Fix panics in the mtcp-client

If the write to reportChan is deferred, and reportChang has already been
closed, then the resulting panic won't be recovered by the deferred
recover-function.

pkg/cla/mtcp/client.go

@@ -116,41 +116,42 @@ func (client *MTCPClient) Send(bndl bpv7.Bundle) (err error) {
 			err = fmt.Errorf("MTCPClient.Send: %v", r)
 		}
 
-		// In case of an error, report our failure upstream
-		if err != nil {
-			client.reportChan <- cla.NewConvergencePeerDisappeared(client, client.GetPeerEndpointID())
-		}
+		client.mutex.Unlock()
 	}()
 
 	client.mutex.Lock()
-	defer client.mutex.Unlock()
 
 	connWriter := bufio.NewWriter(client.conn)
 
 	buff := new(bytes.Buffer)
 	if cborErr := cboring.Marshal(&bndl, buff); cborErr != nil {
 		err = cborErr
+		client.reportChan <- cla.NewConvergencePeerDisappeared(client, client.GetPeerEndpointID())
 		return
 	}
 
 	if bsErr := cboring.WriteByteStringLen(uint64(buff.Len()), connWriter); bsErr != nil {
 		err = bsErr
+		client.reportChan <- cla.NewConvergencePeerDisappeared(client, client.GetPeerEndpointID())
 		return
 	}
 
 	if _, plErr := buff.WriteTo(connWriter); plErr != nil {
 		err = plErr
+		client.reportChan <- cla.NewConvergencePeerDisappeared(client, client.GetPeerEndpointID())
 		return
 	}
 
 	if flushErr := connWriter.Flush(); flushErr != nil {
 		err = flushErr
+		client.reportChan <- cla.NewConvergencePeerDisappeared(client, client.GetPeerEndpointID())
 		return
 	}
 
 	// Check if the connection is still alive with an empty, unbuffered packet
 	if probeErr := cboring.WriteByteStringLen(0, client.conn); probeErr != nil {
 		err = probeErr
+		client.reportChan <- cla.NewConvergencePeerDisappeared(client, client.GetPeerEndpointID())
 		return
 	}