Fix shutdown hang in dict_stats , caused by MDEV-16264mariadb-10.5.0

dict_stats_shutdown() can hang, waiting for timer callback to finish.
This happens because locks the same mutex, which can also used inside
timer callback, within dict_stats_schedule() function.

Fix is to make dict_stats_schedule() use mutex.try_lock() instead of
mutex.lock().

In the unlikely case of simultaneous dict_stats_schedule() setting
different timer delays, now the first one would win, which is fine.
Important is that shutdown won't hang.

1 files changed, 10 insertions, 1 deletions

diff --git a/storage/innobase/dict/dict0stats_bg.cc b/storage/innobase/dict/dict0stats_bg.cc
index ec9fb4851d1..ee26b5bd302 100644
--- a/storage/innobase/dict/dict0stats_bg.cc
+++ b/storage/innobase/dict/dict0stats_bg.cc
@@ -436,7 +436,16 @@ void dict_stats_start()
 
 static void dict_stats_schedule(int ms)
 {
-  std::lock_guard<std::mutex> lk(dict_stats_mutex);
+  std::unique_lock<std::mutex> lk(dict_stats_mutex, std::defer_lock);
+  /*
+    Use try_lock() to avoid deadlock in dict_stats_shutdown(), which
+    uses dict_stats_mutex too. If there is simultaneous timer reschedule,
+    the first one will win, which is fine.
+  */
+  if (!lk.try_lock())
+  {
+    return;
+  }
   if (dict_stats_timer)
     dict_stats_timer->set_time(ms,0);
 }