[UR][L0] Fix event refcount bugs in L0 v1 adapter barrier handling

Fixes three issues in the Level Zero v1 adapter that caused
urEventWait to be called for an internal event crashes
and segfaults when mixing multiple in-order queues with
ext_oneapi_submit_barrier.

1. Fix insertBarrierIntoCmdList passing InterruptBasedEventsEnabled
   as IsMultiDevice to createEventAndAssociateQueue. Barrier events
   do not need multi-device visibility; pass false instead.

2. In urEventWait, replace die() with continue when hasExternalRefs()
   is false. A recycled event (RefCountExternal == 0) was already
   completed before recycling, so skipping the wait is safe.

3. In urEventRelease, use CleanupCompletedEvent (which is a no-op
   when CleanedUp is true) instead of calling urEventReleaseInternal
   directly. The old code double-released the internal refcount when
   CleanupEventListFromResetCmdList had already cleaned the event.

UR_L0_SERIALIZE=2 masked the race by forcing synchronous execution.
UR_L0_DISABLE_EVENTS_CACHING=1 turned the recycling into a delete,
escalating the bug from a stale-data read to a segfault.

Fixes: #21704

Signed-off-by: Zhang, Winston <winston.zhang@intel.com>

Author: winstonzhang-intel

unified-runtime/source/adapters/level_zero/event.cpp

@@ -199,9 +199,10 @@ ur_result_t urEnqueueEventsWaitWithBarrierExt(
       [&Queue](ur_command_list_ptr_t CmdList, ur_ze_event_list_t &EventWaitList,
                ur_event_handle_t &Event, bool IsInternal,
                bool InterruptBasedEventsEnabled) {
+        (void)InterruptBasedEventsEnabled;
         UR_CALL(createEventAndAssociateQueue(
             Queue, &Event, UR_COMMAND_EVENTS_WAIT_WITH_BARRIER, CmdList,
-            IsInternal, InterruptBasedEventsEnabled));
+            IsInternal, /* IsMultiDevice */ false));
 
         Event->WaitList = EventWaitList;
 
@@ -796,7 +797,7 @@ urEventWait(uint32_t NumEvents,
       //
       ur_event_handle_t_ *Event = ur_cast<ur_event_handle_t_ *>(e);
       if (!Event->hasExternalRefs())
-        die("urEventWait must not be called for an internal event");
+        continue;
 
       ze_event_handle_t ZeHostVisibleEvent;
       if (auto Res = Event->getOrCreateHostVisibleEvent(ZeHostVisibleEvent))
@@ -822,7 +823,7 @@ urEventWait(uint32_t NumEvents,
       {
         std::shared_lock<ur_shared_mutex> EventLock(Event->Mutex);
         if (!Event->hasExternalRefs())
-          die("urEventWait must not be called for an internal event");
+          continue;
 
         if (!Event->Completed) {
           auto HostVisibleEvent = Event->HostVisibleEvent;
@@ -894,15 +895,11 @@ urEventRelease(/** [in] handle of the event object */ ur_event_handle_t Event) {
   UR_CALL(urEventReleaseInternal(Event, &isEventDeleted));
   // If this is a Completed Event Wait Out Event, then we need to cleanup the
   // event at user release and not at the time of completion.
-  // If the event is labelled as completed and no additional references are
-  // removed, then we still need to decrement the event, but not mark as
-  // completed.
+  // Use CleanupCompletedEvent which is a no-op if the event was already
+  // cleaned up (e.g. by CleanupEventListFromResetCmdList), preventing a
+  // double-release of the internal reference count.
   if (isEventsWaitCompleted & !isEventDeleted) {
-    if (Event->CleanedUp) {
-      UR_CALL(urEventReleaseInternal(Event));
-    } else {
-      UR_CALL(CleanupCompletedEvent((Event), false, false));
-    }
+    UR_CALL(CleanupCompletedEvent((Event), false, false));
   }
 
   return UR_RESULT_SUCCESS;