danicoin/external/unbound/daemon/unbound.c
Erik de Castro Lopo a85b5759f3 Upgrade unbound library
These files were pulled from the 1.6.3 release tarball.

This new version builds against OpenSSL version 1.1 which will be
the default in the new Debian Stable which is due to be released
RealSoonNow (tm).
2017-06-17 23:04:00 +10:00

739 lines
21 KiB
C

/*
* daemon/unbound.c - main program for unbound DNS resolver daemon.
*
* Copyright (c) 2007, NLnet Labs. All rights reserved.
*
* This software is open source.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* Redistributions in binary form must reproduce the above copyright notice,
* this list of conditions and the following disclaimer in the documentation
* and/or other materials provided with the distribution.
*
* Neither the name of the NLNET LABS nor the names of its contributors may
* be used to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
* TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
* NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
/**
* \file
*
* Main program to start the DNS resolver daemon.
*/
#include "config.h"
#ifdef HAVE_GETOPT_H
#include <getopt.h>
#endif
#include <sys/time.h>
#include "util/log.h"
#include "daemon/daemon.h"
#include "daemon/remote.h"
#include "util/config_file.h"
#include "util/storage/slabhash.h"
#include "services/listen_dnsport.h"
#include "services/cache/rrset.h"
#include "services/cache/infra.h"
#include "util/fptr_wlist.h"
#include "util/data/msgreply.h"
#include "util/module.h"
#include "util/net_help.h"
#include "util/ub_event.h"
#include <signal.h>
#include <fcntl.h>
#include <openssl/crypto.h>
#ifdef HAVE_PWD_H
#include <pwd.h>
#endif
#ifdef HAVE_GRP_H
#include <grp.h>
#endif
#ifndef S_SPLINT_S
/* splint chokes on this system header file */
#ifdef HAVE_SYS_RESOURCE_H
#include <sys/resource.h>
#endif
#endif /* S_SPLINT_S */
#ifdef HAVE_LOGIN_CAP_H
#include <login_cap.h>
#endif
#ifdef UB_ON_WINDOWS
# include "winrc/win_svc.h"
#endif
#ifdef HAVE_NSS
/* nss3 */
# include "nss.h"
#endif
/** print usage. */
static void usage(void)
{
const char** m;
const char *evnm="event", *evsys="", *evmethod="";
time_t t;
struct timeval now;
struct ub_event_base* base;
printf("usage: unbound [options]\n");
printf(" start unbound daemon DNS resolver.\n");
printf("-h this help\n");
printf("-c file config file to read instead of %s\n", CONFIGFILE);
printf(" file format is described in unbound.conf(5).\n");
printf("-d do not fork into the background.\n");
printf("-v verbose (more times to increase verbosity)\n");
#ifdef UB_ON_WINDOWS
printf("-w opt windows option: \n");
printf(" install, remove - manage the services entry\n");
printf(" service - used to start from services control panel\n");
#endif
printf("Version %s\n", PACKAGE_VERSION);
base = ub_default_event_base(0,&t,&now);
ub_get_event_sys(base, &evnm, &evsys, &evmethod);
printf("linked libs: %s %s (it uses %s), %s\n",
evnm, evsys, evmethod,
#ifdef HAVE_SSL
# ifdef SSLEAY_VERSION
SSLeay_version(SSLEAY_VERSION)
# else
OpenSSL_version(OPENSSL_VERSION)
# endif
#elif defined(HAVE_NSS)
NSS_GetVersion()
#elif defined(HAVE_NETTLE)
"nettle"
#endif
);
printf("linked modules:");
for(m = module_list_avail(); *m; m++)
printf(" %s", *m);
printf("\n");
printf("BSD licensed, see LICENSE in source package for details.\n");
printf("Report bugs to %s\n", PACKAGE_BUGREPORT);
ub_event_base_free(base);
}
#ifndef unbound_testbound
int replay_var_compare(const void* ATTR_UNUSED(a), const void* ATTR_UNUSED(b))
{
log_assert(0);
return 0;
}
#endif
/** check file descriptor count */
static void
checkrlimits(struct config_file* cfg)
{
#ifndef S_SPLINT_S
#ifdef HAVE_GETRLIMIT
/* list has number of ports to listen to, ifs number addresses */
int list = ((cfg->do_udp?1:0) + (cfg->do_tcp?1 +
(int)cfg->incoming_num_tcp:0));
size_t listen_ifs = (size_t)(cfg->num_ifs==0?
((cfg->do_ip4 && !cfg->if_automatic?1:0) +
(cfg->do_ip6?1:0)):cfg->num_ifs);
size_t listen_num = list*listen_ifs;
size_t outudpnum = (size_t)cfg->outgoing_num_ports;
size_t outtcpnum = cfg->outgoing_num_tcp;
size_t misc = 4; /* logfile, pidfile, stdout... */
size_t perthread_noudp = listen_num + outtcpnum +
2/*cmdpipe*/ + 2/*libevent*/ + misc;
size_t perthread = perthread_noudp + outudpnum;
#if !defined(HAVE_PTHREAD) && !defined(HAVE_SOLARIS_THREADS)
int numthread = 1; /* it forks */
#else
int numthread = (cfg->num_threads?cfg->num_threads:1);
#endif
size_t total = numthread * perthread + misc;
size_t avail;
struct rlimit rlim;
if(total > 1024 &&
strncmp(ub_event_get_version(), "mini-event", 10) == 0) {
log_warn("too many file descriptors requested. The builtin"
"mini-event cannot handle more than 1024. Config "
"for less fds or compile with libevent");
if(numthread*perthread_noudp+15 > 1024)
fatal_exit("too much tcp. not enough fds.");
cfg->outgoing_num_ports = (int)((1024
- numthread*perthread_noudp
- 10 /* safety margin */) /numthread);
log_warn("continuing with less udp ports: %u",
cfg->outgoing_num_ports);
total = 1024;
}
if(perthread > 64 &&
strncmp(ub_event_get_version(), "winsock-event", 13) == 0) {
log_err("too many file descriptors requested. The winsock"
" event handler cannot handle more than 64 per "
" thread. Config for less fds");
if(perthread_noudp+2 > 64)
fatal_exit("too much tcp. not enough fds.");
cfg->outgoing_num_ports = (int)((64
- perthread_noudp
- 2/* safety margin */));
log_warn("continuing with less udp ports: %u",
cfg->outgoing_num_ports);
total = numthread*(perthread_noudp+
(size_t)cfg->outgoing_num_ports)+misc;
}
if(getrlimit(RLIMIT_NOFILE, &rlim) < 0) {
log_warn("getrlimit: %s", strerror(errno));
return;
}
if(rlim.rlim_cur == (rlim_t)RLIM_INFINITY)
return;
if((size_t)rlim.rlim_cur < total) {
avail = (size_t)rlim.rlim_cur;
rlim.rlim_cur = (rlim_t)(total + 10);
rlim.rlim_max = (rlim_t)(total + 10);
#ifdef HAVE_SETRLIMIT
if(setrlimit(RLIMIT_NOFILE, &rlim) < 0) {
log_warn("setrlimit: %s", strerror(errno));
#endif
log_warn("cannot increase max open fds from %u to %u",
(unsigned)avail, (unsigned)total+10);
/* check that calculation below does not underflow,
* with 15 as margin */
if(numthread*perthread_noudp+15 > avail)
fatal_exit("too much tcp. not enough fds.");
cfg->outgoing_num_ports = (int)((avail
- numthread*perthread_noudp
- 10 /* safety margin */) /numthread);
log_warn("continuing with less udp ports: %u",
cfg->outgoing_num_ports);
log_warn("increase ulimit or decrease threads, "
"ports in config to remove this warning");
return;
#ifdef HAVE_SETRLIMIT
}
#endif
verbose(VERB_ALGO, "increased limit(open files) from %u to %u",
(unsigned)avail, (unsigned)total+10);
}
#else
(void)cfg;
#endif /* HAVE_GETRLIMIT */
#endif /* S_SPLINT_S */
}
/** set default logfile identity based on value from argv[0] at startup **/
static void
log_ident_set_fromdefault(struct config_file* cfg,
const char *log_default_identity)
{
if(cfg->log_identity == NULL || cfg->log_identity[0] == 0)
log_ident_set(log_default_identity);
else
log_ident_set(cfg->log_identity);
}
/** set verbosity, check rlimits, cache settings */
static void
apply_settings(struct daemon* daemon, struct config_file* cfg,
int cmdline_verbose, int debug_mode, const char* log_default_identity)
{
/* apply if they have changed */
verbosity = cmdline_verbose + cfg->verbosity;
if (debug_mode > 1) {
cfg->use_syslog = 0;
free(cfg->logfile);
cfg->logfile = NULL;
}
daemon_apply_cfg(daemon, cfg);
checkrlimits(cfg);
if (cfg->use_systemd && cfg->do_daemonize) {
log_warn("use-systemd and do-daemonize should not be enabled at the same time");
}
log_ident_set_fromdefault(cfg, log_default_identity);
}
#ifdef HAVE_KILL
/** Read existing pid from pidfile.
* @param file: file name of pid file.
* @return: the pid from the file or -1 if none.
*/
static pid_t
readpid (const char* file)
{
int fd;
pid_t pid;
char pidbuf[32];
char* t;
ssize_t l;
if ((fd = open(file, O_RDONLY)) == -1) {
if(errno != ENOENT)
log_err("Could not read pidfile %s: %s",
file, strerror(errno));
return -1;
}
if (((l = read(fd, pidbuf, sizeof(pidbuf)))) == -1) {
if(errno != ENOENT)
log_err("Could not read pidfile %s: %s",
file, strerror(errno));
close(fd);
return -1;
}
close(fd);
/* Empty pidfile means no pidfile... */
if (l == 0) {
return -1;
}
pidbuf[sizeof(pidbuf)-1] = 0;
pid = (pid_t)strtol(pidbuf, &t, 10);
if (*t && *t != '\n') {
return -1;
}
return pid;
}
/** write pid to file.
* @param pidfile: file name of pid file.
* @param pid: pid to write to file.
*/
static void
writepid (const char* pidfile, pid_t pid)
{
FILE* f;
if ((f = fopen(pidfile, "w")) == NULL ) {
log_err("cannot open pidfile %s: %s",
pidfile, strerror(errno));
return;
}
if(fprintf(f, "%lu\n", (unsigned long)pid) < 0) {
log_err("cannot write to pidfile %s: %s",
pidfile, strerror(errno));
}
fclose(f);
}
/**
* check old pid file.
* @param pidfile: the file name of the pid file.
* @param inchroot: if pidfile is inchroot and we can thus expect to
* be able to delete it.
*/
static void
checkoldpid(char* pidfile, int inchroot)
{
pid_t old;
if((old = readpid(pidfile)) != -1) {
/* see if it is still alive */
if(kill(old, 0) == 0 || errno == EPERM)
log_warn("unbound is already running as pid %u.",
(unsigned)old);
else if(inchroot)
log_warn("did not exit gracefully last time (%u)",
(unsigned)old);
}
}
#endif /* HAVE_KILL */
/** detach from command line */
static void
detach(void)
{
#if defined(HAVE_DAEMON) && !defined(DEPRECATED_DAEMON)
/* use POSIX daemon(3) function */
if(daemon(1, 0) != 0)
fatal_exit("daemon failed: %s", strerror(errno));
#else /* no HAVE_DAEMON */
#ifdef HAVE_FORK
int fd;
/* Take off... */
switch (fork()) {
case 0:
break;
case -1:
fatal_exit("fork failed: %s", strerror(errno));
default:
/* exit interactive session */
exit(0);
}
/* detach */
#ifdef HAVE_SETSID
if(setsid() == -1)
fatal_exit("setsid() failed: %s", strerror(errno));
#endif
if ((fd = open("/dev/null", O_RDWR, 0)) != -1) {
(void)dup2(fd, STDIN_FILENO);
(void)dup2(fd, STDOUT_FILENO);
(void)dup2(fd, STDERR_FILENO);
if (fd > 2)
(void)close(fd);
}
#endif /* HAVE_FORK */
#endif /* HAVE_DAEMON */
}
/** daemonize, drop user priviliges and chroot if needed */
static void
perform_setup(struct daemon* daemon, struct config_file* cfg, int debug_mode,
const char** cfgfile)
{
#ifdef HAVE_KILL
int pidinchroot;
#endif
#ifdef HAVE_GETPWNAM
struct passwd *pwd = NULL;
if(cfg->username && cfg->username[0]) {
if((pwd = getpwnam(cfg->username)) == NULL)
fatal_exit("user '%s' does not exist.", cfg->username);
/* endpwent below, in case we need pwd for setusercontext */
}
#endif
#ifdef UB_ON_WINDOWS
w_config_adjust_directory(cfg);
#endif
/* init syslog (as root) if needed, before daemonize, otherwise
* a fork error could not be printed since daemonize closed stderr.*/
if(cfg->use_syslog) {
log_init(cfg->logfile, cfg->use_syslog, cfg->chrootdir);
}
/* if using a logfile, we cannot open it because the logfile would
* be created with the wrong permissions, we cannot chown it because
* we cannot chown system logfiles, so we do not open at all.
* So, using a logfile, the user does not see errors unless -d is
* given to unbound on the commandline. */
/* read ssl keys while superuser and outside chroot */
#ifdef HAVE_SSL
if(!(daemon->rc = daemon_remote_create(cfg)))
fatal_exit("could not set up remote-control");
if(cfg->ssl_service_key && cfg->ssl_service_key[0]) {
if(!(daemon->listen_sslctx = listen_sslctx_create(
cfg->ssl_service_key, cfg->ssl_service_pem, NULL)))
fatal_exit("could not set up listen SSL_CTX");
}
if(!(daemon->connect_sslctx = connect_sslctx_create(NULL, NULL, NULL)))
fatal_exit("could not set up connect SSL_CTX");
#endif
#ifdef HAVE_KILL
/* true if pidfile is inside chrootdir, or nochroot */
pidinchroot = !(cfg->chrootdir && cfg->chrootdir[0]) ||
(cfg->chrootdir && cfg->chrootdir[0] &&
strncmp(cfg->pidfile, cfg->chrootdir,
strlen(cfg->chrootdir))==0);
/* check old pid file before forking */
if(cfg->pidfile && cfg->pidfile[0]) {
/* calculate position of pidfile */
if(cfg->pidfile[0] == '/')
daemon->pidfile = strdup(cfg->pidfile);
else daemon->pidfile = fname_after_chroot(cfg->pidfile,
cfg, 1);
if(!daemon->pidfile)
fatal_exit("pidfile alloc: out of memory");
checkoldpid(daemon->pidfile, pidinchroot);
}
#endif
/* daemonize because pid is needed by the writepid func */
if(!debug_mode && cfg->do_daemonize) {
detach();
}
/* write new pidfile (while still root, so can be outside chroot) */
#ifdef HAVE_KILL
if(cfg->pidfile && cfg->pidfile[0]) {
writepid(daemon->pidfile, getpid());
if(cfg->username && cfg->username[0] && cfg_uid != (uid_t)-1 &&
pidinchroot) {
# ifdef HAVE_CHOWN
if(chown(daemon->pidfile, cfg_uid, cfg_gid) == -1) {
verbose(VERB_QUERY, "cannot chown %u.%u %s: %s",
(unsigned)cfg_uid, (unsigned)cfg_gid,
daemon->pidfile, strerror(errno));
}
# endif /* HAVE_CHOWN */
}
}
#else
(void)daemon;
#endif /* HAVE_KILL */
/* Set user context */
#ifdef HAVE_GETPWNAM
if(cfg->username && cfg->username[0] && cfg_uid != (uid_t)-1) {
#ifdef HAVE_SETUSERCONTEXT
/* setusercontext does initgroups, setuid, setgid, and
* also resource limits from login config, but we
* still call setresuid, setresgid to be sure to set all uid*/
if(setusercontext(NULL, pwd, cfg_uid, (unsigned)
LOGIN_SETALL & ~LOGIN_SETUSER & ~LOGIN_SETGROUP) != 0)
log_warn("unable to setusercontext %s: %s",
cfg->username, strerror(errno));
#endif /* HAVE_SETUSERCONTEXT */
}
#endif /* HAVE_GETPWNAM */
/* box into the chroot */
#ifdef HAVE_CHROOT
if(cfg->chrootdir && cfg->chrootdir[0]) {
if(chdir(cfg->chrootdir)) {
fatal_exit("unable to chdir to chroot %s: %s",
cfg->chrootdir, strerror(errno));
}
verbose(VERB_QUERY, "chdir to %s", cfg->chrootdir);
if(chroot(cfg->chrootdir))
fatal_exit("unable to chroot to %s: %s",
cfg->chrootdir, strerror(errno));
if(chdir("/"))
fatal_exit("unable to chdir to / in chroot %s: %s",
cfg->chrootdir, strerror(errno));
verbose(VERB_QUERY, "chroot to %s", cfg->chrootdir);
if(strncmp(*cfgfile, cfg->chrootdir,
strlen(cfg->chrootdir)) == 0)
(*cfgfile) += strlen(cfg->chrootdir);
/* adjust stored pidfile for chroot */
if(daemon->pidfile && daemon->pidfile[0] &&
strncmp(daemon->pidfile, cfg->chrootdir,
strlen(cfg->chrootdir))==0) {
char* old = daemon->pidfile;
daemon->pidfile = strdup(old+strlen(cfg->chrootdir));
free(old);
if(!daemon->pidfile)
log_err("out of memory in pidfile adjust");
}
daemon->chroot = strdup(cfg->chrootdir);
if(!daemon->chroot)
log_err("out of memory in daemon chroot dir storage");
}
#else
(void)cfgfile;
#endif
/* change to working directory inside chroot */
if(cfg->directory && cfg->directory[0]) {
char* dir = cfg->directory;
if(cfg->chrootdir && cfg->chrootdir[0] &&
strncmp(dir, cfg->chrootdir,
strlen(cfg->chrootdir)) == 0)
dir += strlen(cfg->chrootdir);
if(dir[0]) {
if(chdir(dir)) {
fatal_exit("Could not chdir to %s: %s",
dir, strerror(errno));
}
verbose(VERB_QUERY, "chdir to %s", dir);
}
}
/* drop permissions after chroot, getpwnam, pidfile, syslog done*/
#ifdef HAVE_GETPWNAM
if(cfg->username && cfg->username[0] && cfg_uid != (uid_t)-1) {
# ifdef HAVE_INITGROUPS
if(initgroups(cfg->username, cfg_gid) != 0)
log_warn("unable to initgroups %s: %s",
cfg->username, strerror(errno));
# endif /* HAVE_INITGROUPS */
# ifdef HAVE_ENDPWENT
endpwent();
# endif
#ifdef HAVE_SETRESGID
if(setresgid(cfg_gid,cfg_gid,cfg_gid) != 0)
#elif defined(HAVE_SETREGID) && !defined(DARWIN_BROKEN_SETREUID)
if(setregid(cfg_gid,cfg_gid) != 0)
#else /* use setgid */
if(setgid(cfg_gid) != 0)
#endif /* HAVE_SETRESGID */
fatal_exit("unable to set group id of %s: %s",
cfg->username, strerror(errno));
#ifdef HAVE_SETRESUID
if(setresuid(cfg_uid,cfg_uid,cfg_uid) != 0)
#elif defined(HAVE_SETREUID) && !defined(DARWIN_BROKEN_SETREUID)
if(setreuid(cfg_uid,cfg_uid) != 0)
#else /* use setuid */
if(setuid(cfg_uid) != 0)
#endif /* HAVE_SETRESUID */
fatal_exit("unable to set user id of %s: %s",
cfg->username, strerror(errno));
verbose(VERB_QUERY, "drop user privileges, run as %s",
cfg->username);
}
#endif /* HAVE_GETPWNAM */
/* file logging inited after chroot,chdir,setuid is done so that
* it would succeed on SIGHUP as well */
if(!cfg->use_syslog)
log_init(cfg->logfile, cfg->use_syslog, cfg->chrootdir);
}
/**
* Run the daemon.
* @param cfgfile: the config file name.
* @param cmdline_verbose: verbosity resulting from commandline -v.
* These increase verbosity as specified in the config file.
* @param debug_mode: if set, do not daemonize.
* @param log_default_identity: Default identity to report in logs
*/
static void
run_daemon(const char* cfgfile, int cmdline_verbose, int debug_mode, const char* log_default_identity)
{
struct config_file* cfg = NULL;
struct daemon* daemon = NULL;
int done_setup = 0;
if(!(daemon = daemon_init()))
fatal_exit("alloc failure");
while(!daemon->need_to_exit) {
if(done_setup)
verbose(VERB_OPS, "Restart of %s.", PACKAGE_STRING);
else verbose(VERB_OPS, "Start of %s.", PACKAGE_STRING);
/* config stuff */
if(!(cfg = config_create()))
fatal_exit("Could not alloc config defaults");
if(!config_read(cfg, cfgfile, daemon->chroot)) {
if(errno != ENOENT)
fatal_exit("Could not read config file: %s",
cfgfile);
log_warn("Continuing with default config settings");
}
apply_settings(daemon, cfg, cmdline_verbose, debug_mode, log_default_identity);
if(!done_setup)
config_lookup_uid(cfg);
/* prepare */
if(!daemon_open_shared_ports(daemon))
fatal_exit("could not open ports");
if(!done_setup) {
perform_setup(daemon, cfg, debug_mode, &cfgfile);
done_setup = 1;
} else {
/* reopen log after HUP to facilitate log rotation */
if(!cfg->use_syslog)
log_init(cfg->logfile, 0, cfg->chrootdir);
}
/* work */
daemon_fork(daemon);
/* clean up for restart */
verbose(VERB_ALGO, "cleanup.");
daemon_cleanup(daemon);
config_delete(cfg);
}
verbose(VERB_ALGO, "Exit cleanup.");
/* this unlink may not work if the pidfile is located outside
* of the chroot/workdir or we no longer have permissions */
if(daemon->pidfile) {
int fd;
/* truncate pidfile */
fd = open(daemon->pidfile, O_WRONLY | O_TRUNC, 0644);
if(fd != -1)
close(fd);
/* delete pidfile */
unlink(daemon->pidfile);
}
daemon_delete(daemon);
}
/** getopt global, in case header files fail to declare it. */
extern int optind;
/** getopt global, in case header files fail to declare it. */
extern char* optarg;
/**
* main program. Set options given commandline arguments.
* @param argc: number of commandline arguments.
* @param argv: array of commandline arguments.
* @return: exit status of the program.
*/
int
main(int argc, char* argv[])
{
int c;
const char* cfgfile = CONFIGFILE;
const char* winopt = NULL;
const char* log_ident_default;
int cmdline_verbose = 0;
int debug_mode = 0;
#ifdef UB_ON_WINDOWS
int cmdline_cfg = 0;
#endif
log_init(NULL, 0, NULL);
log_ident_default = strrchr(argv[0],'/')?strrchr(argv[0],'/')+1:argv[0];
log_ident_set(log_ident_default);
/* parse the options */
while( (c=getopt(argc, argv, "c:dhvw:")) != -1) {
switch(c) {
case 'c':
cfgfile = optarg;
#ifdef UB_ON_WINDOWS
cmdline_cfg = 1;
#endif
break;
case 'v':
cmdline_verbose++;
verbosity++;
break;
case 'd':
debug_mode++;
break;
case 'w':
winopt = optarg;
break;
case '?':
case 'h':
default:
usage();
return 1;
}
}
argc -= optind;
argv += optind;
if(winopt) {
#ifdef UB_ON_WINDOWS
wsvc_command_option(winopt, cfgfile, cmdline_verbose,
cmdline_cfg);
#else
fatal_exit("option not supported");
#endif
}
if(argc != 0) {
usage();
return 1;
}
run_daemon(cfgfile, cmdline_verbose, debug_mode, log_ident_default);
log_init(NULL, 0, NULL); /* close logfile */
return 0;
}