fix(plugins): fix skipped-frame health/force_change tracking, config validation, and lock-lifetime gaps in async updates

Addresses PR #407 review findings:
- display_controller: only clear force_change / record health success
  when display() actually ran this frame, not when the frame was skipped
  because the plugin's lock was busy (a skip must preserve a pending
  mode-switch force_clear).
- plugin_manager: replace bool() coercion of synchronous_updates with
  explicit isinstance validation of plugin_system/synchronous_updates,
  failing safe to synchronous mode (with a logged reason) on malformed
  config instead of silently defaulting to async.
- plugin_manager: _update_worker_loop now acquires the plugin lock before
  looking up its instance and re-checks under the lock, so an unloaded
  plugin's lifecycle state is never resurrected to ENABLED.
- plugin_manager + display_controller: move lock ownership (and, for
  updates, RUNNING/pending lifecycle bookkeeping) into the actual update()/
  display() call itself rather than the timeout-wrapped caller, so the
  lock stays held for the real operation's duration even after
  PluginExecutor's own join(timeout) elapses and a lingering daemon thread
  keeps running in the background.
- DisplayController.cleanup() now stops the update worker before tearing
  down display/cache resources; stop_update_worker() logs when the join
  times out instead of failing silently.
- test_async_plugin_updates: rewrite test_unloaded_while_queued_is_harmless
  to exercise the public unload_plugin() lifecycle (via a deterministic
  blocker) instead of deleting pm.plugins directly, and add a regression
  test proving the plugin lock stays held through PluginExecutor's own
  timeout while the real update() call is still running.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KEZK1P1Q1fu5pcuVrkrCFZ

src/display_controller.py

@@ -23,6 +23,8 @@ Entry point: :func:`main` — instantiates :class:`DisplayController` and calls
 import time
 import os
 import json
+import threading
+import types
 from contextlib import contextmanager
 from pathlib import Path
 from typing import Dict, Any, List, Optional, Callable
@@ -1851,6 +1853,7 @@ class DisplayController:
                     plugin_id = getattr(manager_to_display, 'plugin_id', active_mode)
                     try:
                         logger.debug(f"Calling display() for {active_mode} with force_clear={self.force_change}")
+                        can_display = False
                         if hasattr(manager_to_display, 'display'):
                             # Opt #1: look up (or compute once) whether display() accepts display_mode
                             _cache_key = plugin_id
@@ -1861,30 +1864,79 @@ class DisplayController:
                                 )
                             _accepts_display_mode = self._plugin_accepts_display_mode[_cache_key]
 
-                            with self._display_lock_or_skip(plugin_id) as can_display:
+                            pm = self.plugin_manager
-                                if not can_display:
+                            display_lock = None
-                                    # update() in flight on the worker — hold
+                            can_display = True
-                                    # the last frame; not a plugin failure
+                            if pm and hasattr(pm, 'get_plugin_lock'):
-                                    result = True
+                                display_lock = pm.get_plugin_lock(plugin_id)
-                                elif self.plugin_manager and hasattr(self.plugin_manager, 'plugin_executor'):
+                                can_display = display_lock.acquire(blocking=False)
-                                    # Use PluginExecutor for safe execution with timeout
+
+                            if not can_display:
+                                # update() in flight on the worker — hold
+                                # the last frame; not a plugin failure
+                                result = True
+                            elif pm and hasattr(pm, 'plugin_executor'):
+                                # PluginExecutor's own thread.join(timeout) can
+                                # return before the real display() call
+                                # finishes (a lingering daemon thread keeps
+                                # running it) -- so the lock is released from
+                                # inside the wrapped call itself, whichever
+                                # thread actually finishes it, rather than
+                                # here when this dispatch merely returns.
+                                release_guard = threading.Lock()
+                                released = {'done': False}
+
+                                def _release_display_lock():
+                                    with release_guard:
+                                        if released['done']:
+                                            return
+                                        released['done'] = True
+                                    if display_lock is not None:
+                                        display_lock.release()
+
+                                if _accepts_display_mode:
+                                    def _display_target(display_mode=None, force_clear=False):
+                                        try:
+                                            return manager_to_display.display(
+                                                display_mode=display_mode, force_clear=force_clear)
+                                        finally:
+                                            _release_display_lock()
+                                else:
+                                    def _display_target(force_clear=False):
+                                        try:
+                                            return manager_to_display.display(force_clear=force_clear)
+                                        finally:
+                                            _release_display_lock()
+
+                                try:
                                     result = self.plugin_manager.plugin_executor.execute_display(
-                                        manager_to_display,
+                                        types.SimpleNamespace(display=_display_target),
                                         plugin_id,
                                         force_clear=self.force_change,
                                         display_mode=active_mode if _accepts_display_mode else None
                                     )
-                                    # execute_display returns bool, convert to expected format
+                                except Exception:  # pragma: no cover - defensive;
-                                    if result:
+                                    # execute_display catches everything
-                                        result = True  # Success
+                                    # internally, but guarantee the lock is
-                                    else:
+                                    # never leaked if something unexpected
-                                        result = False  # Failed
+                                    # slips through.
+                                    _release_display_lock()
+                                    raise
+                                # execute_display returns bool, convert to expected format
+                                if result:
+                                    result = True  # Success
                                 else:
-                                    # Fallback to direct call if executor not available
+                                    result = False  # Failed
+                            else:
+                                # Fallback to direct call if executor not available
+                                try:
                                     if _accepts_display_mode:
                                         result = manager_to_display.display(display_mode=active_mode, force_clear=self.force_change)
                                     else:
                                         result = manager_to_display.display(force_clear=self.force_change)
+                                finally:
+                                    if display_lock is not None:
+                                        display_lock.release()
                             
                             logger.debug(f"display() returned: {result} (type: {type(result)})")
                             # Check if display() returned a boolean (new behavior)
@@ -1893,11 +1945,15 @@ class DisplayController:
                                 if not display_result:
                                     logger.info(f"Plugin {plugin_id} display() returned False for mode {active_mode}")
                         
-                        # Record success if display completed without exception
+                        # Record success only when display() actually ran this
-                        if self.plugin_manager and hasattr(self.plugin_manager, 'health_tracker') and self.plugin_manager.health_tracker:
+                        # frame -- a skipped frame (lock busy) held the last
-                            self.plugin_manager.health_tracker.record_success(plugin_id)
+                        # frame, not a real success, and must not clear
-                        
+                        # force_change or the pending mode-switch clear will
-                        self.force_change = False
+                        # be lost when display() finally does run.
+                        if can_display:
+                            if self.plugin_manager and hasattr(self.plugin_manager, 'health_tracker') and self.plugin_manager.health_tracker:
+                                self.plugin_manager.health_tracker.record_success(plugin_id)
+                            self.force_change = False
                     except Exception as exc:  # pylint: disable=broad-except
                         logger.exception("Error displaying %s", self.current_display_mode)
                         # Record failure
@@ -2754,6 +2810,14 @@ class DisplayController:
 
     def cleanup(self):
         """Clean up resources."""
+        # Stop the async update worker first so no in-flight update() call
+        # is still touching display/cache-backed resources while they're
+        # torn down below.
+        if self.plugin_manager and hasattr(self.plugin_manager, 'stop_update_worker'):
+            try:
+                self.plugin_manager.stop_update_worker()
+            except Exception as e:
+                logger.warning("Error stopping plugin update worker: %s", e)
         # Shutdown config service if it exists
         if hasattr(self, 'config_service'):
             try:

src/plugin_system/plugin_manager.py

@@ -111,13 +111,31 @@ class PluginManager:
         self._plugin_locks_guard = threading.Lock()
         self._update_worker: Optional[threading.Thread] = None
         self._synchronous_updates = False
-        try:
+        if self.config_manager is not None:
-            if self.config_manager is not None:
+            try:
                 cfg = self.config_manager.get_config() or {}
-                self._synchronous_updates = bool(
+            except (OSError, ValueError) as exc:
-                    cfg.get('plugin_system', {}).get('synchronous_updates', False))
+                self.logger.warning(
-        except Exception:
+                    "Could not load config to check plugin_system.synchronous_updates "
-            self._synchronous_updates = False
+                    "(%s: %s); defaulting to synchronous updates", type(exc).__name__, exc)
+                self._synchronous_updates = True
+            else:
+                plugin_system_cfg = cfg.get('plugin_system', {})
+                if not isinstance(plugin_system_cfg, dict):
+                    self.logger.warning(
+                        "config plugin_system must be a mapping, got %s; "
+                        "defaulting to synchronous updates",
+                        type(plugin_system_cfg).__name__)
+                    self._synchronous_updates = True
+                else:
+                    sync_value = plugin_system_cfg.get('synchronous_updates', False)
+                    if not isinstance(sync_value, bool):
+                        self.logger.warning(
+                            "config plugin_system.synchronous_updates must be a boolean, "
+                            "got %r; defaulting to synchronous updates", sync_value)
+                        self._synchronous_updates = True
+                    else:
+                        self._synchronous_updates = sync_value
         
         # Ensure plugins directory exists with proper permissions
         try:
@@ -804,84 +822,145 @@ class PluginManager:
         self._update_worker.start()
 
     def _update_worker_loop(self) -> None:
-        """Single worker: serializes all plugin updates (matching the old
+        """Single worker: dispatches queued updates off the render thread
-        inline behavior — no thundering herd of concurrent fetches), off the
+        (matching the old inline behavior — no thundering herd of
-        render thread."""
+        concurrent fetches).
+
+        The plugin's lock is acquired here, before its instance is looked
+        up, and the instance is re-fetched under the lock — a concurrent
+        unload_plugin() can't leave this loop about to run update() on an
+        instance that's already been torn down. The lock — and RUNNING/
+        pending lifecycle state — is released by the update itself once the
+        real update() call genuinely finishes (see _execute_update_now),
+        which can be after this dispatch returns if PluginExecutor's own
+        timeout elapses first.
+        """
         while True:
             item = self._update_queue.get()
             if item is None:  # shutdown sentinel
                 return
             plugin_id, scheduled_time = item
-            try:
+            lock = self.get_plugin_lock(plugin_id)
-                plugin_instance = self.plugins.get(plugin_id)
+            lock.acquire()
-                if plugin_instance is None:  # unloaded while queued
+            plugin_instance = self.plugins.get(plugin_id)
-                    self.state_manager.set_state(plugin_id, PluginState.ENABLED)
+            if plugin_instance is None:  # unloaded while queued; its
-                    continue
+                # lifecycle state was already cleared by unload_plugin —
-                lock = self.get_plugin_lock(plugin_id)
+                # leave it alone rather than resurrecting it to ENABLED
-                with lock:
+                lock.release()
-                    self._execute_update_now(plugin_id, plugin_instance,
-                                             scheduled_time)
-            except Exception:  # pylint: disable=broad-except
-                self.logger.exception("update worker: unexpected error for %s",
-                                      plugin_id)
-            finally:
                 with self._pending_lock:
                     self._pending_updates.discard(plugin_id)
+                continue
+            try:
+                self._execute_update_now(plugin_id, plugin_instance,
+                                         scheduled_time, lock=lock)
+            except Exception:  # pylint: disable=broad-except
+                # _execute_update_now guarantees the lock/pending bookkeeping
+                # is released via its own _finish() before returning or
+                # raising; this is a last-resort log only.
+                self.logger.exception("update worker: unexpected error for %s",
+                                      plugin_id)
 
     def stop_update_worker(self, timeout: float = 5.0) -> None:
         """Signal the worker to exit (used by cleanup; thread is a daemon)."""
         if self._update_worker is not None and self._update_worker.is_alive():
             self._update_queue.put(None)
             self._update_worker.join(timeout=timeout)
+            if self._update_worker.is_alive():
+                self.logger.warning(
+                    "Update worker did not stop within %.1fs; it is a daemon "
+                    "thread and will be abandoned on shutdown", timeout)
 
     def _execute_update_now(self, plugin_id: str, plugin_instance: Any,
-                            scheduled_time: float) -> None:
+                            scheduled_time: float,
-        """The (unchanged) update execution block: executor + bookkeeping.
+                            lock: Optional[threading.Lock] = None) -> None:
+        """Execute a plugin's update() via PluginExecutor, then bookkeep.
 
-        Caller is responsible for having set RUNNING state and, on the async
+        Caller is responsible for having set RUNNING state.
-        path, for holding the plugin's lock. plugin_executor's internal
+
-        thread.join(timeout) blocks only the calling thread; on timeout the
+        On the synchronous path (``lock=None``) this is the original,
-        lingering daemon update-thread keeps running unkillable — exactly the
+        unchanged inline behavior. On the async worker path, PluginExecutor's
-        pre-async behavior, documented here so nobody "fixes" it into a
+        internal thread.join(timeout) blocks only the calling thread -- on
-        thread leak hunt.
+        timeout the lingering daemon update-thread keeps running the real
+        plugin.update() call unkillable in the background. So that the
+        plugin's lock (and its RUNNING/pending lifecycle state) stays held
+        for that real duration rather than just this bounded wait, ownership
+        of both is carried by the wrapped update callable itself, released
+        from whichever thread actually finishes it -- see _finish() below.
         """
-        try:
+        finish_guard = threading.Lock()
-            # Use PluginExecutor for safe execution
+        finished = {'done': False}
-            success = False
-            if self.resource_monitor:
-                # If resource monitor exists, wrap the call
-                def monitored_update():
-                    self.resource_monitor.monitor_call(plugin_id, plugin_instance.update)
-                # SimpleNamespace stores `update` as an *instance*
-                # attribute, so attribute lookup returns the plain
-                # function object as-is. A dynamically-built class
-                # (`type(..., {'update': monitored_update})`) instead
-                # stores it as a *class* attribute, which the
-                # descriptor protocol turns into a bound method on
-                # access -- silently prepending the instance as an
-                # implicit first argument to a function that takes
-                # none, raising "monitored_update() takes 0
-                # positional arguments but 1 was given" on every call.
-                success = self.plugin_executor.execute_update(
-                    types.SimpleNamespace(update=monitored_update),
-                    plugin_id
-                )
-            else:
-                success = self.plugin_executor.execute_update(plugin_instance, plugin_id)
 
-            if success:
+        def _finish(success: bool, exc: Optional[Exception] = None) -> None:
-                self.plugin_last_update[plugin_id] = scheduled_time
+            with finish_guard:
-                self.state_manager.record_update(plugin_id)
+                if finished['done']:
-                # Update state back to ENABLED
+                    return
-                self.state_manager.set_state(plugin_id, PluginState.ENABLED)
+                finished['done'] = True
-                # Record success
+            try:
-                if self.health_tracker:
+                if success:
-                    self.health_tracker.record_success(plugin_id)
+                    self.plugin_last_update[plugin_id] = scheduled_time
+                    self.state_manager.record_update(plugin_id)
+                    self.state_manager.set_state(plugin_id, PluginState.ENABLED)
+                    if self.health_tracker:
+                        self.health_tracker.record_success(plugin_id)
+                else:
+                    self._record_update_failure(plugin_id, exc=exc)
+            finally:
+                if lock is not None:
+                    lock.release()
+                    with self._pending_lock:
+                        self._pending_updates.discard(plugin_id)
+
+        if lock is None:
+            # Synchronous / no-lock path: unchanged behavior.
+            try:
+                if self.resource_monitor:
+                    def monitored_update():
+                        self.resource_monitor.monitor_call(plugin_id, plugin_instance.update)
+                    # SimpleNamespace stores `update` as an *instance*
+                    # attribute, so attribute lookup returns the plain
+                    # function object as-is. A dynamically-built class
+                    # (`type(..., {'update': monitored_update})`) instead
+                    # stores it as a *class* attribute, which the
+                    # descriptor protocol turns into a bound method on
+                    # access -- silently prepending the instance as an
+                    # implicit first argument to a function that takes
+                    # none, raising "monitored_update() takes 0
+                    # positional arguments but 1 was given" on every call.
+                    success = self.plugin_executor.execute_update(
+                        types.SimpleNamespace(update=monitored_update),
+                        plugin_id
+                    )
+                else:
+                    success = self.plugin_executor.execute_update(plugin_instance, plugin_id)
+                _finish(success)
+            except Exception as exc:  # pylint: disable=broad-except
+                self.logger.exception("Error updating plugin %s: %s", plugin_id, exc)
+                _finish(False, exc=exc)
+            return
+
+        # Async worker path: the real update() call -- through the resource
+        # monitor, if configured -- owns finishing the lock/lifecycle
+        # bookkeeping, from whichever thread actually runs it to completion.
+        def _target_update() -> None:
+            try:
+                if self.resource_monitor:
+                    self.resource_monitor.monitor_call(plugin_id, plugin_instance.update)
+                else:
+                    plugin_instance.update()
+            except Exception as exc:
+                _finish(False, exc=exc)
+                raise
             else:
-                self._record_update_failure(plugin_id)
+                _finish(True)
-        except Exception as exc:  # pylint: disable=broad-except
+
-            self.logger.exception("Error updating plugin %s: %s", plugin_id, exc)
+        try:
-            self._record_update_failure(plugin_id, exc=exc)
+            self.plugin_executor.execute_update(
+                types.SimpleNamespace(update=_target_update), plugin_id)
+        except Exception as exc:  # pragma: no cover - defensive; execute_update
+            # catches everything internally, but guarantee _finish still
+            # runs (releasing the lock) if something unexpected slips through.
+            self.logger.exception("Unexpected error dispatching update for %s: %s", plugin_id, exc)
+            _finish(False, exc=exc)
 
     def update_all_plugins(self) -> None:
         """

test/test_async_plugin_updates.py

@@ -4,8 +4,9 @@ The invariants that keep this change safe:
 1. run_scheduled_updates returns immediately — a slow update() can never
    again freeze the render loop (the original defect: 30s scroll freezes).
 2. A plugin's update() and display() are NEVER concurrent — the per-plugin
-   lock makes the old implicit no-overlap guarantee explicit (and, unlike
+   lock makes the old implicit no-overlap guarantee explicit, and it stays
-   before, holds it across the post-timeout window too).
+   held through PluginExecutor's own timeout: a lingering, still-running
+   update() keeps the lock even after PluginExecutor gives up waiting on it.
 3. Failure/timeout bookkeeping is unchanged (same executor, same
    _record_update_failure path, same last-update stamping).
 4. The kill switch (plugin_system.synchronous_updates) restores the
@@ -138,6 +139,40 @@ class TestUpdateDisplayExclusion:
         assert plugin.display_calls > 10
         assert plugin.overlap_detected is False
 
+    def test_lock_held_through_timeout_until_real_update_finishes(self, pm):
+        """PluginExecutor's join(timeout) can return before the real
+        update() call does -- the lingering daemon thread keeps running it.
+        The plugin's lock must stay held for that whole real duration, not
+        just PluginExecutor's bounded wait, or display() could run
+        concurrently with a still-executing update()."""
+        plugin = SlowPlugin(update_seconds=0.3)
+        plugin_id = _install(pm, plugin)
+        pm.plugin_executor.default_timeout = 0.05  # times out well before
+                                                    # update_seconds elapses
+
+        pm.run_scheduled_updates()
+
+        deadline = time.monotonic() + 2
+        while not plugin.in_update and time.monotonic() < deadline:
+            time.sleep(0.01)
+        assert plugin.in_update is True
+
+        # PluginExecutor's own timeout has now elapsed, but the real
+        # update() (0.3s) is still running in its lingering daemon thread.
+        time.sleep(0.15)
+        assert plugin.in_update is True, "test setup: update should still be running"
+        lock = pm.get_plugin_lock(plugin_id)
+        assert lock.acquire(blocking=False) is False, \
+            "lock must stay held through PluginExecutor's timeout while the real update() runs"
+
+        # Once the real update() genuinely finishes, the lock is released.
+        deadline = time.monotonic() + 2
+        while plugin.in_update and time.monotonic() < deadline:
+            time.sleep(0.02)
+        time.sleep(0.1)
+        assert lock.acquire(blocking=False) is True
+        lock.release()
+
     def test_state_returns_to_enabled_after_update(self, pm):
         """RUNNING is set at enqueue (blocks re-entry via can_execute) and
         must return to an executable state once the update finishes."""
@@ -175,11 +210,49 @@ class TestFailurePaths:
         pm.get_plugin_lock(plugin_id).release()
 
     def test_unloaded_while_queued_is_harmless(self, pm):
-        plugin_id = _install(pm, SlowPlugin(update_seconds=0.1))
+        """Exercise the public unload_plugin() lifecycle rather than
-        pm._enqueue_update(plugin_id, time.time())
+        deleting pm.plugins directly: queue the target's update behind a
-        del pm.plugins[plugin_id]
+        deterministic blocker (occupying the single worker), unload the
-        time.sleep(0.3)  # worker drains the item without crashing
+        target while its item still sits queued, then release the blocker
-        assert pm._update_worker is None or pm._update_worker.is_alive()
+        and confirm the target's update never ran and its state stayed
+        unloaded rather than being resurrected to ENABLED."""
+        blocker_event = threading.Event()
+
+        class BlockerPlugin(SlowPlugin):
+            def update(self):
+                self.update_calls += 1
+                blocker_event.wait(timeout=5)
+                return True
+
+        blocker_id = _install(pm, BlockerPlugin(), plugin_id="blocker-plugin")
+        target = SlowPlugin(update_seconds=0.05)
+        target_id = _install(pm, target, plugin_id="slow-plugin")
+
+        # Dispatch the blocker first so it occupies the single worker
+        # thread, then enqueue the target behind it -- deterministically
+        # queued, not yet started.
+        pm._enqueue_update(blocker_id, time.time())
+        deadline = time.monotonic() + 2
+        while pm.plugins[blocker_id].update_calls == 0 and time.monotonic() < deadline:
+            time.sleep(0.01)
+        assert pm.plugins[blocker_id].update_calls == 1
+
+        pm._enqueue_update(target_id, time.time())
+        assert target_id in pm._pending_updates
+
+        assert pm.unload_plugin(target_id) is True
+        assert target_id not in pm.plugins
+
+        blocker_event.set()  # let the blocker finish; worker moves on to
+                              # the target's queued item
+
+        deadline = time.monotonic() + 3
+        while target_id in pm._pending_updates and time.monotonic() < deadline:
+            time.sleep(0.02)
+
+        assert target.update_calls == 0, "update() must not run for an unloaded plugin"
+        assert target_id not in pm._pending_updates
+        assert pm.state_manager.get_state(target_id) == PluginState.UNLOADED
 
 
 class TestKillSwitch: