The logic for synchronising host and guest time stamps uses pthreads and
mutexes. The guest pthread locks a mutex on the beginning of each
synchronization iteration and uses it to wait on a conditional variable.
When the synchronization is done and the thread exists, this mutex
should be unlocked.
fixes: 08b9d507 ("trace-cmd: Basic infrastructure for host - guest timestamp synchronization")
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=215808
Reported-by: Ryan <ycaibb@xxxxxxxxx>
Signed-off-by: Tzvetomir Stoyanov (VMware) <tz.stoyanov@xxxxxxxxx>
---
lib/trace-cmd/trace-timesync.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/lib/trace-cmd/trace-timesync.c b/lib/trace-cmd/trace-timesync.c
index f8ec15a3..90ffa6d0 100644
--- a/lib/trace-cmd/trace-timesync.c
+++ b/lib/trace-cmd/trace-timesync.c
@@ -807,8 +807,10 @@ static int tsync_with_guest(struct tracecmd_time_sync *tsync)
first = false;
pthread_barrier_wait(&tsync->first_sync);
}
- if (end || i < tsync->vcpu_count)
+ if (end || i < tsync->vcpu_count) {
+ pthread_mutex_unlock(&tsync->lock);
break;
+ }
if (tsync->loop_interval > 0) {
get_ts_loop_delay(&timeout, tsync->loop_interval);
ret = pthread_cond_timedwait(&tsync->cond, &tsync->lock, &timeout);
--
2.35.1
|