discover the corrupted Lisp object or data structure, it is | |
8e92a96a EZ |
488 | useful to look at it in a fresh Emacs session and compare its contents |
489 | with a session that you are debugging. | |
437368fe | 490 | |
93699019 EZ |
491 | ** Debugging problems with non-ASCII characters |
492 | ||
493 | If you experience problems which seem to be related to non-ASCII | |
494 | characters, such as \201 characters appearing in the buffer or in your | |
495 | files, set the variable byte-debug-flag to t. This causes Emacs to do | |
496 | some extra checks, such as look for broken relations between byte and | |
497 | character positions in buffers and strings; the resulting diagnostics | |
498 | might pinpoint the cause of the problem. | |
499 | ||
036cb5a2 EZ |
500 | ** Debugging the TTY (non-windowed) version |
501 | ||
502 | The most convenient method of debugging the character-terminal display | |
503 | is to do that on a window system such as X. Begin by starting an | |
504 | xterm window, then type these commands inside that window: | |
505 | ||
506 | $ tty | |
507 | $ echo $TERM | |
508 | ||
509 | Let's say these commands print "/dev/ttyp4" and "xterm", respectively. | |
510 | ||
511 | Now start Emacs (the normal, windowed-display session, i.e. without | |
512 | the `-nw' option), and invoke "M-x gdb RET emacs RET" from there. Now | |
513 | type these commands at GDB's prompt: | |
514 | ||
515 | (gdb) set args -nw -t /dev/ttyp4 | |
516 | (gdb) set environment TERM xterm | |
517 | (gdb) run | |
518 | ||
519 | The debugged Emacs should now start in no-window mode with its display | |
520 | directed to the xterm window you opened above. | |
521 | ||
e039053d EZ |
522 | Similar arrangement is possible on a character terminal by using the |
523 | `screen' package. | |
524 | ||
19cf8f36 EZ |
525 | ** Running Emacs built with malloc debugging packages |
526 | ||
527 | If Emacs exhibits bugs that seem to be related to use of memory | |
528 | allocated off the heap, it might be useful to link Emacs with a | |
529 | special debugging library, such as Electric Fence (a.k.a. efence) or | |
530 | GNU Checker, which helps find such problems. | |
531 | ||
532 | Emacs compiled with such packages might not run without some hacking, | |
533 | because Emacs replaces the system's memory allocation functions with | |
534 | its own versions, and because the dumping process might be | |
535 | incompatible with the way these packages use to track allocated | |
536 | memory. Here are some of the changes you might find necessary | |
537 | (SYSTEM-NAME and MACHINE-NAME are the names of your OS- and | |
538 | CPU-specific headers in the subdirectories of `src'): | |
539 | ||
540 | - In src/s/SYSTEM-NAME.h add "#define SYSTEM_MALLOC". | |
541 | ||
542 | - In src/m/MACHINE-NAME.h add "#define CANNOT_DUMP" and | |
543 | "#define CANNOT_UNEXEC". | |
544 | ||
545 | - Configure with a different --prefix= option. If you use GCC, | |
546 | version 2.7.2 is preferred, as some malloc debugging packages | |
547 | work a lot better with it than with 2.95 or later versions. | |
548 | ||
549 | - Type "make" then "make -k install". | |
550 | ||
551 | - If required, invoke the package-specific command to prepare | |
552 | src/temacs for execution. | |
553 | ||
554 | - cd ..; src/temacs | |
555 | ||
556 | (Note that this runs `temacs' instead of the usual `emacs' executable. | |
557 | This avoids problems with dumping Emacs mentioned above.) | |
558 | ||
559 | Some malloc debugging libraries might print lots of false alarms for | |
560 | bitfields used by Emacs in some data structures. If you want to get | |
561 | rid of the false alarms, you will have to hack the definitions of | |
562 | these data structures on the respective headers to remove the `:N' | |
563 | bitfield definitions (which will cause each such field to use a full | |
564 | int). | |
565 | ||
437368fe EZ |
566 | ** Some suggestions for debugging on MS Windows: |
567 | ||
568 | (written by Marc Fleischeuers, Geoff Voelker and Andrew Innes) | |
569 | ||
3102e429 | 570 | To debug Emacs with Microsoft Visual C++, you either start emacs from |
961e2394 EZ |
571 | the debugger or attach the debugger to a running emacs process. |
572 | ||
573 | To start emacs from the debugger, you can use the file bin/debug.bat. | |
574 | The Microsoft Developer studio will start and under Project, Settings, | |
3102e429 | 575 | Debug, General you can set the command-line arguments and Emacs's |
437368fe EZ |
576 | startup directory. Set breakpoints (Edit, Breakpoints) at Fsignal and |
577 | other functions that you want to examine. Run the program (Build, | |
578 | Start debug). Emacs will start and the debugger will take control as | |
579 | soon as a breakpoint is hit. | |
580 | ||
3102e429 | 581 | You can also attach the debugger to an already running Emacs process. |
437368fe EZ |
582 | To do this, start up the Microsoft Developer studio and select Build, |
583 | Start debug, Attach to process. Choose the Emacs process from the | |
584 | list. Send a break to the running process (Debug, Break) and you will | |
585 | find that execution is halted somewhere in user32.dll. Open the stack | |
586 | trace window and go up the stack to w32_msg_pump. Now you can set | |
587 | breakpoints in Emacs (Edit, Breakpoints). Continue the running Emacs | |
588 | process (Debug, Step out) and control will return to Emacs, until a | |
589 | breakpoint is hit. | |
590 | ||
3102e429 | 591 | To examine the contents of a Lisp variable, you can use the function |
437368fe EZ |
592 | 'debug_print'. Right-click on a variable, select QuickWatch (it has |
593 | an eyeglass symbol on its button in the toolbar), and in the text | |
594 | field at the top of the window, place 'debug_print(' and ')' around | |
595 | the expression. Press 'Recalculate' and the output is sent to stderr, | |
596 | and to the debugger via the OutputDebugString routine. The output | |
597 | sent to stderr should be displayed in the console window that was | |
598 | opened when the emacs.exe executable was started. The output sent to | |
599 | the debugger should be displayed in the 'Debug' pane in the Output | |
600 | window. If Emacs was started from the debugger, a console window was | |
601 | opened at Emacs' startup; this console window also shows the output of | |
602 | 'debug_print'. | |
603 | ||
604 | For example, start and run Emacs in the debugger until it is waiting | |
605 | for user input. Then click on the `Break' button in the debugger to | |
606 | halt execution. Emacs should halt in `ZwUserGetMessage' waiting for | |
607 | an input event. Use the `Call Stack' window to select the procedure | |
608 | `w32_msp_pump' up the call stack (see below for why you have to do | |
609 | this). Open the QuickWatch window and enter | |
610 | "debug_print(Vexec_path)". Evaluating this expression will then print | |
3102e429 | 611 | out the contents of the Lisp variable `exec-path'. |
437368fe EZ |
612 | |
613 | If QuickWatch reports that the symbol is unknown, then check the call | |
614 | stack in the `Call Stack' window. If the selected frame in the call | |
615 | stack is not an Emacs procedure, then the debugger won't recognize | |
616 | Emacs symbols. Instead, select a frame that is inside an Emacs | |
617 | procedure and try using `debug_print' again. | |
618 | ||
619 | If QuickWatch invokes debug_print but nothing happens, then check the | |
620 | thread that is selected in the debugger. If the selected thread is | |
621 | not the last thread to run (the "current" thread), then it cannot be | |
622 | used to execute debug_print. Use the Debug menu to select the current | |
623 | thread and try using debug_print again. Note that the debugger halts | |
624 | execution (e.g., due to a breakpoint) in the context of the current | |
625 | thread, so this should only be a problem if you've explicitly switched | |
626 | threads. | |
627 | ||
3102e429 | 628 | It is also possible to keep appropriately masked and typecast Lisp |
437368fe EZ |
629 | symbols in the Watch window, this is more convenient when steeping |
630 | though the code. For instance, on entering apply_lambda, you can | |
631 | watch (struct Lisp_Symbol *) (0xfffffff & args[0]). | |
961e2394 EZ |
632 | |
633 | Optimizations often confuse the MS debugger. For example, the | |
634 | debugger will sometimes report wrong line numbers, e.g., when it | |
635 | prints the backtrace for a crash. It is usually best to look at the | |
636 | disassembly to determine exactly what code is being run--the | |
637 | disassembly will probably show several source lines followed by a | |
638 | block of assembler for those lines. The actual point where Emacs | |
639 | crashes will be one of those source lines, but not neccesarily the one | |
640 | that the debugger reports. | |
641 | ||
642 | Another problematic area with the MS debugger is with variables that | |
643 | are stored in registers: it will sometimes display wrong values for | |
644 | those variables. Usually you will not be able to see any value for a | |
645 | register variable, but if it is only being stored in a register | |
646 | temporarily, you will see an old value for it. Again, you need to | |
647 | look at the disassembly to determine which registers are being used, | |
648 | and look at those registers directly, to see the actual current values | |
649 | of these variables. |