Fix: perform statedump before replying to sessiond

[lttng-ust.git] / liblttng-ust / lttng-ust-comm.c

diff --git a/liblttng-ust/lttng-ust-comm.c b/liblttng-ust/lttng-ust-comm.c
index 1e4bec699b4f454c0d01eef2749f6a7204e5f89d..f6c1f9e9e8b7a53262d79284f4f5785872b258fe 100644 (file)
--- a/liblttng-ust/lttng-ust-comm.c
+++ b/liblttng-ust/lttng-ust-comm.c
@@ -370,6 +370,16 @@ void lttng_fixup_urcu_bp_tls(void)
        rcu_read_unlock();
 }
 
+void lttng_ust_fixup_tls(void)
+{
+       lttng_fixup_urcu_bp_tls();
+       lttng_fixup_ringbuffer_tls();
+       lttng_fixup_vtid_tls();
+       lttng_fixup_nest_count_tls();
+       lttng_fixup_procname_tls();
+       lttng_fixup_ust_mutex_nest_tls();
+}
+
 int lttng_get_notify_socket(void *owner)
 {
        struct sock_info *info = owner;
@@ -423,7 +433,7 @@ int setup_local_apps(void)
 }
 
 /*
- * Get notify_sock timeout, in ms.
+ * Get socket timeout, in ms.
  * -1: wait forever. 0: don't wait. >0: timeout, in ms.
  */
 static
@@ -437,15 +447,26 @@ long get_timeout(void)
        }
        if (str_timeout)
                constructor_delay_ms = strtol(str_timeout, NULL, 10);
+       /* All negative values are considered as "-1". */
+       if (constructor_delay_ms < -1)
+               constructor_delay_ms = -1;
        return constructor_delay_ms;
 }
 
+/* Timeout for notify socket send and recv. */
 static
 long get_notify_sock_timeout(void)
 {
        return get_timeout();
 }
 
+/* Timeout for connecting to cmd and notify sockets. */
+static
+long get_connect_sock_timeout(void)
+{
+       return get_timeout();
+}
+
 /*
  * Return values: -1: wait forever. 0: don't wait. 1: timeout wait.
  */
@@ -848,6 +869,21 @@ int handle_message(struct sock_info *sock_info,
                }
        }
        DBG("Return value: %d", lur.ret_val);
+
+       ust_unlock();
+
+       /*
+        * Performed delayed statedump operations outside of the UST
+        * lock. We need to take the dynamic loader lock before we take
+        * the UST lock internally within handle_pending_statedump().
+         */
+       handle_pending_statedump(sock_info);
+
+       if (ust_lock()) {
+               ret = -LTTNG_UST_ERR_EXITING;
+               goto error;
+       }
+
        ret = send_reply(sock, &lur);
        if (ret < 0) {
                DBG("error sending reply");
@@ -878,13 +914,6 @@ int handle_message(struct sock_info *sock_info,
 error:
        ust_unlock();
 
-       /*
-        * Performed delayed statedump operations outside of the UST
-        * lock. We need to take the dynamic loader lock before we take
-        * the UST lock internally within handle_pending_statedump().
-         */
-       handle_pending_statedump(sock_info);
-
        return ret;
 }
 
@@ -1215,6 +1244,8 @@ void *ust_listener_thread(void *arg)
        int sock, ret, prev_connect_failed = 0, has_waited = 0;
        long timeout;
 
+       lttng_ust_fixup_tls();
+
        /* Restart trying to connect to the session daemon */
 restart:
        if (prev_connect_failed) {
@@ -1228,8 +1259,9 @@ restart:
                         * deals with a killed or broken session daemon.
                         */
                        sleep(5);
+               } else {
+                       has_waited = 1;
                }
-               has_waited = 1;
                prev_connect_failed = 0;
        }
 
@@ -1258,7 +1290,8 @@ restart:
         * first connect registration message.
         */
        /* Connect cmd socket */
-       ret = ustcomm_connect_unix_sock(sock_info->sock_path);
+       ret = ustcomm_connect_unix_sock(sock_info->sock_path,
+               get_connect_sock_timeout());
        if (ret < 0) {
                DBG("Info: sessiond not accepting connections to %s apps socket", sock_info->name);
                prev_connect_failed = 1;
@@ -1314,7 +1347,8 @@ restart:
        ust_unlock();
 
        /* Connect notify socket */
-       ret = ustcomm_connect_unix_sock(sock_info->sock_path);
+       ret = ustcomm_connect_unix_sock(sock_info->sock_path,
+               get_connect_sock_timeout());
        if (ret < 0) {
                DBG("Info: sessiond not accepting connections to %s apps socket", sock_info->name);
                prev_connect_failed = 1;
@@ -1477,12 +1511,7 @@ void __attribute__((constructor)) lttng_ust_init(void)
         * to be the dynamic linker mutex) and ust_lock, taken within
         * the ust lock.
         */
-       lttng_fixup_urcu_bp_tls();
-       lttng_fixup_ringbuffer_tls();
-       lttng_fixup_vtid_tls();
-       lttng_fixup_nest_count_tls();
-       lttng_fixup_procname_tls();
-       lttng_fixup_ust_mutex_nest_tls();
+       lttng_ust_fixup_tls();
 
        /*
         * We want precise control over the order in which we construct
@@ -1510,7 +1539,9 @@ void __attribute__((constructor)) lttng_ust_init(void)
        timeout_mode = get_constructor_timeout(&constructor_timeout);
 
        ret = sem_init(&constructor_wait, 0, 0);
-       assert(!ret);
+       if (ret) {
+               PERROR("sem_init");
+       }
 
        ret = setup_local_apps();
        if (ret) {
@@ -1575,17 +1606,34 @@ void __attribute__((constructor)) lttng_ust_init(void)
                        ret = sem_timedwait(&constructor_wait,
                                        &constructor_timeout);
                } while (ret < 0 && errno == EINTR);
-               if (ret < 0 && errno == ETIMEDOUT) {
-                       ERR("Timed out waiting for lttng-sessiond");
-               } else {
-                       assert(!ret);
+               if (ret < 0) {
+                       switch (errno) {
+                       case ETIMEDOUT:
+                               ERR("Timed out waiting for lttng-sessiond");
+                               break;
+                       case EINVAL:
+                               PERROR("sem_timedwait");
+                               break;
+                       default:
+                               ERR("Unexpected error \"%s\" returned by sem_timedwait",
+                                       strerror(errno));
+                       }
                }
                break;
        case -1:/* wait forever */
                do {
                        ret = sem_wait(&constructor_wait);
                } while (ret < 0 && errno == EINTR);
-               assert(!ret);
+               if (ret < 0) {
+                       switch (errno) {
+                       case EINVAL:
+                               PERROR("sem_wait");
+                               break;
+                       default:
+                               ERR("Unexpected error \"%s\" returned by sem_wait",
+                                       strerror(errno));
+                       }
+               }
                break;
        case 0: /* no timeout */
                break;
@@ -1597,6 +1645,7 @@ void lttng_ust_cleanup(int exiting)
 {
        cleanup_sock_info(&global_apps, exiting);
        cleanup_sock_info(&local_apps, exiting);
+       local_apps.allowed = 0;
        /*
         * The teardown in this function all affect data structures
         * accessed under the UST lock by the listener thread. This
@@ -1693,6 +1742,9 @@ void ust_before_fork(sigset_t *save_sigset)
        sigset_t all_sigs;
        int ret;
 
+       /* Fixup lttng-ust TLS. */
+       lttng_ust_fixup_tls();
+
        if (URCU_TLS(lttng_ust_nest_count))
                return;
        /* Disable signals */
@@ -1747,11 +1799,11 @@ void ust_after_fork_child(sigset_t *restore_sigset)
 {
        if (URCU_TLS(lttng_ust_nest_count))
                return;
+       lttng_context_vtid_reset();
        DBG("process %d", getpid());
        /* Release urcu mutexes */
        rcu_bp_after_fork_child();
        lttng_ust_cleanup(0);
-       lttng_context_vtid_reset();
        /* Release mutexes and reenable signals */
        ust_after_fork_common(restore_sigset);
        lttng_ust_init();