Fix: thread exit vs futex wait/wakeup race

relayd_live_stop performs, in this order:

        CMM_STORE_SHARED(live_dispatch_thread_exit, 1);   [A]
        futex_nto1_wake(&viewer_conn_queue.futex);        [B]

whereas thread_dispatcher does:

   while (!CMM_LOAD_SHARED(live_dispatch_thread_exit)) {  [1]

     [...]
     futex_nto1_prepare(&viewer_conn_queue.futex);        [2]
     [...]
     futex_nto1_wait(&viewer_conn_queue.futex);           [3]

Unfortunately, on the following sequence:

[1] [A] [B] [2] [3]

thread_dispatcher will end up hanging.

We need to move the live_dispatch_thread_exit load between "prepare" and
"wait" to fix this.

There are similar scenarios with relay_thread_dispatcher, and the
session daemon thread_dispatch_ust_registration, which are also fixed
here.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

diff --git a/src/bin/lttng-relayd/live.c b/src/bin/lttng-relayd/live.c
index 61acafec2098ecd70a5b1889309992e858a0e3cb..f00270bea4c84ab4727047594443191e028965ef 100644 (file)
--- a/src/bin/lttng-relayd/live.c
+++ b/src/bin/lttng-relayd/live.c
@@ -654,12 +654,16 @@ void *thread_dispatcher(void *data)
 
        health_code_update();
 
-       while (!CMM_LOAD_SHARED(live_dispatch_thread_exit)) {
+       for (;;) {
                health_code_update();
 
                /* Atomically prepare the queue futex */
                futex_nto1_prepare(&viewer_conn_queue.futex);
 
+               if (CMM_LOAD_SHARED(live_dispatch_thread_exit)) {
+                       break;
+               }
+
                do {
                        health_code_update();
 

diff --git a/src/bin/lttng-relayd/main.c b/src/bin/lttng-relayd/main.c
index 2fcc60af00d9d3b29b9809865ed7747ab4dbde45..0eb8e28251f2451ab229dc41cf0e62726fddc7c5 100644 (file)
--- a/src/bin/lttng-relayd/main.c
+++ b/src/bin/lttng-relayd/main.c
@@ -977,12 +977,16 @@ static void *relay_thread_dispatcher(void *data)
 
        health_code_update();
 
-       while (!CMM_LOAD_SHARED(dispatch_thread_exit)) {
+       for (;;) {
                health_code_update();
 
                /* Atomically prepare the queue futex */
                futex_nto1_prepare(&relay_conn_queue.futex);
 
+               if (CMM_LOAD_SHARED(dispatch_thread_exit)) {
+                       break;
+               }
+
                do {
                        health_code_update();
 

diff --git a/src/bin/lttng-sessiond/main.c b/src/bin/lttng-sessiond/main.c
index 9b6f70cd08134acfa45afa1d970735d3d2ee6a4c..e282a9c7699845ae4a15e4a9216fa81f812f9020 100644 (file)
--- a/src/bin/lttng-sessiond/main.c
+++ b/src/bin/lttng-sessiond/main.c
@@ -1934,12 +1934,16 @@ static void *thread_dispatch_ust_registration(void *data)
 
        DBG("[thread] Dispatch UST command started");
 
-       while (!CMM_LOAD_SHARED(dispatch_thread_exit)) {
+       for (;;) {
                health_code_update();
 
                /* Atomically prepare the queue futex */
                futex_nto1_prepare(&ust_cmd_queue.futex);
 
+               if (CMM_LOAD_SHARED(dispatch_thread_exit)) {
+                       break;
+               }
+
                do {
                        struct ust_app *app = NULL;
                        ust_cmd = NULL;