usr.sbin/watchdogd/watchdogd.c

   1 /*
   2  * Copyright (c) 2003-2004  Sean M. Kelly <smkelly@FreeBSD.org>
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  24  * SUCH DAMAGE.
  25  */
  26
  27 /*
  28  * Software watchdog daemon.
  29  */
  30
  31 #include <sys/types.h>
  32 __FBSDID("$FreeBSD$");
  33
  34 #include <sys/param.h>
  35 #include <sys/rtprio.h>
  36 #include <sys/stat.h>
  37 #include <sys/time.h>
  38 #include <sys/watchdog.h>
  39
  40 #include <err.h>
  41 #include <errno.h>
  42 #include <fcntl.h>
  43 #include <libutil.h>
  44 #include <math.h>
  45 #include <paths.h>
  46 #include <signal.h>
  47 #include <stdio.h>
  48 #include <stdlib.h>
  49 #include <string.h>
  50 #include <sysexits.h>
  51 #include <unistd.h>
  52
  53 static void     parseargs(int, char *[]);
  54 static void     sighandler(int);
  55 static void     watchdog_loop(void);
  56 static int      watchdog_init(void);
  57 static int      watchdog_onoff(int onoff);
  58 static int      watchdog_patpat(u_int timeout);
  59 static void     usage(void);
  60
  61 int debugging = 0;
  62 int end_program = 0;
  63 const char *pidfile = _PATH_VARRUN "watchdogd.pid";
  64 int reset_mib[3];
  65 size_t reset_miblen = 3;
  66 u_int timeout = WD_TO_16SEC;
  67 u_int passive = 0;
  68 int is_daemon = 0;
  69 int fd = -1;
  70 int nap = 1;
  71 char *test_cmd = NULL;
  72
  73 /*
  74  * Periodically pat the watchdog, preventing it from firing.
  75  */
  76 int
  77 main(int argc, char *argv[])
  78 {
  79         struct rtprio rtp;
  80         struct pidfh *pfh;
  81         pid_t otherpid;
  82
  83         if (getuid() != 0)
  84                 errx(EX_SOFTWARE, "not super user");
  85
  86         parseargs(argc, argv);
  87
  88         rtp.type = RTP_PRIO_REALTIME;
  89         rtp.prio = 0;
  90         if (rtprio(RTP_SET, 0, &rtp) == -1)
  91                 err(EX_OSERR, "rtprio");
  92
  93         if (watchdog_init() == -1)
  94                 errx(EX_SOFTWARE, "unable to initialize watchdog");
  95
  96         if (is_daemon) {
  97                 if (watchdog_onoff(1) == -1)
  98                         exit(EX_SOFTWARE);
  99
 100                 pfh = pidfile_open(pidfile, 0600, &otherpid);
 101                 if (pfh == NULL) {
 102                         if (errno == EEXIST) {
 103                                 errx(EX_SOFTWARE, "%s already running, pid: %d",
 104                                     getprogname(), otherpid);
 105                         }
 106                         warn("Cannot open or create pidfile");
 107                 }
 108
 109                 if (debugging == 0 && daemon(0, 0) == -1) {
 110                         watchdog_onoff(0);
 111                         pidfile_remove(pfh);
 112                         err(EX_OSERR, "daemon");
 113                 }
 114
 115                 signal(SIGHUP, SIG_IGN);
 116                 signal(SIGINT, sighandler);
 117                 signal(SIGTERM, sighandler);
 118
 119                 pidfile_write(pfh);
 120
 121                 watchdog_loop();
 122
 123                 /* exiting */
 124                 pidfile_remove(pfh);
 125                 return (EX_OK);
 126         } else {
 127                 if (passive)
 128                         timeout |= WD_PASSIVE;
 129                 else
 130                         timeout |= WD_ACTIVE;
 131                 if (watchdog_patpat(timeout) < 0)
 132                         err(EX_OSERR, "patting the dog");
 133                 return (EX_OK);
 134         }
 135 }
 136
 137 /*
 138  * Catch signals and begin shutdown process.
 139  */
 140 static void
 141 sighandler(int signum)
 142 {
 143
 144         if (signum == SIGINT || signum == SIGTERM)
 145                 end_program = 1;
 146 }
 147
 148 /*
 149  * Open the watchdog device.
 150  */
 151 static int
 152 watchdog_init()
 153 {
 154
 155         fd = open("/dev/" _PATH_WATCHDOG, O_RDWR);
 156         if (fd >= 0)
 157                 return (0);
 158         warn("Could not open watchdog device");
 159         return (-1);
 160 }
 161
 162 /*
 163  * Main program loop which is iterated every second.
 164  */
 165 static void
 166 watchdog_loop(void)
 167 {
 168         struct stat sb;
 169         int failed;
 170
 171         while (end_program != 2) {
 172                 failed = 0;
 173
 174                 if (test_cmd != NULL)
 175                         failed = system(test_cmd);
 176                 else
 177                         failed = stat("/etc", &sb);
 178
 179                 if (failed == 0)
 180                         watchdog_patpat(timeout|WD_ACTIVE);
 181                 sleep(nap);
 182
 183                 if (end_program != 0) {
 184                         if (watchdog_onoff(0) == 0) {
 185                                 end_program = 2;
 186                         } else {
 187                                 warnx("Could not stop the watchdog, not exitting");
 188                                 end_program = 0;
 189                         }
 190                 }
 191         }
 192 }
 193
 194 /*
 195  * Reset the watchdog timer. This function must be called periodically
 196  * to keep the watchdog from firing.
 197  */
 198 int
 199 watchdog_patpat(u_int t)
 200 {
 201
 202         return ioctl(fd, WDIOCPATPAT, &t);
 203 }
 204
 205 /*
 206  * Toggle the kernel's watchdog. This routine is used to enable and
 207  * disable the watchdog.
 208  */
 209 static int
 210 watchdog_onoff(int onoff)
 211 {
 212
 213         if (onoff)
 214                 return watchdog_patpat((timeout|WD_ACTIVE));
 215         else
 216                 return watchdog_patpat(0);
 217 }
 218
 219 /*
 220  * Tell user how to use the program.
 221  */
 222 static void
 223 usage()
 224 {
 225         if (is_daemon)
 226                 fprintf(stderr, "usage: watchdogd [-d] [-e cmd] [-I file] [-s sleep] [-t timeout]\n");
 227         else
 228                 fprintf(stderr, "usage: watchdog [-d] [-t timeout]\n");
 229         exit(EX_USAGE);
 230 }
 231
 232 /*
 233  * Handle the few command line arguments supported.
 234  */
 235 static void
 236 parseargs(int argc, char *argv[])
 237 {
 238         int c;
 239         char *p;
 240         double a;
 241
 242         c = strlen(argv[0]);
 243         if (argv[0][c - 1] == 'd')
 244                 is_daemon = 1;
 245         while ((c = getopt(argc, argv,
 246             is_daemon ? "I:de:s:t:?" : "dt:?")) != -1) {
 247                 switch (c) {
 248                 case 'I':
 249                         pidfile = optarg;
 250                         break;
 251                 case 'd':
 252                         debugging = 1;
 253                         break;
 254                 case 'e':
 255                         test_cmd = strdup(optarg);
 256                         break;
 257 #ifdef notyet
 258                 case 'p':
 259                         passive = 1;
 260                         break;
 261 #endif
 262                 case 's':
 263                         p = NULL;
 264                         errno = 0;
 265                         nap = strtol(optarg, &p, 0);
 266                         if ((p != NULL && *p != '\0') || errno != 0)
 267                                 errx(EX_USAGE, "-s argument is not a number");
 268                         break;
 269                 case 't':
 270                         p = NULL;
 271                         errno = 0;
 272                         a = strtod(optarg, &p);
 273                         if ((p != NULL && *p != '\0') || errno != 0)
 274                                 errx(EX_USAGE, "-t argument is not a number");
 275                         if (a < 0)
 276                                 errx(EX_USAGE, "-t argument must be positive");
 277                         if (a == 0)
 278                                 timeout = WD_TO_NEVER;
 279                         else
 280                                 timeout = 1.0 + log(a * 1e9) / log(2.0);
 281                         if (debugging)
 282                                 printf("Timeout is 2^%d nanoseconds\n",
 283                                     timeout);
 284                         break;
 285                 case '?':
 286                 default:
 287                         usage();
 288                         /* NOTREACHED */
 289                 }
 290         }
 291         if (argc != optind)
 292                 errx(EX_USAGE, "extra arguments.");
 293         if (is_daemon && timeout < WD_TO_1SEC)
 294                 errx(EX_USAGE, "-t argument is less than one second.");
 295 }