SSLEngine

abstract class SSLEngine

A class which enables secure communications using protocols such as the Secure Sockets Layer (SSL) or IETF RFC 2246 "Transport Layer Security" (TLS) protocols, but is transport independent.

The secure communications modes include:

• Integrity Protection. SSL/TLS protects against modification of messages by an active wiretapper.
• Authentication. In most modes, SSL/TLS provides peer authentication. Servers are usually authenticated, and clients may be authenticated as requested by servers.
• Confidentiality (Privacy Protection). In most modes, SSL/TLS encrypts data being sent between client and server. This protects the confidentiality of data, so that passive wiretappers won't see sensitive data such as financial information or personal information of many kinds.

The cipher suite used is established by a negotiation process called "handshaking". The goal of this process is to create or rejoin a "session", which may protect many connections over time. After handshaking has completed, you can access session attributes by using the getSession() method.

The SSLSocket class provides much of the same security functionality, but all of the inbound and outbound data is automatically transported using the underlying Socket, which by design uses a blocking model. While this is appropriate for many applications, this model does not provide the scalability required by large servers.

The primary distinction of an SSLEngine is that it operates on inbound and outbound byte streams, independent of the transport mechanism. It is the responsibility of the SSLEngine user to arrange for reliable I/O transport to the peer. By separating the SSL/TLS abstraction from the I/O transport mechanism, the SSLEngine can be used for a wide variety of I/O types, such as non-blocking I/O (polling), selectable non-blocking I/O, Socket and the traditional Input/OutputStreams, local ByteBuffers or byte arrays, future asynchronous I/O models , and so on.

At a high level, the SSLEngine appears thus:

app data
 
                 |           ^
                 |     |     |
                 v     |     |
            +----+-----|-----+----+
            |          |          |
            |       SSL|Engine    |
    wrap()  |          |          |  unwrap()
            | OUTBOUND | INBOUND  |
            |          |          |
            +----+-----|-----+----+
                 |     |     ^
                 |     |     |
                 v           |
 
                    net data
  

(In the context of an SSLEngine, the term "handshake data" is taken to mean any data exchanged to establish and control a secure connection. Handshake data includes the SSL/TLS messages "alert", "change_cipher_spec," and "handshake.")

There are five distinct phases to an SSLEngine.

1. Creation - The SSLEngine has been created and initialized, but has not yet been used. During this phase, an application may set any SSLEngine-specific settings (enabled cipher suites, whether the SSLEngine should handshake in client or server mode, and so on). Once handshaking has begun, though, any new settings (except client/server mode, see below) will be used for the next handshake.
2. Initial Handshake - The initial handshake is a procedure by which the two peers exchange communication parameters until an SSLSession is established. Application data can not be sent during this phase.
3. Application Data - Once the communication parameters have been established and the handshake is complete, application data may flow through the SSLEngine. Outbound application messages are encrypted and integrity protected, and inbound messages reverse the process.
4. Rehandshaking - Either side may request a renegotiation of the session at any time during the Application Data phase. New handshaking data can be intermixed among the application data. Before starting the rehandshake phase, the application may reset the SSL/TLS communication parameters such as the list of enabled ciphersuites and whether to use client authentication, but can not change between client/server modes. As before, once handshaking has begun, any new SSLEngine configuration settings will not be used until the next handshake.
5. Closure - When the connection is no longer needed, the application should close the SSLEngine and should send/receive any remaining messages to the peer before closing the underlying transport mechanism. Once an engine is closed, it is not reusable: a new SSLEngine must be created.

Data moves through the engine by calling wrap() or unwrap() on outbound or inbound data, respectively. Depending on the state of the SSLEngine, a wrap() call may consume application data from the source buffer and may produce network data in the destination buffer. The outbound data may contain application and/or handshake data. A call to unwrap() will examine the source buffer and may advance the handshake if the data is handshaking information, or may place application data in the destination buffer if the data is application. The state of the underlying SSL/TLS algorithm will determine when data is consumed and produced.

Calls to wrap() and unwrap() return an SSLEngineResult which indicates the status of the operation, and (optionally) how to interact with the engine to make progress.

The SSLEngine produces/consumes complete SSL/TLS packets only, and does not store application data internally between calls to wrap()/unwrap(). Thus input and output ByteBuffers must be sized appropriately to hold the maximum record that can be produced. Calls to SSLSession.getPacketBufferSize() and SSLSession.getApplicationBufferSize() should be used to determine the appropriate buffer sizes. The size of the outbound application data buffer generally does not matter. If buffer conditions do not allow for the proper consumption/production of data, the application must determine (via SSLEngineResult) and correct the problem, and then try the call again.

For example, unwrap() will return a SSLEngineResult.Status.BUFFER_OVERFLOW result if the engine determines that there is not enough destination buffer space available. Applications should call SSLSession.getApplicationBufferSize() and compare that value with the space available in the destination buffer, enlarging the buffer if necessary. Similarly, if unwrap() were to return a SSLEngineResult.Status.BUFFER_UNDERFLOW, the application should call SSLSession.getPacketBufferSize() to ensure that the source buffer has enough room to hold a record (enlarging if necessary), and then obtain more inbound data.

<code>SSLEngineResult r = engine.unwrap(src, dst);
    switch (r.getStatus()) {
    BUFFER_OVERFLOW:
        // Could attempt to drain the dst buffer of any already obtained
        // data, but we'll just increase it to the size needed.
        int appSize = engine.getSession().getApplicationBufferSize();
        ByteBuffer b = ByteBuffer.allocate(appSize + dst.position());
        dst.flip();
        b.put(dst);
        dst = b;
        // retry the operation.
        break;
    BUFFER_UNDERFLOW:
        int netSize = engine.getSession().getPacketBufferSize();
        // Resize buffer if needed.
        if (netSize &gt; dst.capacity()) {
            ByteBuffer b = ByteBuffer.allocate(netSize);
            src.flip();
            b.put(src);
            src = b;
        }
        // Obtain more inbound network data for src,
        // then retry the operation.
        break;
    // other cases: CLOSED, OK.
    }
  </code>

Unlike SSLSocket, all methods of SSLEngine are non-blocking. SSLEngine implementations may require the results of tasks that may take an extended period of time to complete, or may even block. For example, a TrustManager may need to connect to a remote certificate validation service, or a KeyManager might need to prompt a user to determine which certificate to use as part of client authentication. Additionally, creating cryptographic signatures and verifying them can be slow, seemingly blocking.

For any operation which may potentially block, the SSLEngine will create a java.lang.Runnable delegated task. When SSLEngineResult indicates that a delegated task result is needed, the application must call getDelegatedTask() to obtain an outstanding delegated task and call its run() method (possibly using a different thread depending on the compute strategy). The application should continue obtaining delegated tasks until no more exist, and try the original operation again.

At the end of a communication session, applications should properly close the SSL/TLS link. The SSL/TLS protocols have closure handshake messages, and these messages should be communicated to the peer before releasing the SSLEngine and closing the underlying transport mechanism. A close can be initiated by one of: an SSLException, an inbound closure handshake message, or one of the close methods. In all cases, closure handshake messages are generated by the engine, and wrap() should be repeatedly called until the resulting SSLEngineResult's status returns "CLOSED", or isOutboundDone() returns true. All data obtained from the wrap() method should be sent to the peer.

closeOutbound() is used to signal the engine that the application will not be sending any more data.

A peer will signal its intent to close by sending its own closure handshake message. After this message has been received and processed by the local SSLEngine's unwrap() call, the application can detect the close by calling unwrap() and looking for a SSLEngineResult with status "CLOSED", or if isInboundDone() returns true. If for some reason the peer closes the communication link without sending the proper SSL/TLS closure message, the application can detect the end-of-stream and can signal the engine via closeInbound() that there will no more inbound messages to process. Some applications might choose to require orderly shutdown messages from a peer, in which case they can check that the closure was generated by a handshake message and not by an end-of-stream condition.

There are two groups of cipher suites which you will need to know about when managing cipher suites:

• Supported cipher suites: all the suites which are supported by the SSL implementation. This list is reported using getSupportedCipherSuites().
• Enabled cipher suites, which may be fewer than the full set of supported suites. This group is set using the setEnabledCipherSuites(String[]) method, and queried using the getEnabledCipherSuites() method. Initially, a default set of cipher suites will be enabled on a new engine that represents the minimum suggested configuration.

Each SSL/TLS connection must have one client and one server, thus each endpoint must decide which role to assume. This choice determines who begins the handshaking process as well as which type of messages should be sent by each party. The method setUseClientMode(boolean) configures the mode. Once the initial handshaking has started, an SSLEngine can not switch between client and server modes, even when performing renegotiations.

Applications might choose to process delegated tasks in different threads. When an SSLEngine is created, the current java.security.AccessControlContext is saved. All future delegated tasks will be processed using this context: that is, all access control decisions will be made using the context captured at engine creation.

1. The wrap() and unwrap() methods may execute concurrently of each other.
2. The SSL/TLS protocols employ ordered packets. Applications must take care to ensure that generated packets are delivered in sequence. If packets arrive out-of-order, unexpected or fatal results may occur.

   For example: