fix: ensure first NGAP message is NGSetup to prevent crash

[wiwi878]

This PR is for issue#768

[c9274326]

Hi @wiwi878, This check can be removed. If we properly refactor the logic above, when AmfRanFindByConn() successfully finds a ran, it must have already completed NGSetup, so RanId should never be nil.

[c9274326]

Current Problem: At line 25, we execute ran = amfSelf.NewAmfRan(conn) to create a new AmfRan, but we don't check if len(msg) == 0 until line 33.

This causes an issue: when a new connection is established and immediately closed (msg is empty), the program will first create an AmfRan at L25, then delete it at L33-36. This is unnecessary resource overhead and could potentially be exploited for DoS attacks.

Suggested refactoring (after L15 amfSelf := context.GetSelf()):

func Dispatch(conn net.Conn, msg []byte) {
    amfSelf := context.GetSelf()

    // Step 1: Handle connection close (len(msg) == 0) first
    if len(msg) == 0 {
        ran, ok := amfSelf.AmfRanFindByConn(conn)
        if !ok {
            // Connection closed before NGSetup, no RAN context exists
            logger.NgapLog.Warnf("Connection closed before NGSetup: %s", conn.RemoteAddr())
            return
        }
        // Registered RAN closing connection
        ran.Log.Infof("RAN close the connection.")
        ran.Remove()
        return
    }

    // Step 2: Normal message processing - find or create AmfRan
    ran, ok := amfSelf.AmfRanFindByConn(conn)
    if !ok {
        addr := conn.RemoteAddr()
        if addr == nil {
            logger.NgapLog.Warn("Addr of new NG connection is nil")
            return
        }
        logger.NgapLog.Infof("Create a new NG connection for: %s", addr.String())
        ran = amfSelf.NewAmfRan(conn)
    }

    // Step 3: Decode and dispatch...
    pdu, err := ngap.Decoder(msg)
    // ...
    dispatchMain(ran, pdu)
}

[c9274326]

@Alonza0314 LGTM!