2 * Bacula thread watchdog routine. General routine that monitors
3 * the daemon and signals a thread if it is blocked on a BSOCK
4 * too long. This prevents catastropic long waits -- generally
5 * due to Windows "hanging" the app.
7 * Kern Sibbald, January MMII
11 Copyright (C) 2000-2003 Kern Sibbald and John Walker
13 This program is free software; you can redistribute it and/or
14 modify it under the terms of the GNU General Public License as
15 published by the Free Software Foundation; either version 2 of
16 the License, or (at your option) any later version.
18 This program is distributed in the hope that it will be useful,
19 but WITHOUT ANY WARRANTY; without even the implied warranty of
20 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
21 General Public License for more details.
23 You should have received a copy of the GNU General Public
24 License along with this program; if not, write to the Free
25 Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
33 /* This breaks Kern's #include rules, but I don't want to put it into bacula.h
34 * until it has been discussed with him */
35 #include "bsd_queue.h"
37 /* Exported globals */
38 time_t watchdog_time; /* this has granularity of SLEEP_TIME */
40 #define SLEEP_TIME 1 /* examine things every second */
42 /* Forward referenced functions */
43 static void *watchdog_thread(void *arg);
46 static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
47 static pthread_cond_t timer = PTHREAD_COND_INITIALIZER;
49 static bool wd_is_init = false;
51 /* Forward referenced callback functions */
52 static pthread_t wd_tid;
55 static TAILQ_HEAD(/* no struct */, s_watchdog_t) wd_queue =
56 TAILQ_HEAD_INITIALIZER(wd_queue);
57 static TAILQ_HEAD(/* no struct */, s_watchdog_t) wd_inactive =
58 TAILQ_HEAD_INITIALIZER(wd_inactive);
61 * Start watchdog thread
63 * Returns: 0 on success
66 int start_watchdog(void)
70 Dmsg0(200, "Initialising NicB-hacked watchdog thread\n");
71 watchdog_time = time(NULL);
73 if ((stat = pthread_create(&wd_tid, NULL, watchdog_thread, NULL)) != 0) {
81 * Terminate the watchdog thread
83 * Returns: 0 on success
86 int stop_watchdog(void)
95 Dmsg0(200, "Sending stop signal to NicB-hacked watchdog thread\n");
98 stat = pthread_cond_signal(&timer);
103 stat = pthread_join(wd_tid, NULL);
105 TAILQ_FOREACH_SAFE(p, &wd_queue, qe, n) {
106 TAILQ_REMOVE(&wd_queue, p, qe);
107 if (p->destructor != NULL) {
113 TAILQ_FOREACH_SAFE(p, &wd_inactive, qe, n) {
114 TAILQ_REMOVE(&wd_inactive, p, qe);
115 if (p->destructor != NULL) {
124 watchdog_t *watchdog_new(void)
126 watchdog_t *wd = (watchdog_t *) malloc(sizeof(watchdog_t));
129 Emsg0(M_ABORT, 0, "BUG! watchdog_new called before start_watchdog\n");
138 wd->destructor = NULL;
144 bool register_watchdog(watchdog_t *wd)
147 Emsg0(M_ABORT, 0, "BUG! register_watchdog called before start_watchdog\n");
149 if (wd->callback == NULL) {
150 Emsg1(M_ABORT, 0, "BUG! Watchdog %p has NULL callback\n", wd);
152 if (wd->interval == 0) {
153 Emsg1(M_ABORT, 0, "BUG! Watchdog %p has zero interval\n", wd);
157 wd->next_fire = watchdog_time + wd->interval;
158 TAILQ_INSERT_TAIL(&wd_queue, wd, qe);
159 Dmsg3(200, "Registered watchdog %p, interval %d%s\n",
160 wd, wd->interval, wd->one_shot ? " one shot" : "");
166 bool unregister_watchdog_unlocked(watchdog_t *wd)
171 Emsg0(M_ABORT, 0, "BUG! unregister_watchdog_unlocked called before start_watchdog\n");
174 TAILQ_FOREACH_SAFE(p, &wd_queue, qe, n) {
176 TAILQ_REMOVE(&wd_queue, wd, qe);
177 Dmsg1(200, "Unregistered watchdog %p\n", wd);
182 TAILQ_FOREACH_SAFE(p, &wd_inactive, qe, n) {
184 TAILQ_REMOVE(&wd_inactive, wd, qe);
185 Dmsg1(200, "Unregistered inactive watchdog %p\n", wd);
190 Dmsg1(200, "Failed to unregister watchdog %p\n", wd);
195 bool unregister_watchdog(watchdog_t *wd)
200 Emsg0(M_ABORT, 0, "BUG! unregister_watchdog called before start_watchdog\n");
204 ret = unregister_watchdog_unlocked(wd);
210 static void *watchdog_thread(void *arg)
212 Dmsg0(200, "NicB-reworked watchdog thread entered\n");
223 watchdog_time = time(NULL);
225 TAILQ_FOREACH_SAFE(p, &wd_queue, qe, n) {
226 if (p->next_fire < watchdog_time) {
227 /* Run the callback */
230 /* Reschedule (or move to inactive list if it's a one-shot timer) */
232 TAILQ_REMOVE(&wd_queue, p, qe);
233 TAILQ_INSERT_TAIL(&wd_inactive, p, qe);
235 p->next_fire = watchdog_time + p->interval;
240 bmicrosleep(SLEEP_TIME, 0);
243 Dmsg0(200, "NicB-reworked watchdog thread exited\n");