blob: 0522f88bf7098fbc87fed958e10b40db47c429f7 [file] [log] [blame]
Willy Tarreau4e2b6462019-05-16 17:44:30 +02001/*
2 * Process debugging functions.
3 *
4 * Copyright 2000-2019 Willy Tarreau <willy@haproxy.org>.
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 *
11 */
12
13#include <signal.h>
14#include <time.h>
15#include <stdio.h>
Willy Tarreau6bdf3e92019-05-20 14:25:05 +020016#include <stdlib.h>
Willy Tarreau4e2b6462019-05-16 17:44:30 +020017
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020018#include <common/buf.h>
Willy Tarreau4e2b6462019-05-16 17:44:30 +020019#include <common/config.h>
20#include <common/debug.h>
21#include <common/hathreads.h>
22#include <common/initcall.h>
23#include <common/standard.h>
24
25#include <types/global.h>
26
27#include <proto/cli.h>
28#include <proto/fd.h>
29#include <proto/stream_interface.h>
30#include <proto/task.h>
31
Willy Tarreaua37cb182019-07-31 19:20:39 +020032/* mask of threads still having to dump, used to respect ordering. Only used
33 * when USE_THREAD_DUMP is set.
34 */
35volatile unsigned long threads_to_dump = 0;
36
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020037/* Dumps to the buffer some known information for the desired thread, and
38 * optionally extra info for the current thread. The dump will be appended to
39 * the buffer, so the caller is responsible for preliminary initializing it.
40 * The calling thread ID needs to be passed in <calling_tid> to display a star
Willy Tarreaue6a02fa2019-05-22 07:06:44 +020041 * in front of the calling thread's line (usually it's tid). Any stuck thread
42 * is also prefixed with a '>'.
Willy Tarreau4e2b6462019-05-16 17:44:30 +020043 */
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020044void ha_thread_dump(struct buffer *buf, int thr, int calling_tid)
Willy Tarreau4e2b6462019-05-16 17:44:30 +020045{
46 unsigned long thr_bit = 1UL << thr;
Willy Tarreau9c8800a2019-05-20 20:52:20 +020047 unsigned long long p = thread_info[thr].prev_cpu_time;
48 unsigned long long n = now_cpu_time_thread(&thread_info[thr]);
Willy Tarreaue6a02fa2019-05-22 07:06:44 +020049 int stuck = !!(thread_info[thr].flags & TI_FL_STUCK);
Willy Tarreau4e2b6462019-05-16 17:44:30 +020050
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020051 chunk_appendf(buf,
Willy Tarreaue6a02fa2019-05-22 07:06:44 +020052 "%c%cThread %-2u: act=%d glob=%d wq=%d rq=%d tl=%d tlsz=%d rqsz=%d\n"
Olivier Houchard305d5ab2019-07-24 18:07:06 +020053 " stuck=%d prof=%d",
Willy Tarreaue6a02fa2019-05-22 07:06:44 +020054 (thr == calling_tid) ? '*' : ' ', stuck ? '>' : ' ', thr + 1,
Olivier Houchardcfbb3e62019-05-29 19:22:43 +020055 thread_has_tasks(),
Willy Tarreau4e2b6462019-05-16 17:44:30 +020056 !!(global_tasks_mask & thr_bit),
57 !eb_is_empty(&task_per_thread[thr].timers),
58 !eb_is_empty(&task_per_thread[thr].rqueue),
59 !LIST_ISEMPTY(&task_per_thread[thr].task_list),
60 task_per_thread[thr].task_list_size,
61 task_per_thread[thr].rqueue_size,
Willy Tarreaue6a02fa2019-05-22 07:06:44 +020062 stuck,
Willy Tarreau4e2b6462019-05-16 17:44:30 +020063 !!(task_profiling_mask & thr_bit));
64
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020065 chunk_appendf(buf,
Willy Tarreau4e2b6462019-05-16 17:44:30 +020066 " harmless=%d wantrdv=%d",
67 !!(threads_harmless_mask & thr_bit),
68 !!(threads_want_rdv_mask & thr_bit));
Willy Tarreau4e2b6462019-05-16 17:44:30 +020069
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020070 chunk_appendf(buf, "\n");
Willy Tarreau9c8800a2019-05-20 20:52:20 +020071 chunk_appendf(buf, " cpu_ns: poll=%llu now=%llu diff=%llu\n", p, n, n-p);
Willy Tarreau4e2b6462019-05-16 17:44:30 +020072
73 /* this is the end of what we can dump from outside the thread */
74
75 if (thr != tid)
76 return;
77
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020078 chunk_appendf(buf, " curr_task=");
79 ha_task_dump(buf, curr_task, " ");
Willy Tarreau4e2b6462019-05-16 17:44:30 +020080}
81
82
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020083/* dumps into the buffer some information related to task <task> (which may
Willy Tarreau4e2b6462019-05-16 17:44:30 +020084 * either be a task or a tasklet, and prepend each line except the first one
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020085 * with <pfx>. The buffer is only appended and the first output starts by the
86 * pointer itself. The caller is responsible for making sure the task is not
87 * going to vanish during the dump.
Willy Tarreau4e2b6462019-05-16 17:44:30 +020088 */
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020089void ha_task_dump(struct buffer *buf, const struct task *task, const char *pfx)
Willy Tarreau4e2b6462019-05-16 17:44:30 +020090{
Willy Tarreau578ea8b2019-05-22 09:43:09 +020091 const struct stream *s = NULL;
92
Willy Tarreau14a1ab72019-05-17 10:34:25 +020093 if (!task) {
Willy Tarreau5cf64dd2019-05-17 10:36:08 +020094 chunk_appendf(buf, "0\n");
Willy Tarreau231ec392019-05-17 10:39:47 +020095 return;
96 }
97
Willy Tarreau20db9112019-05-17 14:14:35 +020098 if (TASK_IS_TASKLET(task))
99 chunk_appendf(buf,
100 "%p (tasklet) calls=%u\n",
101 task,
102 task->calls);
103 else
104 chunk_appendf(buf,
105 "%p (task) calls=%u last=%llu%s\n",
106 task,
107 task->calls,
108 task->call_date ? (unsigned long long)(now_mono_time() - task->call_date) : 0,
109 task->call_date ? " ns ago" : "");
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200110
Willy Tarreau5cf64dd2019-05-17 10:36:08 +0200111 chunk_appendf(buf, "%s"
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200112 " fct=%p (%s) ctx=%p\n",
113 pfx,
Willy Tarreau14a1ab72019-05-17 10:34:25 +0200114 task->process,
115 task->process == process_stream ? "process_stream" :
116 task->process == task_run_applet ? "task_run_applet" :
117 task->process == si_cs_io_cb ? "si_cs_io_cb" :
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200118 "?",
Willy Tarreau14a1ab72019-05-17 10:34:25 +0200119 task->context);
Willy Tarreau578ea8b2019-05-22 09:43:09 +0200120
121 if (task->process == process_stream && task->context)
122 s = (struct stream *)task->context;
123 else if (task->process == task_run_applet && task->context)
124 s = si_strm(((struct appctx *)task->context)->owner);
125 else if (task->process == si_cs_io_cb && task->context)
126 s = si_strm((struct stream_interface *)task->context);
127
128 if (s)
129 stream_dump(buf, s, pfx, '\n');
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200130}
131
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200132/* This function dumps all profiling settings. It returns 0 if the output
133 * buffer is full and it needs to be called again, otherwise non-zero.
134 */
135static int cli_io_handler_show_threads(struct appctx *appctx)
136{
137 struct stream_interface *si = appctx->owner;
138 int thr;
139
140 if (unlikely(si_ic(si)->flags & (CF_WRITE_ERROR|CF_SHUTW)))
141 return 1;
142
143 if (appctx->st0)
144 thr = appctx->st1;
145 else
146 thr = 0;
147
Willy Tarreau5cf64dd2019-05-17 10:36:08 +0200148 chunk_reset(&trash);
Willy Tarreauc7091d82019-05-17 10:08:49 +0200149 ha_thread_dump_all_to_trash();
Willy Tarreau5cf64dd2019-05-17 10:36:08 +0200150
151 if (ci_putchk(si_ic(si), &trash) == -1) {
152 /* failed, try again */
153 si_rx_room_blk(si);
154 appctx->st1 = thr;
155 return 0;
156 }
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200157 return 1;
158}
159
Willy Tarreau56131ca2019-05-20 13:48:29 +0200160/* dumps a state of all threads into the trash and on fd #2, then aborts. */
161void ha_panic()
162{
163 chunk_reset(&trash);
Willy Tarreaua9f9fc92019-05-20 17:45:35 +0200164 chunk_appendf(&trash, "Thread %u is about to kill the process.\n", tid + 1);
Willy Tarreau56131ca2019-05-20 13:48:29 +0200165 ha_thread_dump_all_to_trash();
Tim Duesterhusdda11552019-06-12 20:47:30 +0200166 shut_your_big_mouth_gcc(write(2, trash.area, trash.data));
Willy Tarreau56131ca2019-05-20 13:48:29 +0200167 for (;;)
168 abort();
169}
170
Willy Tarreau6bdf3e92019-05-20 14:25:05 +0200171#if defined(DEBUG_DEV)
172/* parse a "debug dev exit" command. It always returns 1, though it should never return. */
173static int debug_parse_cli_exit(char **args, char *payload, struct appctx *appctx, void *private)
174{
175 int code = atoi(args[3]);
176
177 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
178 return 1;
179
180 exit(code);
181 return 1;
182}
183
184/* parse a "debug dev close" command. It always returns 1. */
185static int debug_parse_cli_close(char **args, char *payload, struct appctx *appctx, void *private)
186{
187 int fd;
188
189 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
190 return 1;
191
192 if (!*args[3]) {
193 appctx->ctx.cli.msg = "Missing file descriptor number.\n";
194 goto reterr;
195 }
196
197 fd = atoi(args[3]);
198 if (fd < 0 || fd >= global.maxsock) {
199 appctx->ctx.cli.msg = "File descriptor out of range.\n";
200 goto reterr;
201 }
202
203 if (!fdtab[fd].owner) {
204 appctx->ctx.cli.msg = "File descriptor was already closed.\n";
205 goto retinfo;
206 }
207
208 fd_delete(fd);
209 return 1;
210 retinfo:
211 appctx->ctx.cli.severity = LOG_INFO;
212 appctx->st0 = CLI_ST_PRINT;
213 return 1;
214 reterr:
215 appctx->ctx.cli.severity = LOG_ERR;
216 appctx->st0 = CLI_ST_PRINT;
217 return 1;
218}
219
220/* parse a "debug dev delay" command. It always returns 1. */
221static int debug_parse_cli_delay(char **args, char *payload, struct appctx *appctx, void *private)
222{
223 int delay = atoi(args[3]);
224
225 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
226 return 1;
227
228 usleep((long)delay * 1000);
229 return 1;
230}
231
232/* parse a "debug dev log" command. It always returns 1. */
233static int debug_parse_cli_log(char **args, char *payload, struct appctx *appctx, void *private)
234{
235 int arg;
236
237 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
238 return 1;
239
240 chunk_reset(&trash);
241 for (arg = 3; *args[arg]; arg++) {
242 if (arg > 3)
243 chunk_strcat(&trash, " ");
244 chunk_strcat(&trash, args[arg]);
245 }
246
247 send_log(NULL, LOG_INFO, "%s\n", trash.area);
248 return 1;
249}
250
251/* parse a "debug dev loop" command. It always returns 1. */
252static int debug_parse_cli_loop(char **args, char *payload, struct appctx *appctx, void *private)
253{
254 struct timeval deadline, curr;
255 int loop = atoi(args[3]);
256
257 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
258 return 1;
259
260 gettimeofday(&curr, NULL);
261 tv_ms_add(&deadline, &curr, loop);
262
263 while (tv_ms_cmp(&curr, &deadline) < 0)
264 gettimeofday(&curr, NULL);
265
266 return 1;
267}
268
269/* parse a "debug dev panic" command. It always returns 1, though it should never return. */
270static int debug_parse_cli_panic(char **args, char *payload, struct appctx *appctx, void *private)
271{
272 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
273 return 1;
274
275 ha_panic();
276 return 1;
277}
278
279/* parse a "debug dev exec" command. It always returns 1. */
280static int debug_parse_cli_exec(char **args, char *payload, struct appctx *appctx, void *private)
281{
282 FILE *f;
283 int arg;
284
285 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
286 return 1;
287
288 chunk_reset(&trash);
289 for (arg = 3; *args[arg]; arg++) {
290 if (arg > 3)
291 chunk_strcat(&trash, " ");
292 chunk_strcat(&trash, args[arg]);
293 }
294
295 f = popen(trash.area, "re");
296 if (!f) {
297 appctx->ctx.cli.severity = LOG_ERR;
298 appctx->ctx.cli.msg = "Failed to execute command.\n";
299 appctx->st0 = CLI_ST_PRINT;
300 return 1;
301 }
302
303 chunk_reset(&trash);
304 while (1) {
305 size_t ret = fread(trash.area + trash.data, 1, trash.size - 20 - trash.data, f);
306 if (!ret)
307 break;
308 trash.data += ret;
309 if (trash.data + 20 == trash.size) {
310 chunk_strcat(&trash, "\n[[[TRUNCATED]]]\n");
311 break;
312 }
313 }
314
315 fclose(f);
316 trash.area[trash.data] = 0;
317 appctx->ctx.cli.severity = LOG_INFO;
318 appctx->ctx.cli.msg = trash.area;
319 appctx->st0 = CLI_ST_PRINT;
320 return 1;
321}
322
323/* parse a "debug dev hex" command. It always returns 1. */
324static int debug_parse_cli_hex(char **args, char *payload, struct appctx *appctx, void *private)
325{
326 unsigned long start, len;
327
328 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
329 return 1;
330
331 if (!*args[3]) {
332 appctx->ctx.cli.msg = "Missing memory address to dump from.\n";
333 goto reterr;
334 }
335
336 start = strtoul(args[3], NULL, 0);
337 if (!start) {
338 appctx->ctx.cli.msg = "Will not dump from NULL address.\n";
339 goto reterr;
340 }
341
342 /* by default, dump ~128 till next block of 16 */
343 len = strtoul(args[4], NULL, 0);
344 if (!len)
345 len = ((start + 128) & -16) - start;
346
347 chunk_reset(&trash);
Willy Tarreau37101052019-05-20 16:48:20 +0200348 dump_hex(&trash, " ", (const void *)start, len, 1);
Willy Tarreau6bdf3e92019-05-20 14:25:05 +0200349 trash.area[trash.data] = 0;
350 appctx->ctx.cli.severity = LOG_INFO;
351 appctx->ctx.cli.msg = trash.area;
352 appctx->st0 = CLI_ST_PRINT;
353 return 1;
354 reterr:
355 appctx->ctx.cli.severity = LOG_ERR;
356 appctx->st0 = CLI_ST_PRINT;
357 return 1;
358}
359
360/* parse a "debug dev tkill" command. It always returns 1. */
361static int debug_parse_cli_tkill(char **args, char *payload, struct appctx *appctx, void *private)
362{
363 int thr = 0;
364 int sig = SIGABRT;
365
366 if (!cli_has_level(appctx, ACCESS_LVL_ADMIN))
367 return 1;
368
369 if (*args[3])
370 thr = atoi(args[3]);
371
372 if (thr < 0 || thr > global.nbthread) {
373 appctx->ctx.cli.severity = LOG_ERR;
374 appctx->ctx.cli.msg = "Thread number out of range (use 0 for current).\n";
375 appctx->st0 = CLI_ST_PRINT;
376 return 1;
377 }
378
379 if (*args[4])
380 sig = atoi(args[4]);
381
Willy Tarreau6bdf3e92019-05-20 14:25:05 +0200382 if (thr)
Willy Tarreaufade80d2019-05-22 08:46:59 +0200383 ha_tkill(thr - 1, sig);
Willy Tarreau6bdf3e92019-05-20 14:25:05 +0200384 else
Willy Tarreau6bdf3e92019-05-20 14:25:05 +0200385 raise(sig);
386 return 1;
387}
388
389#endif
390
Willy Tarreauc7091d82019-05-17 10:08:49 +0200391#ifndef USE_THREAD_DUMP
392
393/* This function dumps all threads' state to the trash. This version is the
394 * most basic one, which doesn't inspect other threads.
395 */
396void ha_thread_dump_all_to_trash()
397{
398 unsigned int thr;
399
400 for (thr = 0; thr < global.nbthread; thr++)
401 ha_thread_dump(&trash, thr, tid);
402}
403
404#else /* below USE_THREAD_DUMP is set */
405
Willy Tarreauddd85332019-05-22 06:28:54 +0200406/* The signal to trigger a debug dump on a thread is SIGURG. It has the benefit
407 * of not stopping gdb by default, so that issuing "show threads" in a process
408 * being debugged has no adverse effect.
409 */
410#define DEBUGSIG SIGURG
Willy Tarreauc7091d82019-05-17 10:08:49 +0200411
Willy Tarreauc7091d82019-05-17 10:08:49 +0200412/* ID of the thread requesting the dump */
413static unsigned int thread_dump_tid;
414
415/* points to the buffer where the dump functions should write. It must
416 * have already been initialized by the requester. Nothing is done if
417 * it's NULL.
418 */
419struct buffer *thread_dump_buffer = NULL;
420
421void ha_thread_dump_all_to_trash()
422{
Willy Tarreauc7091d82019-05-17 10:08:49 +0200423 unsigned long old;
424
425 while (1) {
426 old = 0;
427 if (HA_ATOMIC_CAS(&threads_to_dump, &old, all_threads_mask))
428 break;
429 ha_thread_relax();
430 }
431
432 thread_dump_buffer = &trash;
433 thread_dump_tid = tid;
Willy Tarreaufade80d2019-05-22 08:46:59 +0200434 ha_tkillall(DEBUGSIG);
Willy Tarreauc7091d82019-05-17 10:08:49 +0200435}
436
437/* handles DEBUGSIG to dump the state of the thread it's working on */
438void debug_handler(int sig, siginfo_t *si, void *arg)
439{
440 /* There are 4 phases in the dump process:
441 * 1- wait for our turn, i.e. when all lower bits are gone.
442 * 2- perform the action if our bit is set
443 * 3- remove our bit to let the next one go, unless we're
Willy Tarreauc0773622019-07-31 19:15:45 +0200444 * the last one and have to put them all as a signal
445 * 4- wait out bit to re-appear, then clear it and quit.
Willy Tarreauc7091d82019-05-17 10:08:49 +0200446 */
447
448 /* wait for all previous threads to finish first */
449 while (threads_to_dump & (tid_bit - 1))
450 ha_thread_relax();
451
452 /* dump if needed */
453 if (threads_to_dump & tid_bit) {
454 if (thread_dump_buffer)
455 ha_thread_dump(thread_dump_buffer, tid, thread_dump_tid);
456 if ((threads_to_dump & all_threads_mask) == tid_bit) {
457 /* last one */
Willy Tarreauc0773622019-07-31 19:15:45 +0200458 HA_ATOMIC_STORE(&threads_to_dump, all_threads_mask);
Willy Tarreauc7091d82019-05-17 10:08:49 +0200459 thread_dump_buffer = NULL;
460 }
461 else
462 HA_ATOMIC_AND(&threads_to_dump, ~tid_bit);
463 }
464
465 /* now wait for all others to finish dumping. The last one will set all
Willy Tarreauc0773622019-07-31 19:15:45 +0200466 * bits again to broadcast the leaving condition so we'll see ourselves
467 * present again. This way the threads_to_dump variable never passes to
468 * zero until all visitors have stopped waiting.
Willy Tarreauc7091d82019-05-17 10:08:49 +0200469 */
Willy Tarreauc0773622019-07-31 19:15:45 +0200470 while (!(threads_to_dump & tid_bit))
471 ha_thread_relax();
472 HA_ATOMIC_AND(&threads_to_dump, ~tid_bit);
Willy Tarreaue6a02fa2019-05-22 07:06:44 +0200473
474 /* mark the current thread as stuck to detect it upon next invocation
475 * if it didn't move.
476 */
477 if (!((threads_harmless_mask|sleeping_thread_mask) & tid_bit))
478 ti->flags |= TI_FL_STUCK;
Willy Tarreauc7091d82019-05-17 10:08:49 +0200479}
480
481static int init_debug_per_thread()
482{
483 sigset_t set;
484
485 /* unblock the DEBUGSIG signal we intend to use */
486 sigemptyset(&set);
487 sigaddset(&set, DEBUGSIG);
488 ha_sigmask(SIG_UNBLOCK, &set, NULL);
489 return 1;
490}
491
492static int init_debug()
493{
494 struct sigaction sa;
495
496 sa.sa_handler = NULL;
497 sa.sa_sigaction = debug_handler;
498 sigemptyset(&sa.sa_mask);
499 sa.sa_flags = SA_SIGINFO;
500 sigaction(DEBUGSIG, &sa, NULL);
501 return 0;
502}
503
504REGISTER_POST_CHECK(init_debug);
505REGISTER_PER_THREAD_INIT(init_debug_per_thread);
506
507#endif /* USE_THREAD_DUMP */
508
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200509/* register cli keywords */
510static struct cli_kw_list cli_kws = {{ },{
Willy Tarreau6bdf3e92019-05-20 14:25:05 +0200511#if defined(DEBUG_DEV)
512 {{ "debug", "dev", "close", NULL }, "debug dev close <fd> : close this file descriptor", debug_parse_cli_close, NULL },
513 {{ "debug", "dev", "delay", NULL }, "debug dev delay [ms] : sleep this long", debug_parse_cli_delay, NULL },
514 {{ "debug", "dev", "exec", NULL }, "debug dev exec [cmd] ... : show this command's output", debug_parse_cli_exec, NULL },
515 {{ "debug", "dev", "exit", NULL }, "debug dev exit [code] : immediately exit the process", debug_parse_cli_exit, NULL },
516 {{ "debug", "dev", "hex", NULL }, "debug dev hex <addr> [len]: dump a memory area", debug_parse_cli_hex, NULL },
517 {{ "debug", "dev", "log", NULL }, "debug dev log [msg] ... : send this msg to global logs", debug_parse_cli_log, NULL },
518 {{ "debug", "dev", "loop", NULL }, "debug dev loop [ms] : loop this long", debug_parse_cli_loop, NULL },
519 {{ "debug", "dev", "panic", NULL }, "debug dev panic : immediately trigger a panic", debug_parse_cli_panic, NULL },
520 {{ "debug", "dev", "tkill", NULL }, "debug dev tkill [thr] [sig] : send signal to thread", debug_parse_cli_tkill, NULL },
521#endif
Willy Tarreau4e2b6462019-05-16 17:44:30 +0200522 { { "show", "threads", NULL }, "show threads : show some threads debugging information", NULL, cli_io_handler_show_threads, NULL },
523 {{},}
524}};
525
526INITCALL1(STG_REGISTER, cli_register_kw, &cli_kws);