Question

I have the following .NET code. Most of it was written long before I was hired, and none of the original devs still work for us.

private void SendTCPMessage(string IpAddress, string Message)
    {
        ...

        //original code that fails because the Host entry produced 
        //has no elements in AddressList.
        //IPHostEntry remoteMachineInfo = Dns.GetHostEntry(IpAddress);

        //New code that fails when connecting
        IPHostEntry remoteMachineInfo;

        try
        {
            remoteMachineInfo = Dns.GetHostEntry(IpAddress);

            if (remoteMachineInfo.AddressList.Length == 0)
               remoteMachineInfo.AddressList = 
                  new[]
                     {
                        new IPAddress(
                           //Parse the string into the byte array needed by the constructor;
                           //I double-checked that the correct address is produced
                           IpAddress.Split('.')
                           .Select(s => byte.Parse(s))
                           .ToArray())
                     };
        }
        catch (Exception)
        {
            //caught and displayed in a status textbox
            throw new Exception(String.Format("Could not resolve or parse remote host {0} into valid IP address", IpAddress));
        }

        socketClient.Connect(remoteMachineInfo, 12345, ProtocolType.Tcp);

        ...
    }

The SocketClient code of note is as follows:

    public void Connect(IPHostEntry serverHostEntry, int serverPort, ProtocolType socketProtocol)
    {
        //this line was causing the original error; 
        //now AddressList always has at least one element.
        m_serverAddress = serverHostEntry.AddressList[0];
        m_serverPort = serverPort;
        m_socketProtocol = socketProtocol;
        Connect();
    }

    ...

    public void Connect()
    {
        try
        {
            Disconnect();
            SocketConnect();
        }
        catch (Exception exception) ...
    }

    ...

    private void SocketConnect()
    {
        try
        {
            if (SetupLocalSocket())
            {
                IPEndPoint serverEndpoint = new IPEndPoint(m_serverAddress, m_serverPort);

                //This line is the new point of failure
                socket.Connect(serverEndpoint); 

                ...
            }
            else
            {
                throw new Exception("Could not connect!");
            }
        }
        ...
        catch (SocketException se)
        {
            throw new Exception(se.Message);
        }
        ...
    }

    ...

    private bool SetupLocalSocket()
    {
        bool return_value = false;
        try
        {
            IPEndPoint myEndpoint = new IPEndPoint(m_localAddress, 0);
            socket = new Socket(myEndpoint.Address.AddressFamily, SocketType.Stream, m_socketProtocol);
            return_value = true;
        }
        catch (SocketException)
        {
            return_value = false;
        }
        catch (Exception)
        {
            return_value = false;
        }
        return return_value;
    }

When connecting to the endpoint within SocketConnect, I get a SocketException stating:

The system detected an invalid pointer address in attempting to use a pointer argument in a call.

Information online is a bit light on how to fix this. AFAICT, the address is parsing properly, and it’s retrieved properly once passed in to the SocketClient class. Honestly, I don’t know if this code has ever worked; I have never personally seen it do what it’s supposed to, and the functionality that uses all this was created for the benefit of a single client of ours, and has apparently not been functional since before I was hired.

I need to know what to look for to resolve the error. If it helps, the remote computer to which I am trying to establish a connection is on the remote side of a VPN tunnel, and we do have connectivity via other pieces of software we use.

Help?

Answer 1

Found it. The address used as the local endpoint for the socket, in SetupLocalSocket(), used a similarly naive method of getting the address; by resolving the local host and getting the first address. That first address, more often than not, is an IPv6 address, not the IPv4 address that was obviously expected. So, I had it look for the first IPv4 address in the list and use that as the endpoint, and it worked.