Closing bug 730. libbb run_parts is using scandir (a GNUism),
[oweals/busybox.git] / sysklogd / syslogd.c
index 03fea9c761252e1aef94305146ce48244158407e..9e030bd63dd957a66acdbae4ef57e1bf6d1aa51a 100644 (file)
-/* userver.c - simple server for Unix domain sockets */
+/* vi: set sw=4 ts=4: */
+/*
+ * Mini syslogd implementation for busybox
+ *
+ * Copyright (C) 1999-2004 by Erik Andersen <andersen@codepoet.org>
+ *
+ * Copyright (C) 2000 by Karl M. Hegbloom <karlheg@debian.org>
+ *
+ * "circular buffer" Copyright (C) 2001 by Gennady Feldman <gfeldman@gena01.com>
+ *
+ * Maintainer: Gennady Feldman <gfeldman@gena01.com> as of Mar 12, 2001
+ *
+ * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
+ */
 
-/* Waits for a connection on the ./sample-socket Unix domain
-   socket. Once a connection has been established, copy data
-   from the socket to stdout until the other end closes the
-   connection, and then wait for another connection to the socket. */
-
-#include <stdio.h>
-#include <sys/socket.h>
+#include "busybox.h"
+#include <paths.h>
+#include <stdbool.h>
 #include <sys/un.h>
-#include <unistd.h>
 
-#define _PATH_LOG       "/dev/log"
+/* SYSLOG_NAMES defined to pull some extra junk from syslog.h */
+#define SYSLOG_NAMES
+#include <sys/syslog.h>
+#include <sys/uio.h>
+
+/* Path to the unix socket */
+static char lfile[MAXPATHLEN];
+
+/* Path for the file where all log messages are written */
+static const char *logFilePath = "/var/log/messages";
+
+#ifdef CONFIG_FEATURE_ROTATE_LOGFILE
+/* max size of message file before being rotated */
+static int logFileSize = 200 * 1024;
+
+/* number of rotated message files */
+static int logFileRotate = 1;
+#endif
+
+/* interval between marks in seconds */
+static int MarkInterval = 20 * 60;
+
+/* level of messages to be locally logged */
+static int logLevel = 8;
+
+/* localhost's name */
+static char LocalHostName[64];
+
+#ifdef CONFIG_FEATURE_REMOTE_LOG
+#include <netinet/in.h>
+/* udp socket for logging to remote host */
+static int remotefd = -1;
+static struct sockaddr_in remoteaddr;
+
+#endif
+
+/* options */
+/* Correct regardless of combination of CONFIG_xxx */
+enum {
+       OPTBIT_mark = 0, // -m
+       OPTBIT_nofork, // -n
+       OPTBIT_outfile, // -O
+       OPTBIT_loglevel, // -l
+       OPTBIT_small, // -S
+       USE_FEATURE_ROTATE_LOGFILE(OPTBIT_filesize   ,) // -s
+       USE_FEATURE_ROTATE_LOGFILE(OPTBIT_rotatecnt  ,) // -b
+       USE_FEATURE_REMOTE_LOG(    OPTBIT_remote     ,) // -R
+       USE_FEATURE_REMOTE_LOG(    OPTBIT_localtoo   ,) // -L
+       USE_FEATURE_IPC_SYSLOG(    OPTBIT_circularlog,) // -C
+
+       OPT_mark        = 1 << OPTBIT_mark    ,
+       OPT_nofork      = 1 << OPTBIT_nofork  ,
+       OPT_outfile     = 1 << OPTBIT_outfile ,
+       OPT_loglevel    = 1 << OPTBIT_loglevel,
+       OPT_small       = 1 << OPTBIT_small   ,
+       OPT_filesize    = USE_FEATURE_ROTATE_LOGFILE((1 << OPTBIT_filesize   )) + 0,
+       OPT_rotatecnt   = USE_FEATURE_ROTATE_LOGFILE((1 << OPTBIT_rotatecnt  )) + 0,
+       OPT_remotelog   = USE_FEATURE_REMOTE_LOG(    (1 << OPTBIT_remote     )) + 0,
+       OPT_locallog    = USE_FEATURE_REMOTE_LOG(    (1 << OPTBIT_localtoo   )) + 0,
+       OPT_circularlog = USE_FEATURE_IPC_SYSLOG(    (1 << OPTBIT_circularlog)) + 0,
+};
+#define OPTION_STR "m:nO:l:S" \
+       USE_FEATURE_ROTATE_LOGFILE("s:" ) \
+       USE_FEATURE_ROTATE_LOGFILE("b:" ) \
+       USE_FEATURE_REMOTE_LOG(    "R:" ) \
+       USE_FEATURE_REMOTE_LOG(    "L"  ) \
+       USE_FEATURE_IPC_SYSLOG(    "C::")
+#define OPTION_DECL *opt_m, *opt_l \
+       USE_FEATURE_ROTATE_LOGFILE(,*opt_s) \
+       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 \
+       USE_FEATURE_ROTATE_LOGFILE(,&opt_s) \
+       USE_FEATURE_ROTATE_LOGFILE(,&opt_b) \
+       USE_FEATURE_REMOTE_LOG(    ,&opt_R) \
+       USE_FEATURE_IPC_SYSLOG(    ,&opt_C)
+
+#define MAXLINE         1024   /* maximum line length */
+
+/* circular buffer variables/structures */
+#ifdef CONFIG_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 {
+       int size;       // size of data written
+       int head;       // start of message list
+       int tail;       // end of message list
+       char data[1];   // data/messages
+} *shbuf = NULL;        // shared memory pointer
+
+static struct sembuf SMwup[1] = { {1, -1, IPC_NOWAIT} };       // set SMwup
+static struct sembuf SMwdn[3] = { {0, 0}, {1, 0}, {1, +1} };   // set SMwdn
+
+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)
+{
+       puts("Exiting syslogd!");
+       if (shmid != -1) {
+               shmdt(shbuf);
+       }
+
+       if (shmid != -1) {
+               shmctl(shmid, IPC_RMID, NULL);
+       }
+       if (s_semid != -1) {
+               semctl(s_semid, 0, IPC_RMID, 0);
+       }
+}
+
+static void ipcsyslog_init(void)
+{
+       if (shbuf == NULL) {
+               shmid = shmget(KEY_ID, shm_size, IPC_CREAT | 1023);
+               if (shmid == -1) {
+                       bb_perror_msg_and_die("shmget");
+               }
+
+               shbuf = shmat(shmid, NULL, 0);
+               if (!shbuf) {
+                       bb_perror_msg_and_die("shmat");
+               }
+
+               shbuf->size = shm_size - sizeof(*shbuf);
+               shbuf->head = 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) {
+                       if (errno == EEXIST) {
+                               s_semid = semget(KEY_ID, 2, 0);
+                               if (s_semid == -1) {
+                                       bb_perror_msg_and_die("semget");
+                               }
+                       } else {
+                               bb_perror_msg_and_die("semget");
+                       }
+               }
+       } else {
+               printf("Buffer already allocated just grab the semaphore?");
+       }
+}
+
+/* write message to buffer */
+static void circ_message(const char *msg)
+{
+       int l = strlen(msg) + 1;        /* count the whole message w/ '\0' included */
+       const char * const fail_msg = "Can't find the terminator token%s?\n";
+
+       if (semop(s_semid, SMwdn, 3) == -1) {
+               bb_perror_msg_and_die("SMwdn");
+       }
+
+       /*
+        * Circular Buffer Algorithm:
+        * --------------------------
+        *
+        * Start-off w/ empty buffer of specific size SHM_SIZ
+        * Start filling it up w/ messages. I use '\0' as separator to break up messages.
+        * This is also very handy since we can do printf on message.
+        *
+        * Once the buffer is full we need to get rid of the first message in buffer and
+        * insert the new message. (Note: if the message being added is >1 message then
+        * we will need to "remove" >1 old message from the buffer). The way this is done
+        * is the following:
+        *      When we reach the end of the buffer we set a mark and start from the beginning.
+        *      Now what about the beginning and end of the buffer? Well we have the "head"
+        *      index/pointer which is the starting point for the messages and we have "tail"
+        *      index/pointer which is the ending point for the messages. When we "display" the
+        *      messages we start from the beginning and continue until we reach "tail". If we
+        *      reach end of buffer, then we just start from the beginning (offset 0). "head" and
+        *      "tail" are actually offsets from the beginning of the buffer.
+        *
+        * Note: This algorithm uses Linux IPC mechanism w/ shared memory and semaphores to provide
+        *       a threadsafe way of handling shared memory operations.
+        */
+       if ((shbuf->tail + l) < shbuf->size) {
+               /* before we append the message we need to check the HEAD so that we won't
+                  overwrite any of the message that we still need and adjust HEAD to point
+                  to the next message! */
+               if (shbuf->tail < shbuf->head) {
+                       if ((shbuf->tail + l) >= shbuf->head) {
+                               /* we need to move the HEAD to point to the next message
+                                * Theoretically we have enough room to add the whole message to the
+                                * buffer, because of the first outer IF statement, so we don't have
+                                * to worry about overflows here!
+                                */
+                               /* we need to know how many bytes we are overwriting to make enough room */
+                               int k = shbuf->tail + l - shbuf->head;
+                               char *c =
+                                       memchr(shbuf->data + shbuf->head + k, '\0',
+                                                  shbuf->size - (shbuf->head + k));
+                               if (c != NULL) { /* do a sanity check just in case! */
+                                       /* we need to convert pointer to offset + skip the '\0'
+                                          since we need to point to the beginning of the next message */
+                                       shbuf->head = c - shbuf->data + 1;
+                                       /* Note: HEAD is only used to "retrieve" messages, it's not used
+                                          when writing messages into our buffer */
+                               } else { /* show an error message to know we messed up? */
+                                       printf(fail_msg,"");
+                                       shbuf->head = 0;
+                               }
+                       }
+               }
+
+               /* in other cases no overflows have been done yet, so we don't care! */
+               /* we should be ok to append the message now */
+               strncpy(shbuf->data + shbuf->tail, msg, l);     /* append our message */
+               shbuf->tail += l;       /* count full message w/ '\0' terminating char */
+       } else {
+               /* we need to break up the message and "circle" it around */
+               char *c;
+               int k = shbuf->tail + l - shbuf->size;  /* count # of bytes we don't fit */
 
+               /* We need to move HEAD! This is always the case since we are going
+                * to "circle" the message. */
+               c = memchr(shbuf->data + k, '\0', shbuf->size - k);
 
-/* issue an error message via perror() and terminate the program */
-void die(char * message) {
-    perror(message);
-    exit(1);
+               if (c != NULL) {        /* if we don't have '\0'??? weird!!! */
+                       /* move head pointer */
+                       shbuf->head = c - shbuf->data + 1;
+
+                       /* now write the first part of the message */
+                       strncpy(shbuf->data + shbuf->tail, msg, l - k - 1);
+
+                       /* ALWAYS terminate end of buffer w/ '\0' */
+                       shbuf->data[shbuf->size - 1] = '\0';
+
+                       /* now write out the rest of the string to the beginning of the buffer */
+                       strcpy(shbuf->data, &msg[l - k - 1]);
+
+                       /* we need to place the TAIL at the end of the message */
+                       shbuf->tail = k + 1;
+               } else {
+                       printf(fail_msg, " from the beginning");
+                       shbuf->head = shbuf->tail = 0;  /* reset buffer, since it's probably corrupted */
+               }
+
+       }
+       if (semop(s_semid, SMwup, 1) == -1) {
+               bb_perror_msg_and_die("SMwup");
+       }
+
+}
+#else
+void ipcsyslog_cleanup(void);
+void ipcsyslog_init(void);
+void circ_message(const char *msg);
+#endif                                                 /* CONFIG_FEATURE_IPC_SYSLOG */
+
+/* Note: There is also a function called "message()" in init.c */
+/* Print a message to the log file. */
+static void message(char *fmt, ...) __attribute__ ((format(printf, 1, 2)));
+static void message(char *fmt, ...)
+{
+       int fd = -1;
+       struct flock fl;
+       va_list arguments;
+
+       fl.l_whence = SEEK_SET;
+       fl.l_start = 0;
+       fl.l_len = 1;
+
+#ifdef CONFIG_FEATURE_IPC_SYSLOG
+       if ((option_mask32 & OPT_circularlog) && shbuf) {
+               char b[1024];
+
+               va_start(arguments, fmt);
+               vsnprintf(b, sizeof(b) - 1, fmt, arguments);
+               va_end(arguments);
+               circ_message(b);
+
+       } else
+#endif
+       fd = device_open(logFilePath, O_WRONLY | O_CREAT
+                                       | O_NOCTTY | O_APPEND | O_NONBLOCK);
+       if (fd >= 0) {
+               fl.l_type = F_WRLCK;
+               fcntl(fd, F_SETLKW, &fl);
+
+#ifdef CONFIG_FEATURE_ROTATE_LOGFILE
+               if (ENABLE_FEATURE_ROTATE_LOGFILE && logFileSize > 0 ) {
+                       struct stat statf;
+                       int r = fstat(fd, &statf);
+                       if (!r && (statf.st_mode & S_IFREG)
+                                       && (lseek(fd,0,SEEK_END) > logFileSize)) {
+                               if (logFileRotate > 0) {
+                                       int i = strlen(logFilePath) + 4;
+                                       char oldFile[i];
+                                       char newFile[i];
+                                       for (i=logFileRotate-1; i>0; i--) {
+                                               sprintf(oldFile, "%s.%d", logFilePath, i-1);
+                                               sprintf(newFile, "%s.%d", logFilePath, i);
+                                               rename(oldFile, newFile);
+                                       }
+                                       sprintf(newFile, "%s.%d", logFilePath, 0);
+                                       fl.l_type = F_UNLCK;
+                                       fcntl(fd, F_SETLKW, &fl);
+                                       close(fd);
+                                       rename(logFilePath, newFile);
+                                       fd = device_open(logFilePath,
+                                                  O_WRONLY | O_CREAT | O_NOCTTY | O_APPEND |
+                                                  O_NONBLOCK);
+                                       fl.l_type = F_WRLCK;
+                                       fcntl(fd, F_SETLKW, &fl);
+                               } else {
+                                       ftruncate(fd, 0);
+                               }
+                       }
+               }
+#endif
+               va_start(arguments, fmt);
+               vdprintf(fd, fmt, arguments);
+               va_end(arguments);
+               fl.l_type = F_UNLCK;
+               fcntl(fd, F_SETLKW, &fl);
+               close(fd);
+       } else {
+               /* Always send console messages to /dev/console so people will see them. */
+               fd = device_open(_PATH_CONSOLE, O_WRONLY | O_NOCTTY | O_NONBLOCK);
+               if (fd >= 0) {
+                       va_start(arguments, fmt);
+                       vdprintf(fd, fmt, arguments);
+                       va_end(arguments);
+                       close(fd);
+               } else {
+                       fprintf(stderr, "Bummer, can't print: ");
+                       va_start(arguments, fmt);
+                       vfprintf(stderr, fmt, arguments);
+                       fflush(stderr);
+                       va_end(arguments);
+               }
+       }
+}
+
+static void logMessage(int pri, char *msg)
+{
+       time_t now;
+       char *timestamp;
+       char res[20];
+       CODE *c_pri, *c_fac;
+
+       if (pri != 0) {
+               c_fac = facilitynames;
+               while (c_fac->c_name && !(c_fac->c_val == LOG_FAC(pri) << 3))
+                       c_fac++;
+               c_pri = prioritynames;
+               while (c_pri->c_name && !(c_pri->c_val == LOG_PRI(pri)))
+                       c_pri++;
+               if (c_fac->c_name == NULL || c_pri->c_name == NULL) {
+                       snprintf(res, sizeof(res), "<%d>", pri);
+               } else {
+                       snprintf(res, sizeof(res), "%s.%s", c_fac->c_name, c_pri->c_name);
+               }
+       }
+
+       if (strlen(msg) < 16 || msg[3] != ' ' || msg[6] != ' ' ||
+               msg[9] != ':' || msg[12] != ':' || msg[15] != ' ') {
+               time(&now);
+               timestamp = ctime(&now) + 4;
+               timestamp[15] = '\0';
+       } else {
+               timestamp = msg;
+               timestamp[15] = '\0';
+               msg += 16;
+       }
+
+       /* todo: supress duplicates */
+
+#ifdef CONFIG_FEATURE_REMOTE_LOG
+       if (option_mask32 & OPT_remotelog) {
+               char line[MAXLINE + 1];
+               /* trying connect the socket */
+               if (-1 == remotefd) {
+                       remotefd = socket(AF_INET, SOCK_DGRAM, 0);
+               }
+               /* if we have a valid socket, send the message */
+               if (-1 != remotefd) {
+                       snprintf(line, sizeof(line), "<%d>%s", pri, msg);
+                       /* send message to remote logger, ignore possible error */
+                       sendto(remotefd, line, strlen(line), 0,
+                                       (struct sockaddr *) &remoteaddr, sizeof(remoteaddr));
+               }
+       }
+
+       if (option_mask32 & OPT_locallog)
+#endif
+       {
+               /* now spew out the message to wherever it is supposed to go */
+               if (pri == 0 || LOG_PRI(pri) < logLevel) {
+                       if (option_mask32 & OPT_small)
+                               message("%s %s\n", timestamp, msg);
+                       else
+                               message("%s %s %s %s\n", timestamp, LocalHostName, res, msg);
+               }
+       }
 }
 
-/* Copies data from file descriptor 'from' to file descriptor 'to'
-   until nothing is left to be copied. Exits if an error occurs.   */
-void copyData(int from, int to) {
-    char buf[1024];
-    int amount;
-    
-    while ((amount = read(from, buf, sizeof(buf))) > 0) {
-        if (write(to, buf, amount) != amount) {
-            die("write");
-            return;
-        }
-    }
-    if (amount < 0)
-        die("read");
+static void quit_signal(int sig)
+{
+       logMessage(LOG_SYSLOG | LOG_INFO, "System log daemon exiting.");
+       unlink(lfile);
+       if (ENABLE_FEATURE_IPC_SYSLOG)
+               ipcsyslog_cleanup();
+
+       exit(1);
+}
+
+static void domark(int sig)
+{
+       if (MarkInterval > 0) {
+               logMessage(LOG_SYSLOG | LOG_INFO, "-- MARK --");
+               alarm(MarkInterval);
+       }
 }
 
+/* This must be a #define, since when CONFIG_DEBUG and BUFFERS_GO_IN_BSS are
+ * enabled, we otherwise get a "storage size isn't constant error. */
+static int serveConnection(char *tmpbuf, int n_read)
+{
+       char *p = tmpbuf;
+
+       while (p < tmpbuf + n_read) {
 
-int main(void) {
-    struct sockaddr_un address;
-    int fd, conn;
-    size_t addrLength;
+               int pri = (LOG_USER | LOG_NOTICE);
+               int num_lt = 0;
+               char line[MAXLINE + 1];
+               unsigned char c;
+               char *q = line;
 
-    /* Remove any preexisting socket (or other file) */
-    unlink(_PATH_LOG);
+               while ((c = *p) && q < &line[sizeof(line) - 1]) {
+                       if (c == '<' && num_lt == 0) {
+                               /* Parse the magic priority number. */
+                               num_lt++;
+                               pri = 0;
+                               while (isdigit(*(++p))) {
+                                       pri = 10 * pri + (*p - '0');
+                               }
+                               if (pri & ~(LOG_FACMASK | LOG_PRIMASK)) {
+                                       pri = (LOG_USER | LOG_NOTICE);
+                               }
+                       } else if (c == '\n') {
+                               *q++ = ' ';
+                       } else if (iscntrl(c) && (c < 0177)) {
+                               *q++ = '^';
+                               *q++ = c ^ 0100;
+                       } else {
+                               *q++ = c;
+                       }
+                       p++;
+               }
+               *q = '\0';
+               p++;
+               /* Now log it */
+               logMessage(pri, line);
+       }
+       return n_read;
+}
+
+static void doSyslogd(void) ATTRIBUTE_NORETURN;
+static void doSyslogd(void)
+{
+       struct sockaddr_un sunx;
+       socklen_t addrLength;
 
-    memset(&address, 0, sizeof(address));
-    address.sun_family = AF_UNIX;      /* Unix domain socket */
-    strncpy(address.sun_path, _PATH_LOG, sizeof(address.sun_path));
+       int sock_fd;
+       fd_set fds;
 
-    if ((fd = socket(PF_UNIX, SOCK_STREAM, 0)) < 0)
-        die("socket");
+       /* Set up signal handlers. */
+       signal(SIGINT, quit_signal);
+       signal(SIGTERM, quit_signal);
+       signal(SIGQUIT, quit_signal);
+       signal(SIGHUP, SIG_IGN);
+       signal(SIGCHLD, SIG_IGN);
+#ifdef SIGCLD
+       signal(SIGCLD, SIG_IGN);
+#endif
+       signal(SIGALRM, domark);
+       alarm(MarkInterval);
 
+       /* Create the syslog file so realpath() can work. */
+       if (realpath(_PATH_LOG, lfile) != NULL) {
+               unlink(lfile);
+       }
 
-    /* The total length of the address includes the sun_family 
-       element */
-    addrLength = sizeof(address.sun_family) + 
-                 strlen(address.sun_path);
+       memset(&sunx, 0, sizeof(sunx));
+       sunx.sun_family = AF_UNIX;
+       strncpy(sunx.sun_path, lfile, sizeof(sunx.sun_path));
+       sock_fd = xsocket(AF_UNIX, SOCK_DGRAM, 0);
+       addrLength = sizeof(sunx.sun_family) + strlen(sunx.sun_path);
+       if (bind(sock_fd, (struct sockaddr *) &sunx, addrLength) < 0) {
+               bb_perror_msg_and_die("cannot connect to socket %s", lfile);
+       }
 
-    if (bind(fd, (struct sockaddr *) &address, addrLength))
-        die("bind");
+       if (chmod(lfile, 0666) < 0) {
+               bb_perror_msg_and_die("cannot set permission on %s", lfile);
+       }
+       if (ENABLE_FEATURE_IPC_SYSLOG && (option_mask32 & OPT_circularlog)) {
+               ipcsyslog_init();
+       }
 
-    if (chmod(_PATH_LOG, 0666) < 0)
-        die("chmod");
+       logMessage(LOG_SYSLOG | LOG_INFO, "syslogd started: " "BusyBox v" BB_VER );
 
-    if (listen(fd, 5))
-        die("listen");
+       for (;;) {
+               FD_ZERO(&fds);
+               FD_SET(sock_fd, &fds);
 
+               if (select(sock_fd + 1, &fds, NULL, NULL, NULL) < 0) {
+                       if (errno == EINTR) {
+                               /* alarm may have happened. */
+                               continue;
+                       }
+                       bb_perror_msg_and_die("select");
+               }
 
-    while ((conn = accept(fd, (struct sockaddr *) &address, 
-                          &addrLength)) >= 0) {
-        printf("---- getting data\n");
-        copyData(conn, fileno(stdout));
-        printf("---- done\n");
-        close(conn);
-    }
+               if (FD_ISSET(sock_fd, &fds)) {
+                       int i;
+#if MAXLINE > BUFSIZ
+# define TMP_BUF_SZ BUFSIZ
+#else
+# define TMP_BUF_SZ MAXLINE
+#endif
+#define tmpbuf bb_common_bufsiz1
 
-    if (conn < 0) 
-        die("accept");
-    
-    close(fd);
-    return 0;
+                       if ((i = recv(sock_fd, tmpbuf, TMP_BUF_SZ, 0)) > 0) {
+                               tmpbuf[i] = '\0';
+                               serveConnection(tmpbuf, i);
+                       } else {
+                               bb_perror_msg_and_die("UNIX socket error");
+                       }
+               }                               /* FD_ISSET() */
+       }                                       /* for main loop */
 }
 
 
+int syslogd_main(int argc, char **argv)
+{
+       char OPTION_DECL;
+       char *p;
+
+       /* do normal option parsing */
+       getopt32(argc, argv, OPTION_STR, OPTION_PARAM);
+       if (option_mask32 & OPT_mark) MarkInterval = xatoul_range(opt_m, 0, INT_MAX/60) * 60; // -m
+       //if (option_mask32 & OPT_nofork) // -n
+       //if (option_mask32 & OPT_outfile) // -O
+       if (option_mask32 & OPT_loglevel) { // -l
+               logLevel = xatoi_u(opt_l);
+               /* Valid levels are between 1 and 8 */
+               if (logLevel < 1 || logLevel > 8)
+                       bb_show_usage();
+       }
+       //if (option_mask32 & OPT_small) // -S
+#if ENABLE_FEATURE_ROTATE_LOGFILE
+       if (option_mask32 & OPT_filesize) logFileSize = xatoul_range(opt_s, 0, INT_MAX/1024) * 1024; // -s
+       if (option_mask32 & OPT_rotatecnt) { // -b
+               logFileRotate = xatoi_u(opt_b);
+               if (logFileRotate > 99) logFileRotate = 99; 
+       }
+#endif
+#if ENABLE_FEATURE_REMOTE_LOG
+       if (option_mask32 & OPT_remotelog) { // -R
+               int port = 514;
+               char *host = xstrdup(opt_R);
+               p = strchr(host, ':');
+               if (p) {
+                       port = xatou16(p + 1);
+                       *p = '\0';
+               }
+               remoteaddr.sin_family = AF_INET;
+               /* FIXME: looks ip4-specific. need to do better */
+               remoteaddr.sin_addr = *(struct in_addr *) *(xgethostbyname(host)->h_addr_list);
+               remoteaddr.sin_port = htons(port);
+               free(host);
+       }
+       //if (option_mask32 & OPT_locallog) // -L
+#endif
+#if ENABLE_FEATURE_IPC_SYSLOG
+       if (option_mask32 & OPT_circularlog) { // -C
+               if (opt_C) {
+                       shm_size = xatoul_range(opt_C, 4, INT_MAX/1024) * 1024;
+               }
+       }
+#endif
+
+       /* If they have not specified remote logging, then log locally */
+       if (ENABLE_FEATURE_REMOTE_LOG && !(option_mask32 & OPT_remotelog))
+               option_mask32 |= OPT_locallog;
 
+       /* Store away localhost's name before the fork */
+       gethostname(LocalHostName, sizeof(LocalHostName));
+       p = strchr(LocalHostName, '.');
+       if (p) {
+               *p = '\0';
+       }
+
+       umask(0);
+
+       if (!(option_mask32 & OPT_nofork)) {
+#ifdef BB_NOMMU
+               vfork_daemon_rexec(0, 1, argc, argv, "-n");
+#else
+               xdaemon(0, 1);
+#endif
+       }
+       doSyslogd();
+
+       return EXIT_SUCCESS;
+}