ctdb-scripts: Do not de-duplicate the interfaces list
[samba.git] / lib / tdb / common / mutex.c
index e57031dc222b5de64c4497f8054d6b34221788fe..a7106164445df258fb4cf8385e37f151ea74c55e 100644 (file)
@@ -149,7 +149,7 @@ static bool tdb_mutex_index(struct tdb_context *tdb, off_t off, off_t len,
 
 static bool tdb_have_mutex_chainlocks(struct tdb_context *tdb)
 {
-       size_t i;
+       int i;
 
        for (i=0; i < tdb->num_lockrecs; i++) {
                bool ret;
@@ -454,7 +454,7 @@ int tdb_mutex_allrecord_upgrade(struct tdb_context *tdb)
 
        /*
         * Our only caller tdb_allrecord_upgrade()
-        * garantees that we already own the allrecord lock.
+        * guarantees that we already own the allrecord lock.
         *
         * Which means m->allrecord_mutex is still locked by us.
         */
@@ -502,7 +502,7 @@ void tdb_mutex_allrecord_downgrade(struct tdb_context *tdb)
 
        /*
         * Our only caller tdb_allrecord_upgrade() (in the error case)
-        * garantees that we already own the allrecord lock.
+        * guarantees that we already own the allrecord lock.
         *
         * Which means m->allrecord_mutex is still locked by us.
         */
@@ -531,7 +531,7 @@ int tdb_mutex_allrecord_unlock(struct tdb_context *tdb)
        /*
         * Our only callers tdb_allrecord_unlock() and
         * tdb_allrecord_lock() (in the error path)
-        * garantee that we already own the allrecord lock.
+        * guarantee that we already own the allrecord lock.
         *
         * Which means m->allrecord_mutex is still locked by us.
         */
@@ -559,7 +559,8 @@ int tdb_mutex_init(struct tdb_context *tdb)
 {
        struct tdb_mutexes *m;
        pthread_mutexattr_t ma;
-       int i, ret;
+       uint32_t i;
+       int ret;
 
        ret = tdb_mutex_mmap(tdb);
        if (ret == -1) {
@@ -603,12 +604,13 @@ int tdb_mutex_init(struct tdb_context *tdb)
 fail:
        pthread_mutexattr_destroy(&ma);
 fail_munmap:
-       tdb_mutex_munmap(tdb);
 
        if (ret == 0) {
                return 0;
        }
 
+       tdb_mutex_munmap(tdb);
+
        errno = ret;
        return -1;
 }
@@ -623,6 +625,10 @@ int tdb_mutex_mmap(struct tdb_context *tdb)
                return 0;
        }
 
+       if (tdb->mutexes != NULL) {
+               return 0;
+       }
+
        ptr = mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_SHARED|MAP_FILE,
                   tdb->fd, 0);
        if (ptr == MAP_FAILED) {
@@ -636,13 +642,20 @@ int tdb_mutex_mmap(struct tdb_context *tdb)
 int tdb_mutex_munmap(struct tdb_context *tdb)
 {
        size_t len;
+       int ret;
 
        len = tdb_mutex_size(tdb);
        if (len == 0) {
                return 0;
        }
 
-       return munmap(tdb->mutexes, len);
+       ret = munmap(tdb->mutexes, len);
+       if (ret == -1) {
+               return -1;
+       }
+       tdb->mutexes = NULL;
+
+       return 0;
 }
 
 static bool tdb_mutex_locking_cached;
@@ -740,12 +753,23 @@ static bool tdb_robust_mutex_setup_sigchild(void (*handler)(int),
 
 static void tdb_robust_mutex_handler(int sig)
 {
-       if (tdb_robust_mutex_pid != -1) {
+       pid_t child_pid = tdb_robust_mutex_pid;
+
+       if (child_pid != -1) {
                pid_t pid;
-               int status;
 
-               pid = waitpid(tdb_robust_mutex_pid, &status, WNOHANG);
-               if (pid == tdb_robust_mutex_pid) {
+               pid = waitpid(child_pid, NULL, WNOHANG);
+               if (pid == -1) {
+                       switch (errno) {
+                       case ECHILD:
+                               tdb_robust_mutex_pid = -1;
+                               return;
+
+                       default:
+                               return;
+                       }
+               }
+               if (pid == child_pid) {
                        tdb_robust_mutex_pid = -1;
                        return;
                }
@@ -764,10 +788,48 @@ static void tdb_robust_mutex_handler(int sig)
        tdb_robust_mutext_old_handler(sig);
 }
 
+static void tdb_robust_mutex_wait_for_child(pid_t *child_pid)
+{
+       int options = WNOHANG;
+
+       if (*child_pid == -1) {
+               return;
+       }
+
+       while (tdb_robust_mutex_pid > 0) {
+               pid_t pid;
+
+               /*
+                * First we try with WNOHANG, as the process might not exist
+                * anymore. Once we've sent SIGKILL we block waiting for the
+                * exit.
+                */
+               pid = waitpid(*child_pid, NULL, options);
+               if (pid == -1) {
+                       if (errno == EINTR) {
+                               continue;
+                       } else if (errno == ECHILD) {
+                               break;
+                       } else {
+                               abort();
+                       }
+               }
+               if (pid == *child_pid) {
+                       break;
+               }
+
+               kill(*child_pid, SIGKILL);
+               options = 0;
+       }
+
+       tdb_robust_mutex_pid = -1;
+       *child_pid = -1;
+}
+
 _PUBLIC_ bool tdb_runtime_check_for_robust_mutexes(void)
 {
-       void *ptr;
-       pthread_mutex_t *m;
+       void *ptr = NULL;
+       pthread_mutex_t *m = NULL;
        pthread_mutexattr_t ma;
        int ret = 1;
        int pipe_down[2] = { -1, -1 };
@@ -776,7 +838,8 @@ _PUBLIC_ bool tdb_runtime_check_for_robust_mutexes(void)
        char c = 0;
        bool ok;
        static bool initialized;
-       sigset_t mask, old_mask, suspend_mask;
+       pid_t saved_child_pid = -1;
+       bool cleanup_ma = false;
 
        if (initialized) {
                return tdb_mutex_locking_cached;
@@ -796,57 +859,46 @@ _PUBLIC_ bool tdb_runtime_check_for_robust_mutexes(void)
        if (ptr == MAP_FAILED) {
                return false;
        }
-       m = (pthread_mutex_t *)ptr;
 
        ret = pipe(pipe_down);
        if (ret != 0) {
-               goto cleanup_mmap;
+               goto cleanup;
        }
        ret = pipe(pipe_up);
        if (ret != 0) {
-               goto cleanup_pipe;
+               goto cleanup;
        }
 
        ret = pthread_mutexattr_init(&ma);
        if (ret != 0) {
-               goto cleanup_pipe;
+               goto cleanup;
        }
+       cleanup_ma = true;
        ret = pthread_mutexattr_settype(&ma, PTHREAD_MUTEX_ERRORCHECK);
        if (ret != 0) {
-               goto cleanup_ma;
+               goto cleanup;
        }
        ret = pthread_mutexattr_setpshared(&ma, PTHREAD_PROCESS_SHARED);
        if (ret != 0) {
-               goto cleanup_ma;
+               goto cleanup;
        }
        ret = pthread_mutexattr_setrobust(&ma, PTHREAD_MUTEX_ROBUST);
        if (ret != 0) {
-               goto cleanup_ma;
+               goto cleanup;
        }
-       ret = pthread_mutex_init(m, &ma);
+       ret = pthread_mutex_init(ptr, &ma);
        if (ret != 0) {
-               goto cleanup_ma;
+               goto cleanup;
        }
-
-       /*
-        * Block SIGCHLD so we can atomically wait for it later with
-        * sigsuspend()
-        */
-       sigemptyset(&mask);
-       sigaddset(&mask, SIGCHLD);
-       ret = pthread_sigmask(SIG_BLOCK, &mask, &old_mask);
-       if (ret != 0) {
-               goto cleanup_m;
-       }
-       suspend_mask = old_mask;
-       sigdelset(&suspend_mask, SIGCHLD);
+       m = (pthread_mutex_t *)ptr;
 
        if (tdb_robust_mutex_setup_sigchild(tdb_robust_mutex_handler,
                        &tdb_robust_mutext_old_handler) == false) {
-               goto cleanup_sigmask;
+               goto cleanup;
        }
 
        tdb_robust_mutex_pid = fork();
+       saved_child_pid = tdb_robust_mutex_pid;
        if (tdb_robust_mutex_pid == 0) {
                size_t nwritten;
                close(pipe_down[1]);
@@ -867,7 +919,7 @@ _PUBLIC_ bool tdb_runtime_check_for_robust_mutexes(void)
                _exit(0);
        }
        if (tdb_robust_mutex_pid == -1) {
-               goto cleanup_sig_child;
+               goto cleanup;
        }
        close(pipe_down[0]);
        pipe_down[0] = -1;
@@ -876,7 +928,7 @@ _PUBLIC_ bool tdb_runtime_check_for_robust_mutexes(void)
 
        nread = read(pipe_up[0], &ret, sizeof(ret));
        if (nread != sizeof(ret)) {
-               goto cleanup_child;
+               goto cleanup;
        }
 
        ret = pthread_mutex_trylock(m);
@@ -884,74 +936,68 @@ _PUBLIC_ bool tdb_runtime_check_for_robust_mutexes(void)
                if (ret == 0) {
                        pthread_mutex_unlock(m);
                }
-               goto cleanup_child;
+               goto cleanup;
        }
 
        if (write(pipe_down[1], &c, 1) != 1) {
-               goto cleanup_child;
+               goto cleanup;
        }
 
        nread = read(pipe_up[0], &c, 1);
        if (nread != 0) {
-               goto cleanup_child;
+               goto cleanup;
        }
 
-       while (tdb_robust_mutex_pid > 0) {
-               ret = sigsuspend(&suspend_mask);
-               if (ret != -1 || errno != EINTR) {
-                       abort();
-               }
-       }
-       tdb_robust_mutex_setup_sigchild(tdb_robust_mutext_old_handler, NULL);
+       tdb_robust_mutex_wait_for_child(&saved_child_pid);
 
        ret = pthread_mutex_trylock(m);
        if (ret != EOWNERDEAD) {
                if (ret == 0) {
                        pthread_mutex_unlock(m);
                }
-               goto cleanup_sigmask;
+               goto cleanup;
        }
 
        ret = pthread_mutex_consistent(m);
        if (ret != 0) {
-               goto cleanup_sigmask;
+               goto cleanup;
        }
 
        ret = pthread_mutex_trylock(m);
-       if (ret != EDEADLK) {
+       if (ret != EDEADLK && ret != EBUSY) {
                pthread_mutex_unlock(m);
-               goto cleanup_sigmask;
+               goto cleanup;
        }
 
        ret = pthread_mutex_unlock(m);
        if (ret != 0) {
-               goto cleanup_sigmask;
+               goto cleanup;
        }
 
        tdb_mutex_locking_cached = true;
-       goto cleanup_sigmask;
 
-cleanup_child:
-       while (tdb_robust_mutex_pid > 0) {
-               kill(tdb_robust_mutex_pid, SIGKILL);
-               ret = sigsuspend(&suspend_mask);
-               if (ret != -1 || errno != EINTR) {
-                       abort();
-               }
-       }
+cleanup:
+       /*
+        * Note that we don't reset the signal handler we just reset
+        * tdb_robust_mutex_pid to -1. This is ok as this code path is only
+        * called once per process.
+        *
+        * Leaving our signal handler avoids races with other threads potentially
+        * setting up their SIGCHLD handlers.
+        *
+        * The worst thing that can happen is that the other newer signal
+        * handler will get the SIGCHLD signal for our child and/or reap the
+        * child with a wait() function. tdb_robust_mutex_wait_for_child()
+        * handles the case where waitpid returns ECHILD.
+        */
+       tdb_robust_mutex_wait_for_child(&saved_child_pid);
 
-cleanup_sig_child:
-       tdb_robust_mutex_setup_sigchild(tdb_robust_mutext_old_handler, NULL);
-cleanup_sigmask:
-       ret = pthread_sigmask(SIG_SETMASK, &old_mask, NULL);
-       if (ret != 0) {
-               abort();
+       if (m != NULL) {
+               pthread_mutex_destroy(m);
+       }
+       if (cleanup_ma) {
+               pthread_mutexattr_destroy(&ma);
        }
-cleanup_m:
-       pthread_mutex_destroy(m);
-cleanup_ma:
-       pthread_mutexattr_destroy(&ma);
-cleanup_pipe:
        if (pipe_down[0] != -1) {
                close(pipe_down[0]);
        }
@@ -964,8 +1010,9 @@ cleanup_pipe:
        if (pipe_up[1] != -1) {
                close(pipe_up[1]);
        }
-cleanup_mmap:
-       munmap(ptr, sizeof(pthread_mutex_t));
+       if (ptr != NULL) {
+               munmap(ptr, sizeof(pthread_mutex_t));
+       }
 
        return tdb_mutex_locking_cached;
 }