Index: sys/sys/refcount.h
===================================================================
--- sys/sys/refcount.h
+++ sys/sys/refcount.h
@@ -50,7 +50,7 @@
 {
 
 	KASSERT(*count < UINT_MAX, ("refcount %p overflowed", count));
-	atomic_add_acq_int(count, 1);	
+	atomic_add_int(count, 1);
 }
 
 static __inline int
@@ -58,10 +58,22 @@
 {
 	u_int old;
 
-	/* XXX: Should this have a rel membar? */
 	old = atomic_fetchadd_int(count, -1);
 	KASSERT(old > 0, ("negative refcount %p", count));
-	return (old == 1);
+	if (old > 1)
+		return (0);
+
+	/*
+	 * Last reference.  Signal the user to call the desctructor.
+	 *
+	 * Ensure that the destructing code sees all updates.  The
+	 * constructor and updating consumers must have executed the
+	 * release operations, which are synchronized with this fence.
+	 * XXXKIB.  The recommendation there is to use
+	 * fetchadd_acq_rel(), but we do not have this atomic variant.
+	 */
+	atomic_thread_fence_acq();
+	return (1);
 }
 
 #endif	/* ! __SYS_REFCOUNT_H__ */