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 cc88032dc99abac75d8b011d251b4eafe8493b1a..ffec5977e54984d99e72ae83fb2c79f3029481a7 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 56efaae3409748ba6134d72de26692aae6689649..9c96af127473a5080807b30df943cd8993e1910e 100644 (file)
--- a/src/bin/lttng-relayd/main.c
+++ b/src/bin/lttng-relayd/main.c
@@ -969,12 +969,16 @@ 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 8527ad04c8afdbbab57d4890694c68ee2f2571a6..50eead010a3e2e72b053f4c899a49a2a02f86042 100644 (file)
--- a/src/bin/lttng-sessiond/main.c
+++ b/src/bin/lttng-sessiond/main.c
@@ -1916,12 +1916,16 @@ 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;