Docs:Qutexes Update design but it's not reflective of impl

docs/design/qutexes.md

@@ -374,12 +374,17 @@ public:
 		 */
 		if (nRequiredLocks == 1)
 		{
+			bool ret=false;
+
 			if (tryingLockvoker == &queue.front())
 			{
 				currOwner = tryingLockvoker;
-				lock.release();
+				ret = true;
-				return true;
 			}
+
+			ret = false;
+			lock.release();
+			return ret;
 		}
 
 		auto rIt = queue.rbegin();
@@ -494,6 +499,22 @@ public:
 
 		lock.release();
 
+		/**	EXPLANATION:
+		 * Why should this never happen? Well, if we were at the front of the queue
+		 * and we failed to acquire the lock, we should have been rotated away from
+		 * the front. On the other hand, if we were not at the front of the queue
+		 * and we failed to acquire the lock, then we weren't at the front of the
+		 * queue to begin with.
+		 * The exception is if the queue has only one item in it.
+		 *
+		 * Hence there ought to be no way for the failedAcquirer to be at the front
+		 * of the queue at this point UNLESS the queue has only one item in it.
+		 */
+		if (newFront == failedAcquirer && nQItems > 1)
+		{
+			throw;
+		}
+
 		/**	EXPLANATION:
 		 * We should always awaken whoever is at the front of the queue, even if
 		 * we didn't rotate. Why? Consider this scenario:
@@ -521,7 +542,7 @@ public:
 		 * itself. This can happen if, for example a multi-locking lockvoker
 		 * is backing off of a qutex within which it's the only waiter.
 		 */
-		if (newFront != failedAcquirer) {
+		if (nQItems > 1) {
 			wakeUp(newFront);
 		}
 	}