Fix for dual decoder in VCM receiver

In VCMReceiver::FrameForDecoding, one of the if-cases could sometimes
extract an incomplete frame without first copying the state to the
dual decoder.

Review URL: http://webrtc-codereview.appspot.com/328006

git-svn-id: http://webrtc.googlecode.com/svn/trunk@1221 4adac7df-926f-26a2-2b94-8c16560cd09d

src/modules/video_coding/main/source/receiver.cc

@@ -275,7 +275,19 @@ VCMReceiver::FrameForDecoding(WebRtc_UWord16 maxWaitTimeMs,
     {
         // No time to wait for a complete frame,
         // check if we have an incomplete
-        frame = _jitterBuffer.GetFrameForDecoding();
+        const bool dualReceiverEnabledAndPassive = (dualReceiver != NULL &&
+                                     dualReceiver->State() == kPassive &&
+                                     dualReceiver->NackMode() == kNackInfinite);
+        if (dualReceiverEnabledAndPassive &&
+            !_jitterBuffer.CompleteSequenceWithNextFrame())
+        {
+            // Jitter buffer state might get corrupt with this frame.
+            dualReceiver->CopyJitterBufferStateFromReceiver(*this);
+            frame = _jitterBuffer.GetFrameForDecoding();
+            assert(frame);
+        } else {
+            frame = _jitterBuffer.GetFrameForDecoding();
+        }
     }
     if (frame == NULL)
     {