chiark / gitweb /
[PATCH] PATCH udev close on exec
[elogind.git] / udevd.c
diff --git a/udevd.c b/udevd.c
index 26e6b23486b23963fa34cfb09ed102520deded59..81f4474a89b6bc7df5ddd834cff3180d0cb12bdf 100644 (file)
--- a/udevd.c
+++ b/udevd.c
@@ -1,10 +1,8 @@
 /*
- * udevd.c
+ * udevd.c - hotplug event serializer
  *
- * Userspace devfs
- *
- * Copyright (C) 2004 Ling, Xiaofeng <xiaofeng.ling@intel.com>
  * Copyright (C) 2004 Kay Sievers <kay.sievers@vrfy.org>
+ * Copyright (C) 2004 Chris Friesen <chris_friesen@sympatico.ca>
  *
  *
  *     This program is free software; you can redistribute it and/or modify it
  */
 
 #include <stddef.h>
-#include <sys/types.h>
-#include <sys/ipc.h>
 #include <sys/wait.h>
-#include <sys/msg.h>
 #include <signal.h>
 #include <unistd.h>
 #include <errno.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <time.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/socket.h>
+#include <sys/un.h>
 #include <fcntl.h>
+#include "klibc_fixups.h"
+#include <sys/sysinfo.h>
 
 #include "list.h"
 #include "udev.h"
+#include "udev_lib.h"
+#include "udev_version.h"
 #include "udevd.h"
 #include "logging.h"
 
-#define BUFFER_SIZE                    1024
+static int pipefds[2];
+static unsigned long long expected_seqnum = 0;
+static volatile int children_waiting;
+static volatile int run_msg_q;
+static volatile int sig_flag;
+static int run_exec_q;
+
+static LIST_HEAD(msg_list);
+static LIST_HEAD(exec_list);
+static LIST_HEAD(running_list);
+
+static void exec_queue_manager(void);
+static void msg_queue_manager(void);
+static void user_sighandler(void);
+static void reap_kids(void);
+char *udev_bin;
+
+#ifdef LOG
+unsigned char logname[LOGNAME_SIZE];
+void log_message (int level, const char *format, ...)
+{
+       va_list args;
 
-static int expect_seqnum = 0;
-static int lock_file = -1;
-static char *lock_filename = ".udevd_lock";
+       va_start(args, format);
+       vsyslog(level, format, args);
+       va_end(args);
+}
+#endif
 
-LIST_HEAD(msg_list);
+#define msg_dump(msg) \
+       dbg("msg_dump: sequence %llu, '%s', '%s', '%s'", \
+       msg->seqnum, msg->action, msg->devpath, msg->subsystem);
 
-static void sig_handler(int signum)
+static void msg_dump_queue(void)
 {
-       dbg("caught signal %d", signum);
-       switch (signum) {
-       case SIGALRM:
-               dbg("event timeout reached");
-               break;
-       case SIGINT:
-       case SIGTERM:
-       case SIGKILL:
-               if (lock_file >= 0) {
-                       close(lock_file);
-                       unlink(lock_filename);
-               }
-               exit(20 + signum);
-               break;
-       default:
-               dbg("unhandled signal");
-       }
+#ifdef DEBUG
+       struct hotplug_msg *msg;
+
+       list_for_each_entry(msg, &msg_list, list)
+               dbg("sequence %llu in queue", msg->seqnum);
+#endif
 }
 
-static void dump_queue(void)
+static struct hotplug_msg *msg_create(void)
 {
-       struct hotplug_msg *msg;
+       struct hotplug_msg *new_msg;
 
-       list_for_each_entry(msg, &msg_list, list)
-               dbg("sequence %d in queue", msg->seqnum);
+       new_msg = malloc(sizeof(struct hotplug_msg));
+       if (new_msg == NULL)
+               dbg("error malloc");
+       return new_msg;
 }
 
-static void dump_msg(struct hotplug_msg *msg)
+static void run_queue_delete(struct hotplug_msg *msg)
 {
-       dbg("sequence %d, '%s', '%s', '%s'",
-           msg->seqnum, msg->action, msg->devpath, msg->subsystem);
+       list_del(&msg->list);
+       free(msg);
 }
 
-static int dispatch_msg(struct hotplug_msg *msg)
+/* orders the message in the queue by sequence number */
+static void msg_queue_insert(struct hotplug_msg *msg)
 {
-       pid_t pid;
+       struct hotplug_msg *loop_msg;
+       struct sysinfo info;
+
+       /* sort message by sequence number into list. events
+        * will tend to come in order, so scan the list backwards
+        */
+       list_for_each_entry_reverse(loop_msg, &msg_list, list)
+               if (loop_msg->seqnum < msg->seqnum)
+                       break;
+
+       /* store timestamp of queuing */
+       sysinfo(&info);
+       msg->queue_time = info.uptime;
 
-       dump_msg(msg);
+       list_add(&msg->list, &loop_msg->list);
+       dbg("queued message seq %llu", msg->seqnum);
 
-       setenv("ACTION", msg->action, 1);
-       setenv("DEVPATH", msg->devpath, 1);
+       /* run msg queue manager */
+       run_msg_q = 1;
+
+       return ;
+}
+
+/* forks event and removes event from run queue when finished */
+static void udev_run(struct hotplug_msg *msg)
+{
+       pid_t pid;
+       char action[ACTION_SIZE];
+       char devpath[DEVPATH_SIZE];
+       char seqnum[SEQNUM_SIZE];
+       char *env[] = { action, devpath, seqnum, NULL };
+
+       snprintf(action, ACTION_SIZE-1, "ACTION=%s", msg->action);
+       action[ACTION_SIZE-1] = '\0';
+       snprintf(devpath, DEVPATH_SIZE-1, "DEVPATH=%s", msg->devpath);
+       devpath[DEVPATH_SIZE-1] = '\0';
+       sprintf(seqnum, "SEQNUM=%llu", msg->seqnum);
 
        pid = fork();
        switch (pid) {
        case 0:
                /* child */
-               execl(UDEV_EXEC, "udev", msg->subsystem, NULL);
+               execle(udev_bin, "udev", msg->subsystem, NULL, env);
                dbg("exec of child failed");
                exit(1);
                break;
        case -1:
                dbg("fork of child failed");
-               return -1;
+               run_queue_delete(msg);
+               /* note: we never managed to run, so we had no impact on 
+                * running_with_devpath(), so don't bother setting run_exec_q
+                */
+               break;
        default:
-               wait(NULL);
+               /* get SIGCHLD in main loop */
+               dbg("==> exec seq %llu [%d] working at '%s'", msg->seqnum, pid, msg->devpath);
+               msg->pid = pid;
        }
-       return 0;
 }
 
-static void set_timeout(int seconds)
+/* returns already running task with devpath */
+static struct hotplug_msg *running_with_devpath(struct hotplug_msg *msg)
 {
-       alarm(seconds);
-       dbg("set timeout in %d seconds", seconds);
+       struct hotplug_msg *loop_msg;
+       list_for_each_entry(loop_msg, &running_list, list)
+               if (strncmp(loop_msg->devpath, msg->devpath, sizeof(loop_msg->devpath)) == 0)
+                       return loop_msg;
+       return NULL;
 }
 
-static void check_queue(void)
+/* exec queue management routine executes the events and delays events for the same devpath */
+static void exec_queue_manager()
 {
+       struct hotplug_msg *loop_msg;
+       struct hotplug_msg *tmp_msg;
        struct hotplug_msg *msg;
+
+       list_for_each_entry_safe(loop_msg, tmp_msg, &exec_list, list) {
+               msg = running_with_devpath(loop_msg);
+               if (!msg) {
+                       /* move event to run list */
+                       list_move_tail(&loop_msg->list, &running_list);
+                       udev_run(loop_msg);
+                       dbg("moved seq %llu to running list", loop_msg->seqnum);
+               } else {
+                       dbg("delay seq %llu, cause seq %llu already working on '%s'",
+                               loop_msg->seqnum, msg->seqnum, msg->devpath);
+               }
+       }
+}
+
+static void msg_move_exec(struct hotplug_msg *msg)
+{
+       list_move_tail(&msg->list, &exec_list);
+       run_exec_q = 1;
+       expected_seqnum = msg->seqnum+1;
+       dbg("moved seq %llu to exec, next expected is %llu",
+               msg->seqnum, expected_seqnum);
+}
+
+/* msg queue management routine handles the timeouts and dispatches the events */
+static void msg_queue_manager()
+{
+       struct hotplug_msg *loop_msg;
        struct hotplug_msg *tmp_msg;
-       time_t msg_age;
+       struct sysinfo info;
+       long msg_age = 0;
 
+       dbg("msg queue manager, next expected is %llu", expected_seqnum);
 recheck:
-       /* dispatch events until one is missing */
-       list_for_each_entry_safe(msg, tmp_msg, &msg_list, list) {
-               if (msg->seqnum != expect_seqnum)
-                       break;
-               dispatch_msg(msg);
-               expect_seqnum++;
-               list_del_init(&msg->list);
-               free(msg);
-       }
+       list_for_each_entry_safe(loop_msg, tmp_msg, &msg_list, list) {
+               /* move event with expected sequence to the exec list */
+               if (loop_msg->seqnum == expected_seqnum) {
+                       msg_move_exec(loop_msg);
+                       continue;
+               }
 
-       /* recalculate timeout */
-       if (list_empty(&msg_list) == 0) {
-               msg_age = time(NULL) - msg->queue_time;
-               if (msg_age > EVENT_TIMEOUT_SECONDS-1) {
-                       info("event %d, age %li seconds, skip event %d-%d",
-                            msg->seqnum, msg_age, expect_seqnum, msg->seqnum-1);
-                       expect_seqnum = msg->seqnum;
+               /* move event with expired timeout to the exec list */
+               sysinfo(&info);
+               msg_age = info.uptime - loop_msg->queue_time;
+               dbg("seq %llu is %li seconds old", loop_msg->seqnum, msg_age);
+               if (msg_age > EVENT_TIMEOUT_SEC-1) {
+                       msg_move_exec(loop_msg);
                        goto recheck;
+               } else {
+                       break;
                }
-               set_timeout(EVENT_TIMEOUT_SECONDS - msg_age);
-               return;
        }
 
-       /* queue is empty */
-       set_timeout(UDEVD_TIMEOUT_SECONDS);
+       msg_dump_queue();
+
+       /* set timeout for remaining queued events */
+       if (list_empty(&msg_list) == 0) {
+               struct itimerval itv = {{0, 0}, {EVENT_TIMEOUT_SEC - msg_age, 0}};
+               dbg("next event expires in %li seconds", EVENT_TIMEOUT_SEC - msg_age);
+               setitimer(ITIMER_REAL, &itv, 0);
+       }
 }
 
-static int queue_msg(struct hotplug_msg *msg)
+/* receive the msg, do some basic sanity checks, and queue it */
+static void handle_msg(int sock)
 {
-       struct hotplug_msg *new_msg;
-       struct hotplug_msg *tmp_msg;
+       struct hotplug_msg *msg;
+       int retval;
+       struct msghdr smsg;
+       struct cmsghdr *cmsg;
+       struct iovec iov;
+       struct ucred *cred;
+       char cred_msg[CMSG_SPACE(sizeof(struct ucred))];
+
+       msg = msg_create();
+       if (msg == NULL) {
+               dbg("unable to store message");
+               return;
+       }
 
-       new_msg = malloc(sizeof(*new_msg));
-       if (new_msg == NULL) {
-               dbg("error malloc");
-               return -ENOMEM;
+       iov.iov_base = msg;
+       iov.iov_len = sizeof(struct hotplug_msg);
+
+       memset(&smsg, 0x00, sizeof(struct msghdr));
+       smsg.msg_iov = &iov;
+       smsg.msg_iovlen = 1;
+       smsg.msg_control = cred_msg;
+       smsg.msg_controllen = sizeof(cred_msg);
+
+       retval = recvmsg(sock, &smsg, 0);
+       if (retval <  0) {
+               if (errno != EINTR)
+                       dbg("unable to receive message");
+               return;
        }
-       memcpy(new_msg, msg, sizeof(*new_msg));
+       cmsg = CMSG_FIRSTHDR(&smsg);
+       cred = (struct ucred *) CMSG_DATA(cmsg);
 
-       /* store timestamp of queuing */
-       new_msg->queue_time = time(NULL);
+       if (cmsg == NULL || cmsg->cmsg_type != SCM_CREDENTIALS) {
+               dbg("no sender credentials received, message ignored");
+               goto skip;
+       }
 
-       /* sort message by sequence number into list*/
-       list_for_each_entry(tmp_msg, &msg_list, list)
-               if (tmp_msg->seqnum > new_msg->seqnum)
-                       break;
-       list_add_tail(&new_msg->list, &tmp_msg->list);
+       if (cred->uid != 0) {
+               dbg("sender uid=%i, message ignored", cred->uid);
+               goto skip;
+       }
+
+       if (strncmp(msg->magic, UDEV_MAGIC, sizeof(UDEV_MAGIC)) != 0 ) {
+               dbg("message magic '%s' doesn't match, ignore it", msg->magic);
+               goto skip;
+       }
+
+       /* if no seqnum is given, we move straight to exec queue */
+       if (msg->seqnum == 0) {
+               list_add(&msg->list, &exec_list);
+               run_exec_q = 1;
+       } else {
+               msg_queue_insert(msg);
+       }
+       return;
 
-       return 0;
+skip:
+       free(msg);
+       return;
 }
 
-static void work(void)
+asmlinkage static void sig_handler(int signum)
 {
-       struct hotplug_msg *msg;
-       int msgid;
-       key_t key;
-       char buf[BUFFER_SIZE];
-       int ret;
-
-       key = ftok(UDEVD_EXEC, IPC_KEY_ID);
-       msg = (struct hotplug_msg *) buf;
-       msgid = msgget(key, IPC_CREAT);
-       if (msgid == -1) {
-               dbg("open message queue error");
-               exit(1);
+       int rc;
+
+       switch (signum) {
+               case SIGINT:
+               case SIGTERM:
+                       exit(20 + signum);
+                       break;
+               case SIGALRM:
+                       /* set flag, then write to pipe if needed */
+                       run_msg_q = 1;
+                       goto do_write;
+                       break;
+               case SIGCHLD:
+                       /* set flag, then write to pipe if needed */
+                       children_waiting = 1;
+                       goto do_write;
+                       break;
+               default:
+                       dbg("unhandled signal %d", signum);
+                       return;
        }
-       while (1) {
-               ret = msgrcv(msgid, (struct msgbuf *) buf, BUFFER_SIZE-4, HOTPLUGMSGTYPE, 0);
-               if (ret != -1) {
-                       /* init the expected sequence with value from first call */
-                       if (expect_seqnum == 0) {
-                               expect_seqnum = msg->seqnum;
-                               dbg("init next expected sequence number to %d", expect_seqnum);
-                       }
-                       dbg("current sequence %d, expected sequence %d", msg->seqnum, expect_seqnum);
-                       if (msg->seqnum == expect_seqnum) {
-                               /* execute expected event */
-                               dispatch_msg(msg);
-                               expect_seqnum++;
-                               check_queue();
-                               dump_queue();
-                               continue;
-                       }
-                       if (msg->seqnum > expect_seqnum) {
-                               /* something missing, queue event*/
-                               queue_msg(msg);
-                               check_queue();
-                               dump_queue();
-                               continue;
-                       }
-                       dbg("too late for event with sequence %d, even skipped ", msg->seqnum);
-               } else {
-                       if (errno == EINTR) {
-                               /* timeout */
-                               if (list_empty(&msg_list)) {
-                                       info("we have nothing to do, so daemon exits...");
-                                       if (lock_file >= 0) {
-                                               close(lock_file);
-                                               unlink(lock_filename);
-                                       }
-                                       exit(0);
-                               }
-                               check_queue();
-                               dump_queue();
-                               continue;
-                       }
-                       dbg("ipc message receive error '%s'", strerror(errno));
-               }
+       
+do_write:
+       /* if pipe is empty, write to pipe to force select to return
+        * immediately when it gets called
+        */
+       if (!sig_flag) {
+               rc = write(pipefds[1],&signum,sizeof(signum));
+               if (rc < 0)
+                       dbg("unable to write to pipe");
+               else
+                       sig_flag = 1;
        }
 }
 
-static int one_and_only(void)
+static void udev_done(int pid)
 {
-       char string[100];
+       /* find msg associated with pid and delete it */
+       struct hotplug_msg *msg;
 
-       lock_file = open(lock_filename, O_RDWR | O_CREAT, 0x640);
+       list_for_each_entry(msg, &running_list, list) {
+               if (msg->pid == pid) {
+                       dbg("<== exec seq %llu came back", msg->seqnum);
+                       run_queue_delete(msg);
+
+                       /* we want to run the exec queue manager since there may
+                        * be events waiting with the devpath of the one that
+                        * just finished
+                        */
+                       run_exec_q = 1;
+                       return;
+               }
+       }
+}
 
-       /* see if we can open */
-       if (lock_file < 0)
-               return -1;
-       
-       /* see if we can lock */
-       if (lockf(lock_file, F_TLOCK, 0) < 0) {
-               close(lock_file);
-               unlink(lock_filename);
-               return -1;
+static void reap_kids()
+{
+       /* reap all dead children */
+       while(1) {
+               int pid = waitpid(-1, 0, WNOHANG);
+               if ((pid == -1) || (pid == 0))
+                       break;
+               udev_done(pid);
        }
+}
 
-       snprintf(string, sizeof(string), "%d\n", getpid());
-       write(lock_file, string, strlen(string));
+/* just read everything from the pipe and clear the flag,
+ * the useful flags were set in the signal handler
+ */
+static void user_sighandler()
+{
+       int sig;
+       while(1) {
+               int rc = read(pipefds[0],&sig,sizeof(sig));
+               if (rc < 0)
+                       break;
 
-       return 0;
+               sig_flag = 0;
+       }
 }
 
+
 int main(int argc, char *argv[])
 {
-       /* only let one version of the daemon run at any one time */
-       if (one_and_only() != 0)
-               exit(0);
+       int ssock, maxsockplus;
+       struct sockaddr_un saddr;
+       socklen_t addrlen;
+       int retval, fd;
+       const int on = 1;
+       struct sigaction act;
+       fd_set readfds;
+
+       init_logging("udevd");
+       dbg("version %s", UDEV_VERSION);
+
+       if (getuid() != 0) {
+               dbg("need to be root, exit");
+               exit(1);
+       }
+       /* make sure we are at top of dir */
+       chdir("/");
+       umask( umask( 077 ) | 022 );
+       /* Set fds to dev/null */
+       fd = open( "/dev/null", O_RDWR );
+       if ( fd < 0 ) {
+               dbg("error opening /dev/null %s", strerror(errno));
+               exit(1);
+       }
+       dup2(fd, 0);
+       dup2(fd, 1);
+       dup2(fd, 2);
+       if (fd > 2) 
+               close(fd);
+       /* Get new session id so stray signals don't come our way. */
+       setsid();
+
+       /* setup signal handler pipe */
+       retval = pipe(pipefds);
+       if (retval < 0) {
+               dbg("error getting pipes: %s", strerror(errno));
+               exit(1);
+       }
 
-       /* set up signal handler */
-       signal(SIGINT, sig_handler);
-       signal(SIGTERM, sig_handler);
-       signal(SIGKILL, sig_handler);
-       signal(SIGALRM, sig_handler);
+       retval = fcntl(pipefds[0], F_SETFL, O_NONBLOCK);
+       if (retval < 0) {
+               dbg("error fcntl on read pipe: %s", strerror(errno));
+               exit(1);
+       }
+       retval = fcntl(pipefds[0], F_SETFD, FD_CLOEXEC);
+       if (retval < 0) {
+               dbg("error fcntl on read pipe: %s", strerror(errno));
+               exit(1);
+       }
+
+       retval = fcntl(pipefds[1], F_SETFL, O_NONBLOCK);
+       if (retval < 0) {
+               dbg("error fcntl on write pipe: %s", strerror(errno));
+               exit(1);
+       }
+       retval = fcntl(pipefds[1], F_SETFD, FD_CLOEXEC);
+       if (retval < 0) {
+               dbg("error fcntl on write pipe: %s", strerror(errno));
+               exit(1);
+       }
 
-       /* we exit if we have nothing to do, next event will start us again */
-       set_timeout(UDEVD_TIMEOUT_SECONDS);
+       
+       /* set signal handlers */
+       act.sa_handler = sig_handler;
+       sigemptyset(&act.sa_mask);
+       act.sa_flags = SA_RESTART;
+       sigaction(SIGINT, &act, NULL);
+       sigaction(SIGTERM, &act, NULL);
+       sigaction(SIGALRM, &act, NULL);
+       sigaction(SIGCHLD, &act, NULL);
+
+       memset(&saddr, 0x00, sizeof(saddr));
+       saddr.sun_family = AF_LOCAL;
+       /* use abstract namespace for socket path */
+       strcpy(&saddr.sun_path[1], UDEVD_SOCK_PATH);
+       addrlen = offsetof(struct sockaddr_un, sun_path) + strlen(saddr.sun_path+1) + 1;
+
+       ssock = socket(AF_LOCAL, SOCK_DGRAM, 0);
+       if (ssock == -1) {
+               dbg("error getting socket, exit");
+               exit(1);
+       }
+
+       set_cloexec_flag(ssock, 1);
+
+       /* the bind takes care of ensuring only one copy running */
+       retval = bind(ssock, (struct sockaddr *) &saddr, addrlen);
+       if (retval < 0) {
+               dbg("bind failed, exit");
+               goto exit;
+       }
+       retval = fcntl(ssock, F_SETFD, FD_CLOEXEC);
+       if (retval < 0) {
+               dbg("error fcntl on ssock: %s", strerror(errno));
+               exit(1);
+       }
 
-       work();
-       exit(0);
+       /* enable receiving of the sender credentials */
+       setsockopt(ssock, SOL_SOCKET, SO_PASSCRED, &on, sizeof(on));
+
+       /* possible override of udev binary, used for testing */
+       udev_bin = getenv("UDEV_BIN");
+       if (udev_bin != NULL)
+               dbg("udev binary is set to '%s'", udev_bin);
+       else
+               udev_bin = UDEV_BIN;
+
+       FD_ZERO(&readfds);
+       FD_SET(ssock, &readfds);
+       FD_SET(pipefds[0], &readfds);
+       maxsockplus = ssock+1;
+       while (1) {
+               fd_set workreadfds = readfds;
+               retval = select(maxsockplus, &workreadfds, NULL, NULL, NULL);
+
+               if (retval < 0) {
+                       if (errno != EINTR)
+                               dbg("error in select: %s", strerror(errno));
+                       continue;
+               }
+
+               if (FD_ISSET(ssock, &workreadfds))
+                       handle_msg(ssock);
+
+               if (FD_ISSET(pipefds[0], &workreadfds))
+                       user_sighandler();
+
+               if (children_waiting) {
+                       children_waiting = 0;
+                       reap_kids();
+               }
+
+               if (run_msg_q) {
+                       run_msg_q = 0;
+                       msg_queue_manager();
+               }
+
+               if (run_exec_q) {
+                       /* this is tricky.  exec_queue_manager() loops over exec_list, and
+                        * calls running_with_devpath(), which loops over running_list. This gives
+                        * O(N*M), which can get *nasty*.  Clean up running_list before
+                        * calling exec_queue_manager().
+                        */
+                       if (children_waiting) {
+                               children_waiting = 0;
+                               reap_kids();
+                       }
+
+                       run_exec_q = 0;
+                       exec_queue_manager();
+               }
+       }
+exit:
+       close(ssock);
+       exit(1);
 }