Fix Core deadlock by ensuring only one ZHA log queue handler thread is running at a time (#142568)

Ensure only one log queue handler is running at a time

Author: puddly

homeassistant/components/zha/helpers.py

@@ -514,6 +514,7 @@ def __init__(
         self._log_queue_handler.listener = logging.handlers.QueueListener(
             log_simple_queue, log_relay_handler
         )
+        self._log_queue_handler_count: int = 0
 
         self._unsubs: list[Callable[[], None]] = []
         self._unsubs.append(self.gateway.on_all_events(self._handle_event_protocol))
@@ -747,7 +748,10 @@ def async_enable_debug_mode(self, filterer: _LogFilterType | None = None) -> Non
         if filterer:
             self._log_queue_handler.addFilter(filterer)
 
-        if self._log_queue_handler.listener:
+        # Only start a new log queue handler if the old one is no longer running
+        self._log_queue_handler_count += 1
+
+        if self._log_queue_handler.listener and self._log_queue_handler_count == 1:
             self._log_queue_handler.listener.start()
 
         for logger_name in DEBUG_RELAY_LOGGERS:
@@ -763,7 +767,10 @@ def async_disable_debug_mode(self, filterer: _LogFilterType | None = None) -> No
         for logger_name in DEBUG_RELAY_LOGGERS:
             logging.getLogger(logger_name).removeHandler(self._log_queue_handler)
 
-        if self._log_queue_handler.listener:
+        # Only stop the log queue handler if nothing else is using it
+        self._log_queue_handler_count -= 1
+
+        if self._log_queue_handler.listener and self._log_queue_handler_count == 0:
             self._log_queue_handler.listener.stop()
 
         if filterer: