NOMMU re-exec trick shuld not depend on existence of "don't daemonize"
[oweals/busybox.git] / sysklogd / syslogd.c
index 4052827967cb1d7d723bad4ce2a0372d11d2a6fc..437212372948361f7e5bb08e0a36d90c32c26451 100644 (file)
 #include <paths.h>
 #include <sys/un.h>
 
-/* SYSLOG_NAMES defined to pull some extra junk from syslog.h */
+/* SYSLOG_NAMES defined to pull some extra junk from syslog.h: */
+/* prioritynames[] and facilitynames[]. uclibc pulls those in _rwdata_! :( */
+
 #define SYSLOG_NAMES
 #include <sys/syslog.h>
 #include <sys/uio.h>
 
+#if ENABLE_FEATURE_REMOTE_LOG
+#include <netinet/in.h>
+#endif
+
+#if ENABLE_FEATURE_IPC_SYSLOG
+#include <sys/ipc.h>
+#include <sys/sem.h>
+#include <sys/shm.h>
+#endif
+
+
 #define DEBUG 0
 
-/* Path to the unix socket */
-static char *dev_log_name;
+/* MARK code is not very useful, is bloat, and broken:
+ * can deadlock if alarmed to make MARK while writing to IPC buffer
+ * (semaphores are down but do_mark routine tries to down them again) */
+#undef SYSLOGD_MARK
 
-/* Path for the file where all log messages are written */
-static const char *logFilePath = "/var/log/messages";
-static int logFD = -1;
+enum { MAX_READ = 256 };
 
-/* interval between marks in seconds */
-static int markInterval = 20 * 60;
+/* Semaphore operation structures */
+struct shbuf_ds {
+       int32_t size;   /* size of data written */
+       int32_t head;   /* start of message list */
+       int32_t tail;   /* end of message list */
+       char data[1];   /* data/messages */
+};
 
-/* level of messages to be locally logged */
-static int logLevel = 8;
+/* Allows us to have smaller initializer. Ugly. */
+#define GLOBALS \
+       const char *logFilePath;                \
+       int logFD;                              \
+       /* interval between marks in seconds */ \
+       /*int markInterval;*/                   \
+       /* level of messages to be logged */    \
+       int logLevel;                           \
+USE_FEATURE_ROTATE_LOGFILE( \
+       /* max size of file before rotation */  \
+       unsigned logFileSize;                   \
+       /* number of rotated message files */   \
+       unsigned logFileRotate;                 \
+       unsigned curFileSize;                   \
+       smallint isRegular;                     \
+) \
+USE_FEATURE_REMOTE_LOG( \
+       /* udp socket for remote logging */     \
+       int remoteFD;                           \
+       len_and_sockaddr* remoteAddr;           \
+) \
+USE_FEATURE_IPC_SYSLOG( \
+       int shmid; /* ipc shared memory id */   \
+       int s_semid; /* ipc semaphore id */     \
+       int shm_size;                           \
+       struct sembuf SMwup[1];                 \
+       struct sembuf SMwdn[3];                 \
+)
+
+struct init_globals {
+       GLOBALS
+};
 
-/* localhost's name */
-static char localHostName[64];
+struct globals {
+       GLOBALS
+#if ENABLE_FEATURE_IPC_SYSLOG
+       struct shbuf_ds *shbuf;
+#endif
+       time_t last_log_time;
+       /* localhost's name */
+       char localHostName[64];
+
+       /* We recv into recvbuf... */
+       char recvbuf[MAX_READ];
+       /* ...then copy to parsebuf, escaping control chars */
+       /* (can grow x2 max) */
+       char parsebuf[MAX_READ*2];
+       /* ...then sprintf into printbuf, adding timestamp (15 chars),
+        * host (64), fac.prio (20) to the message */
+       /* (growth by: 15 + 64 + 20 + delims = ~110) */
+       char printbuf[MAX_READ*2 + 128];
+};
 
+static const struct init_globals init_data = {
+       .logFilePath = "/var/log/messages",
+       .logFD = -1,
+#ifdef SYSLOGD_MARK
+       .markInterval = 20 * 60,
+#endif
+       .logLevel = 8,
 #if ENABLE_FEATURE_ROTATE_LOGFILE
-/* max size of message file before being rotated */
-static unsigned logFileSize = 200 * 1024;
-/* number of rotated message files */
-static unsigned logFileRotate = 1;
-static unsigned curFileSize;
-static smallint isRegular;
+       .logFileSize = 200 * 1024,
+       .logFileRotate = 1,
 #endif
-
 #if ENABLE_FEATURE_REMOTE_LOG
-#include <netinet/in.h>
-/* udp socket for logging to remote host */
-static int remoteFD = -1;
-static len_and_sockaddr* remoteAddr;
+       .remoteFD = -1,
 #endif
+#if ENABLE_FEATURE_IPC_SYSLOG
+       .shmid = -1,
+       .s_semid = -1,
+       .shm_size = ((CONFIG_FEATURE_IPC_SYSLOG_BUFFER_SIZE)*1024), // default shm size
+       .SMwup = { {1, -1, IPC_NOWAIT} },
+       .SMwdn = { {0, 0}, {1, 0}, {1, +1} },
+#endif
+};
 
-/* We are using bb_common_bufsiz1 for buffering: */
-enum { MAX_READ = (BUFSIZ/6) & ~0xf };
-/* We recv into RECVBUF... (size: MAX_READ ~== BUFSIZ/6) */
-#define RECVBUF  bb_common_bufsiz1
-/* ...then copy to PARSEBUF, escaping control chars */
-/* (can grow x2 max ~== BUFSIZ/3) */
-#define PARSEBUF (bb_common_bufsiz1 + MAX_READ)
-/* ...then sprintf into PRINTBUF, adding timestamp (15 chars),
- * host (64), fac.prio (20) to the message */
-/* (growth by: 15 + 64 + 20 + delims = ~110) */
-#define PRINTBUF (bb_common_bufsiz1 + 3*MAX_READ)
-/* totals: BUFSIZ == BUFSIZ/6 + BUFSIZ/3 + (BUFSIZ/3+BUFSIZ/6)
- * -- we have BUFSIZ/6 extra at the ent of PRINTBUF
- * which covers needed ~110 extra bytes (and much more) */
+#define G (*ptr_to_globals)
 
 
 /* Options */
@@ -107,7 +166,7 @@ enum {
        USE_FEATURE_ROTATE_LOGFILE(,*opt_b) \
        USE_FEATURE_REMOTE_LOG(    ,*opt_R) \
        USE_FEATURE_IPC_SYSLOG(    ,*opt_C = NULL)
-#define OPTION_PARAM &opt_m, &logFilePath, &opt_l \
+#define OPTION_PARAM &opt_m, &G.logFilePath, &opt_l \
        USE_FEATURE_ROTATE_LOGFILE(,&opt_s) \
        USE_FEATURE_ROTATE_LOGFILE(,&opt_b) \
        USE_FEATURE_REMOTE_LOG(    ,&opt_R) \
@@ -117,68 +176,51 @@ enum {
 /* circular buffer variables/structures */
 #if ENABLE_FEATURE_IPC_SYSLOG
 
-
 #if CONFIG_FEATURE_IPC_SYSLOG_BUFFER_SIZE < 4
 #error Sorry, you must set the syslogd buffer size to at least 4KB.
 #error Please check CONFIG_FEATURE_IPC_SYSLOG_BUFFER_SIZE
 #endif
 
-#include <sys/ipc.h>
-#include <sys/sem.h>
-#include <sys/shm.h>
-
 /* our shared key */
 #define KEY_ID ((long)0x414e4547) /* "GENA" */
 
-// Semaphore operation structures
-static struct shbuf_ds {
-       int32_t size;   // size of data written
-       int32_t head;   // start of message list
-       int32_t tail;   // end of message list
-       char data[1];   // data/messages
-} *shbuf;               // shared memory pointer
-
-static int shmid = -1;   // ipc shared memory id
-static int s_semid = -1; // ipc semaphore id
-static int shm_size = ((CONFIG_FEATURE_IPC_SYSLOG_BUFFER_SIZE)*1024);  // default shm size
-
 static void ipcsyslog_cleanup(void)
 {
-       if (shmid != -1) {
-               shmdt(shbuf);
+       if (G.shmid != -1) {
+               shmdt(G.shbuf);
        }
-       if (shmid != -1) {
-               shmctl(shmid, IPC_RMID, NULL);
+       if (G.shmid != -1) {
+               shmctl(G.shmid, IPC_RMID, NULL);
        }
-       if (s_semid != -1) {
-               semctl(s_semid, 0, IPC_RMID, 0);
+       if (G.s_semid != -1) {
+               semctl(G.s_semid, 0, IPC_RMID, 0);
        }
 }
 
 static void ipcsyslog_init(void)
 {
        if (DEBUG)
-               printf("shmget(%lx, %d,...)\n", KEY_ID, shm_size);
+               printf("shmget(%lx, %d,...)\n", KEY_ID, G.shm_size);
 
-       shmid = shmget(KEY_ID, shm_size, IPC_CREAT | 1023);
-       if (shmid == -1) {
+       G.shmid = shmget(KEY_ID, G.shm_size, IPC_CREAT | 1023);
+       if (G.shmid == -1) {
                bb_perror_msg_and_die("shmget");
        }
 
-       shbuf = shmat(shmid, NULL, 0);
-       if (!shbuf) {
+       G.shbuf = shmat(G.shmid, NULL, 0);
+       if (!G.shbuf) {
                bb_perror_msg_and_die("shmat");
        }
 
-       shbuf->size = shm_size - offsetof(struct shbuf_ds, data);
-       shbuf->head = shbuf->tail = 0;
+       G.shbuf->size = G.shm_size - offsetof(struct shbuf_ds, data);
+       G.shbuf->head = G.shbuf->tail = 0;
 
        // we'll trust the OS to set initial semval to 0 (let's hope)
-       s_semid = semget(KEY_ID, 2, IPC_CREAT | IPC_EXCL | 1023);
-       if (s_semid == -1) {
+       G.s_semid = semget(KEY_ID, 2, IPC_CREAT | IPC_EXCL | 1023);
+       if (G.s_semid == -1) {
                if (errno == EEXIST) {
-                       s_semid = semget(KEY_ID, 2, 0);
-                       if (s_semid != -1)
+                       G.s_semid = semget(KEY_ID, 2, 0);
+                       if (G.s_semid != -1)
                                return;
                }
                bb_perror_msg_and_die("semget");
@@ -188,20 +230,16 @@ static void ipcsyslog_init(void)
 /* Write message to shared mem buffer */
 static void log_to_shmem(const char *msg, int len)
 {
-       /* Why libc insists on these being rw? */
-       static struct sembuf SMwup[1] = { {1, -1, IPC_NOWAIT} };
-       static struct sembuf SMwdn[3] = { {0, 0}, {1, 0}, {1, +1} };
-
        int old_tail, new_tail;
        char *c;
 
-       if (semop(s_semid, SMwdn, 3) == -1) {
+       if (semop(G.s_semid, G.SMwdn, 3) == -1) {
                bb_perror_msg_and_die("SMwdn");
        }
 
        /* Circular Buffer Algorithm:
         * --------------------------
-         * tail == position where to store next syslog message.
+        * tail == position where to store next syslog message.
         * head == position of next message to retrieve ("print").
         * if head == tail, there is no "unprinted" messages left.
         * head is typically advanced by separate "reader" program,
@@ -210,53 +248,53 @@ static void log_to_shmem(const char *msg, int len)
         */
        len++; /* length with NUL included */
  again:
-       old_tail = shbuf->tail;
+       old_tail = G.shbuf->tail;
        new_tail = old_tail + len;
-       if (new_tail < shbuf->size) {
+       if (new_tail < G.shbuf->size) {
                /* No need to move head if shbuf->head <= old_tail,
                 * else... */
-               if (old_tail < shbuf->head && shbuf->head <= new_tail) {
+               if (old_tail < G.shbuf->head && G.shbuf->head <= new_tail) {
                        /* ...need to move head forward */
-                       c = memchr(shbuf->data + new_tail, '\0',
-                                          shbuf->size - new_tail);
+                       c = memchr(G.shbuf->data + new_tail, '\0',
+                                          G.shbuf->size - new_tail);
                        if (!c) /* no NUL ahead of us, wrap around */
-                               c = memchr(shbuf->data, '\0', old_tail);
+                               c = memchr(G.shbuf->data, '\0', old_tail);
                        if (!c) { /* still nothing? point to this msg... */
-                               shbuf->head = old_tail;
+                               G.shbuf->head = old_tail;
                        } else {
                                /* convert pointer to offset + skip NUL */
-                               shbuf->head = c - shbuf->data + 1;
+                               G.shbuf->head = c - G.shbuf->data + 1;
                        }
                }
                /* store message, set new tail */
-               memcpy(shbuf->data + old_tail, msg, len);
-               shbuf->tail = new_tail;
+               memcpy(G.shbuf->data + old_tail, msg, len);
+               G.shbuf->tail = new_tail;
        } else {
                /* we need to break up the message and wrap it around */
                /* k == available buffer space ahead of old tail */
-               int k = shbuf->size - old_tail - 1;
-               if (shbuf->head > old_tail) {
+               int k = G.shbuf->size - old_tail - 1;
+               if (G.shbuf->head > old_tail) {
                        /* we are going to overwrite head, need to
                         * move it out of the way */
-                       c = memchr(shbuf->data, '\0', old_tail);
+                       c = memchr(G.shbuf->data, '\0', old_tail);
                        if (!c) { /* nothing? point to this msg... */
-                               shbuf->head = old_tail;
+                               G.shbuf->head = old_tail;
                        } else { /* convert pointer to offset + skip NUL */
-                               shbuf->head = c - shbuf->data + 1;
+                               G.shbuf->head = c - G.shbuf->data + 1;
                        }
                }
                /* copy what fits to the end of buffer, and repeat */
-               memcpy(shbuf->data + old_tail, msg, k);
+               memcpy(G.shbuf->data + old_tail, msg, k);
                msg += k;
                len -= k;
-               shbuf->tail = 0;
+               G.shbuf->tail = 0;
                goto again;
        }
-       if (semop(s_semid, SMwup, 1) == -1) {
+       if (semop(G.s_semid, G.SMwup, 1) == -1) {
                bb_perror_msg_and_die("SMwup");
        }
        if (DEBUG)
-               printf("head:%d tail:%d\n", shbuf->head, shbuf->tail);
+               printf("head:%d tail:%d\n", G.shbuf->head, G.shbuf->tail);
 }
 #else
 void ipcsyslog_cleanup(void);
@@ -268,29 +306,28 @@ void log_to_shmem(const char *msg);
 /* Print a message to the log file. */
 static void log_locally(char *msg)
 {
-       static time_t last;
        struct flock fl;
        int len = strlen(msg);
 
 #if ENABLE_FEATURE_IPC_SYSLOG
-       if ((option_mask32 & OPT_circularlog) && shbuf) {
+       if ((option_mask32 & OPT_circularlog) && G.shbuf) {
                log_to_shmem(msg, len);
                return;
        }
 #endif
-       if (logFD >= 0) {
+       if (G.logFD >= 0) {
                time_t cur;
                time(&cur);
-               if (last != cur) {
-                       last = cur; /* reopen log file every second */
-                       close(logFD);
+               if (G.last_log_time != cur) {
+                       G.last_log_time = cur; /* reopen log file every second */
+                       close(G.logFD);
                        goto reopen;
                }
        } else {
  reopen:
-               logFD = device_open(logFilePath, O_WRONLY | O_CREAT
+               G.logFD = device_open(G.logFilePath, O_WRONLY | O_CREAT
                                        | O_NOCTTY | O_APPEND | O_NONBLOCK);
-               if (logFD < 0) {
+               if (G.logFD < 0) {
                        /* cannot open logfile? - print to /dev/console then */
                        int fd = device_open(_PATH_CONSOLE, O_WRONLY | O_NOCTTY | O_NONBLOCK);
                        if (fd < 0)
@@ -304,9 +341,9 @@ static void log_locally(char *msg)
                {
                struct stat statf;
 
-               isRegular = (fstat(logFD, &statf) == 0 && (statf.st_mode & S_IFREG));
+               G.isRegular = (fstat(G.logFD, &statf) == 0 && (statf.st_mode & S_IFREG));
                /* bug (mostly harmless): can wrap around if file > 4gb */
-               curFileSize = statf.st_size;
+               G.curFileSize = statf.st_size;
                }
 #endif
        }
@@ -315,36 +352,36 @@ static void log_locally(char *msg)
        fl.l_start = 0;
        fl.l_len = 1;
        fl.l_type = F_WRLCK;
-       fcntl(logFD, F_SETLKW, &fl);
+       fcntl(G.logFD, F_SETLKW, &fl);
 
 #if ENABLE_FEATURE_ROTATE_LOGFILE
-       if (logFileSize && isRegular && curFileSize > logFileSize) {
-               if (logFileRotate) { /* always 0..99 */
-                       int i = strlen(logFilePath) + 3 + 1;
+       if (G.logFileSize && G.isRegular && G.curFileSize > G.logFileSize) {
+               if (G.logFileRotate) { /* always 0..99 */
+                       int i = strlen(G.logFilePath) + 3 + 1;
                        char oldFile[i];
                        char newFile[i];
-                       i = logFileRotate - 1;
+                       i = G.logFileRotate - 1;
                        /* rename: f.8 -> f.9; f.7 -> f.8; ... */
                        while (1) {
-                               sprintf(newFile, "%s.%d", logFilePath, i);
+                               sprintf(newFile, "%s.%d", G.logFilePath, i);
                                if (i == 0) break;
-                               sprintf(oldFile, "%s.%d", logFilePath, --i);
+                               sprintf(oldFile, "%s.%d", G.logFilePath, --i);
                                rename(oldFile, newFile);
                        }
                        /* newFile == "f.0" now */
-                       rename(logFilePath, newFile);
+                       rename(G.logFilePath, newFile);
                        fl.l_type = F_UNLCK;
-                       fcntl(logFD, F_SETLKW, &fl);
-                       close(logFD);
+                       fcntl(G.logFD, F_SETLKW, &fl);
+                       close(G.logFD);
                        goto reopen;
                }
-               ftruncate(logFD, 0);
+               ftruncate(G.logFD, 0);
        }
-       curFileSize +=
+       G.curFileSize +=
 #endif
-                       full_write(logFD, msg, len);
+                       full_write(G.logFD, msg, len);
        fl.l_type = F_UNLCK;
-       fcntl(logFD, F_SETLKW, &fl);
+       fcntl(G.logFD, F_SETLKW, &fl);
 }
 
 static void parse_fac_prio_20(int pri, char *res20)
@@ -395,15 +432,15 @@ static void timestamp_and_log(int pri, char *msg, int len)
 
        /* Log message locally (to file or shared mem) */
        if (!ENABLE_FEATURE_REMOTE_LOG || (option_mask32 & OPT_locallog)) {
-               if (LOG_PRI(pri) < logLevel) {
+               if (LOG_PRI(pri) < G.logLevel) {
                        if (option_mask32 & OPT_small)
-                               sprintf(PRINTBUF, "%s %s\n", timestamp, msg);
+                               sprintf(G.printbuf, "%s %s\n", timestamp, msg);
                        else {
                                char res[20];
                                parse_fac_prio_20(pri, res);
-                               sprintf(PRINTBUF, "%s %s %s %s\n", timestamp, localHostName, res, msg);
+                               sprintf(G.printbuf, "%s %s %s %s\n", timestamp, G.localHostName, res, msg);
                        }
-                       log_locally(PRINTBUF);
+                       log_locally(G.printbuf);
                }
        }
 }
@@ -415,7 +452,7 @@ static void split_escape_and_log(char *tmpbuf, int len)
        tmpbuf += len;
        while (p < tmpbuf) {
                char c;
-               char *q = PARSEBUF;
+               char *q = G.parsebuf;
                int pri = (LOG_USER | LOG_NOTICE);
 
                if (*p == '<') {
@@ -438,35 +475,36 @@ static void split_escape_and_log(char *tmpbuf, int len)
                }
                *q = '\0';
                /* Now log it */
-               timestamp_and_log(pri, PARSEBUF, q - PARSEBUF);
+               timestamp_and_log(pri, G.parsebuf, q - G.parsebuf);
        }
 }
 
 static void quit_signal(int sig)
 {
-       timestamp_and_log(LOG_SYSLOG | LOG_INFO, "syslogd exiting", 0);
+       timestamp_and_log(LOG_SYSLOG | LOG_INFO, (char*)"syslogd exiting", 0);
        puts("syslogd exiting");
-       unlink(dev_log_name);
        if (ENABLE_FEATURE_IPC_SYSLOG)
                ipcsyslog_cleanup();
        exit(1);
 }
 
+#ifdef SYSLOGD_MARK
 static void do_mark(int sig)
 {
-       if (markInterval) {
-               timestamp_and_log(LOG_SYSLOG | LOG_INFO, "-- MARK --", 0);
-               alarm(markInterval);
+       if (G.markInterval) {
+               timestamp_and_log(LOG_SYSLOG | LOG_INFO, (char*)"-- MARK --", 0);
+               alarm(G.markInterval);
        }
 }
+#endif
 
 static void do_syslogd(void) ATTRIBUTE_NORETURN;
 static void do_syslogd(void)
 {
        struct sockaddr_un sunx;
-       socklen_t addr_len;
        int sock_fd;
        fd_set fds;
+       char *dev_log_name;
 
        /* Set up signal handlers */
        signal(SIGINT, quit_signal);
@@ -477,31 +515,45 @@ static void do_syslogd(void)
 #ifdef SIGCLD
        signal(SIGCLD, SIG_IGN);
 #endif
+#ifdef SYSLOGD_MARK
        signal(SIGALRM, do_mark);
-       alarm(markInterval);
-
-       dev_log_name = xmalloc_realpath(_PATH_LOG);
-       if (!dev_log_name)
-               dev_log_name = _PATH_LOG;
-
-       /* Unlink old /dev/log (or object it points to) */
-       unlink(dev_log_name);
+       alarm(G.markInterval);
+#endif
 
        memset(&sunx, 0, sizeof(sunx));
        sunx.sun_family = AF_UNIX;
-       strncpy(sunx.sun_path, dev_log_name, sizeof(sunx.sun_path));
+       strcpy(sunx.sun_path, "/dev/log");
+
+       /* Unlink old /dev/log or object it points to. */
+       /* (if it exists, bind will fail) */
+       logmode = LOGMODE_NONE;
+       dev_log_name = xmalloc_readlink_or_warn("/dev/log");
+       logmode = LOGMODE_STDIO;
+       if (dev_log_name) {
+               int fd = xopen(".", O_NONBLOCK);
+               xchdir("/dev");
+               /* we do not check whether this is a link also */
+               unlink(dev_log_name);
+               fchdir(fd);
+               close(fd);
+               safe_strncpy(sunx.sun_path, dev_log_name, sizeof(sunx.sun_path));
+               free(dev_log_name);
+       } else {
+               unlink("/dev/log");
+       }
+
        sock_fd = xsocket(AF_UNIX, SOCK_DGRAM, 0);
-       addr_len = sizeof(sunx.sun_family) + strlen(sunx.sun_path);
-       xbind(sock_fd, (struct sockaddr *) &sunx, addr_len);
+       xbind(sock_fd, (struct sockaddr *) &sunx, sizeof(sunx));
 
-       if (chmod(dev_log_name, 0666) < 0) {
-               bb_perror_msg_and_die("cannot set permission on %s", dev_log_name);
+       if (chmod("/dev/log", 0666) < 0) {
+               bb_perror_msg_and_die("cannot set permission on /dev/log");
        }
        if (ENABLE_FEATURE_IPC_SYSLOG && (option_mask32 & OPT_circularlog)) {
                ipcsyslog_init();
        }
 
-       timestamp_and_log(LOG_SYSLOG | LOG_INFO, "syslogd started: BusyBox v" BB_VER, 0);
+       timestamp_and_log(LOG_SYSLOG | LOG_INFO,
+                       (char*)"syslogd started: BusyBox v" BB_VER, 0);
 
        for (;;) {
                FD_ZERO(&fds);
@@ -517,60 +569,65 @@ static void do_syslogd(void)
 
                if (FD_ISSET(sock_fd, &fds)) {
                        int i;
-                       i = recv(sock_fd, RECVBUF, MAX_READ - 1, 0);
+                       i = recv(sock_fd, G.recvbuf, MAX_READ - 1, 0);
                        if (i <= 0)
                                bb_perror_msg_and_die("UNIX socket error");
                        /* TODO: maybe suppress duplicates? */
 #if ENABLE_FEATURE_REMOTE_LOG
                        /* We are not modifying log messages in any way before send */
                        /* Remote site cannot trust _us_ anyway and need to do validation again */
-                       if (remoteAddr) {
-                               if (-1 == remoteFD) {
-                                       remoteFD = socket(remoteAddr->sa.sa_family, SOCK_DGRAM, 0);
+                       if (G.remoteAddr) {
+                               if (-1 == G.remoteFD) {
+                                       G.remoteFD = socket(G.remoteAddr->sa.sa_family, SOCK_DGRAM, 0);
                                }
-                               if (-1 != remoteFD) {
+                               if (-1 != G.remoteFD) {
                                        /* send message to remote logger, ignore possible error */
-                                       sendto(remoteFD, RECVBUF, i, MSG_DONTWAIT,
-                                               &remoteAddr->sa, remoteAddr->len);
+                                       sendto(G.remoteFD, G.recvbuf, i, MSG_DONTWAIT,
+                                               &G.remoteAddr->sa, G.remoteAddr->len);
                                }
                        }
 #endif
-                       RECVBUF[i] = '\0';
-                       split_escape_and_log(RECVBUF, i);
+                       G.recvbuf[i] = '\0';
+                       split_escape_and_log(G.recvbuf, i);
                } /* FD_ISSET() */
        } /* for */
 }
 
+int syslogd_main(int argc, char **argv);
 int syslogd_main(int argc, char **argv)
 {
        char OPTION_DECL;
        char *p;
 
+       PTR_TO_GLOBALS = memcpy(xzalloc(sizeof(G)), &init_data, sizeof(init_data));
+
        /* do normal option parsing */
        opt_complementary = "=0"; /* no non-option params */
        getopt32(argc, argv, OPTION_STR, OPTION_PARAM);
+#ifdef SYSLOGD_MARK
        if (option_mask32 & OPT_mark) // -m
-               markInterval = xatou_range(opt_m, 0, INT_MAX/60) * 60;
+               G.markInterval = xatou_range(opt_m, 0, INT_MAX/60) * 60;
+#endif
        //if (option_mask32 & OPT_nofork) // -n
        //if (option_mask32 & OPT_outfile) // -O
        if (option_mask32 & OPT_loglevel) // -l
-               logLevel = xatou_range(opt_l, 1, 8);
+               G.logLevel = xatou_range(opt_l, 1, 8);
        //if (option_mask32 & OPT_small) // -S
 #if ENABLE_FEATURE_ROTATE_LOGFILE
        if (option_mask32 & OPT_filesize) // -s
-               logFileSize = xatou_range(opt_s, 0, INT_MAX/1024) * 1024;
+               G.logFileSize = xatou_range(opt_s, 0, INT_MAX/1024) * 1024;
        if (option_mask32 & OPT_rotatecnt) // -b
-               logFileRotate = xatou_range(opt_b, 0, 99);
+               G.logFileRotate = xatou_range(opt_b, 0, 99);
 #endif
 #if ENABLE_FEATURE_REMOTE_LOG
        if (option_mask32 & OPT_remotelog) { // -R
-               remoteAddr = host2sockaddr(opt_R, 514);
+               G.remoteAddr = xhost2sockaddr(opt_R, 514);
        }
        //if (option_mask32 & OPT_locallog) // -L
 #endif
 #if ENABLE_FEATURE_IPC_SYSLOG
-       if ((option_mask32 & OPT_circularlog) && opt_C) // -C
-               shm_size = xatoul_range(opt_C, 4, INT_MAX/1024) * 1024;
+       if (opt_C) // -Cn
+               G.shm_size = xatoul_range(opt_C, 4, INT_MAX/1024) * 1024;
 #endif
 
        /* If they have not specified remote logging, then log locally */
@@ -578,17 +635,18 @@ int syslogd_main(int argc, char **argv)
                option_mask32 |= OPT_locallog;
 
        /* Store away localhost's name before the fork */
-       gethostname(localHostName, sizeof(localHostName));
-       p = strchr(localHostName, '.');
+       gethostname(G.localHostName, sizeof(G.localHostName));
+       p = strchr(G.localHostName, '.');
        if (p) {
                *p = '\0';
        }
 
        if (!(option_mask32 & OPT_nofork)) {
 #ifdef BB_NOMMU
-               vfork_daemon_rexec(0, 1, argc, argv, "-n");
+               if (!re_execed)
+                       vfork_daemon_rexec(0, 1, argv);
 #else
-               xdaemon(0, 1);
+               bb_daemonize();
 #endif
        }
        umask(0);