6. The TLS Record Protocol

   The TLS Record Protocol is a layered protocol.  At each layer,
   messages may include fields for length, description, and content.
   The Record Protocol takes messages to be transmitted, fragments the
   data into manageable blocks, optionally compresses the data, applies
   a MAC, encrypts, and transmits the result.  Received data is
   decrypted, verified, decompressed, reassembled, and then delivered to
   higher-level clients.

   Four record protocol clients are described in this document: the
   handshake protocol, the alert protocol, the change cipher spec
   protocol, and the application data protocol.  In order to allow
   extension of the TLS protocol, additional record types can be
   supported by the record protocol.  Any new record types SHOULD
   allocate type values immediately beyond the ContentType values for
   the four record types described here (see Appendix A.1).  All such
   values must be defined by RFC 2434 Standards Action.  See Section 11
   for IANA Considerations for ContentType values.

   If a TLS implementation receives a record type it does not
   understand, it SHOULD just ignore it.  Any protocol designed for use

   over TLS MUST be carefully designed to deal with all possible attacks
   against it.  Note that because the type and length of a record are
   not protected by encryption, care SHOULD be taken to minimize the
   value of traffic analysis of these values.

6.1. Connection States

   A TLS connection state is the operating environment of the TLS Record
   Protocol.  It specifies a compression algorithm, and encryption
   algorithm, and a MAC algorithm.  In addition, the parameters for
   these algorithms are known: the MAC secret and the bulk encryption
   keys for the connection in both the read and the write directions.
   Logically, there are always four connection states outstanding: the
   current read and write states, and the pending read and write states.
   All records are processed under the current read and write states.
   The security parameters for the pending states can be set by the TLS
   Handshake Protocol, and the Change Cipher Spec can selectively make
   either of the pending states current, in which case the appropriate
   current state is disposed of and replaced with the pending state; the
   pending state is then reinitialized to an empty state.  It is illegal
   to make a state that has not been initialized with security
   parameters a current state.  The initial current state always
   specifies that no encryption, compression, or MAC will be used.

   The security parameters for a TLS Connection read and write state are
   set by providing the following values:

   connection end
      Whether this entity is considered the "client" or the "server" in
      this connection.

   bulk encryption algorithm
      An algorithm to be used for bulk encryption.  This specification
      includes the key size of this algorithm, how much of that key is
      secret, whether it is a block or stream cipher, and the block size
      of the cipher (if appropriate).

   MAC algorithm
      An algorithm to be used for message authentication.  This
      specification includes the size of the hash returned by the MAC
      algorithm.

   compression algorithm
      An algorithm to be used for data compression.  This specification
      must include all information the algorithm requires compression.

   master secret
      A 48-byte secret shared between the two peers in the connection.

   client random
      A 32-byte value provided by the client.

   server random
      A 32-byte value provided by the server.

   These parameters are defined in the presentation language as:

       enum { server, client } ConnectionEnd;

       enum { null, rc4, rc2, des, 3des, des40, idea, aes }
       BulkCipherAlgorithm;

       enum { stream, block } CipherType;

       enum { null, md5, sha } MACAlgorithm;

       enum { null(0), (255) } CompressionMethod;

       /* The algorithms specified in CompressionMethod,
          BulkCipherAlgorithm, and MACAlgorithm may be added to. */

       struct {
           ConnectionEnd          entity;
           BulkCipherAlgorithm    bulk_cipher_algorithm;
           CipherType             cipher_type;
           uint8                  key_size;
           uint8                  key_material_length;
           MACAlgorithm           mac_algorithm;
           uint8                  hash_size;
           CompressionMethod      compression_algorithm;
           opaque                 master_secret[48];
           opaque                 client_random[32];
           opaque                 server_random[32];
       } SecurityParameters;

   The record layer will use the security parameters to generate the
   following four items:

       client write MAC secret
       server write MAC secret
       client write key
       server write key

   The client write parameters are used by the server when receiving and
   processing records and vice-versa.  The algorithm used for generating
   these items from the security parameters is described in Section 6.3.

   Once the security parameters have been set and the keys have been
   generated, the connection states can be instantiated by making them
   the current states.  These current states MUST be updated for each
   record processed.  Each connection state includes the following
   elements:

   compression state
      The current state of the compression algorithm.

   cipher state
      The current state of the encryption algorithm.  This will consist
      of the scheduled key for that connection.  For stream ciphers,
      this will also contain whatever state information is necessary to
      allow the stream to continue to encrypt or decrypt data.

   MAC secret
      The MAC secret for this connection, as generated above.

   sequence number
      Each connection state contains a sequence number, which is
      maintained separately for read and write states.  The sequence
      number MUST be set to zero whenever a connection state is made the
      active state.  Sequence numbers are of type uint64 and may not
      exceed 2^64-1.  Sequence numbers do not wrap.  If a TLS
      implementation would need to wrap a sequence number, it must
      renegotiate instead.  A sequence number is incremented after each
      record: specifically, the first record transmitted under a
      particular connection state MUST use sequence number 0.

6.2. Record layer

   The TLS Record Layer receives uninterpreted data from higher layers
   in non-empty blocks of arbitrary size.

6.2.1. Fragmentation

   The record layer fragments information blocks into TLSPlaintext
   records carrying data in chunks of 2^14 bytes or less.  Client
   message boundaries are not preserved in the record layer (i.e.,
   multiple client messages of the same ContentType MAY be coalesced
   into a single TLSPlaintext record, or a single message MAY be
   fragmented across several records).

       struct {
           uint8 major, minor;
       } ProtocolVersion;

       enum {
           change_cipher_spec(20), alert(21), handshake(22),
           application_data(23), (255)
       } ContentType;

       struct {
           ContentType type;
           ProtocolVersion version;
           uint16 length;
           opaque fragment[TLSPlaintext.length];
       } TLSPlaintext;

   type
      The higher-level protocol used to process the enclosed fragment.

   version
      The version of the protocol being employed.  This document
      describes TLS Version 1.1, which uses the version { 3, 2 }.  The
      version value 3.2 is historical: TLS version 1.1 is a minor
      modification to the TLS 1.0 protocol, which was itself a minor
      modification to the SSL 3.0 protocol, which bears the version
      value 3.0.  (See Appendix A.1.)

   length
      The length (in bytes) of the following TLSPlaintext.fragment.  The
      length should not exceed 2^14.

   fragment
      The application data.  This data is transparent and is treated as
      an independent block to be dealt with by the higher-level protocol
      specified by the type field.

   Note: Data of different TLS Record layer content types MAY be
   interleaved.  Application data is generally of lower precedence for
   transmission than other content types.  However, records MUST be
   delivered to the network in the same order as they are protected by
   the record layer.  Recipients MUST receive and process interleaved
   application layer traffic during handshakes subsequent to the first
   one on a connection.

6.2.2. Record Compression and Decompression

   All records are compressed using the compression algorithm defined in
   the current session state.  There is always an active compression
   algorithm; however, initially it is defined as
   CompressionMethod.null.  The compression algorithm translates a
   TLSPlaintext structure into a TLSCompressed structure.  Compression
   functions are initialized with default state information whenever a
   connection state is made active.

   Compression must be lossless and may not increase the content length
   by more than 1024 bytes.  If the decompression function encounters a
   TLSCompressed.fragment that would decompress to a length in excess of
   2^14 bytes, it should report a fatal decompression failure error.

       struct {
           ContentType type;       /* same as TLSPlaintext.type */
           ProtocolVersion version;/* same as TLSPlaintext.version */
           uint16 length;
           opaque fragment[TLSCompressed.length];
       } TLSCompressed;

   length
      The length (in bytes) of the following TLSCompressed.fragment.
      The length should not exceed 2^14 + 1024.

   fragment
      The compressed form of TLSPlaintext.fragment.

   Note: A CompressionMethod.null operation is an identity operation; no
         fields are altered.

   Implementation note: Decompression functions are responsible for
                        ensuring that messages cannot cause internal
                        buffer overflows.

6.2.3. Record Payload Protection

   The encryption and MAC functions translate a TLSCompressed structure
   into a TLSCiphertext.  The decryption functions reverse the process.
   The MAC of the record also includes a sequence number so that
   missing, extra, or repeated messages are detectable.

       struct {
           ContentType type;
           ProtocolVersion version;
           uint16 length;
           select (CipherSpec.cipher_type) {
               case stream: GenericStreamCipher;
               case block: GenericBlockCipher;
           } fragment;
       } TLSCiphertext;

   type
      The type field is identical to TLSCompressed.type.

   version
      The version field is identical to TLSCompressed.version.

   length
      The length (in bytes) of the following TLSCiphertext.fragment.
      The length may not exceed 2^14 + 2048.

   fragment
      The encrypted form of TLSCompressed.fragment, with the MAC.

6.2.3.1. Null or Standard Stream Cipher

   Stream ciphers (including BulkCipherAlgorithm.null, see Appendix A.6)
   convert TLSCompressed.fragment structures to and from stream
   TLSCiphertext.fragment structures.

       stream-ciphered struct {
           opaque content[TLSCompressed.length];
           opaque MAC[CipherSpec.hash_size];
       } GenericStreamCipher;

   The MAC is generated as:

       HMAC_hash(MAC_write_secret, seq_num + TLSCompressed.type +
                     TLSCompressed.version + TLSCompressed.length +
                     TLSCompressed.fragment));

   where "+" denotes concatenation.

   seq_num
      The sequence number for this record.

   hash
      The hashing algorithm specified by
      SecurityParameters.mac_algorithm.

   Note that the MAC is computed before encryption.  The stream cipher
   encrypts the entire block, including the MAC.  For stream ciphers
   that do not use a synchronization vector (such as RC4), the stream
   cipher state from the end of one record is simply used on the
   subsequent packet.  If the CipherSuite is TLS_NULL_WITH_NULL_NULL,
   encryption consists of the identity operation (i.e., the data is not
   encrypted, and the MAC size is zero, implying that no MAC is used).
   TLSCiphertext.length is TLSCompressed.length plus
   CipherSpec.hash_size.

6.2.3.2. CBC Block Cipher

   For block ciphers (such as RC2, DES, or AES), the encryption and MAC
   functions convert TLSCompressed.fragment structures to and from block
   TLSCiphertext.fragment structures.

       block-ciphered struct {
           opaque IV[CipherSpec.block_length];
           opaque content[TLSCompressed.length];
           opaque MAC[CipherSpec.hash_size];
           uint8 padding[GenericBlockCipher.padding_length];
           uint8 padding_length;
       } GenericBlockCipher;

   The MAC is generated as described in Section 6.2.3.1.

   IV
      Unlike previous versions of SSL and TLS, TLS 1.1 uses an explicit
      IV in order to prevent the attacks described by [CBCATT].  We
      recommend the following equivalently strong procedures.  For
      clarity we use the following notation.

      IV
         The transmitted value of the IV field in the GenericBlockCipher
         structure.

      CBC residue
         The last ciphertext block of the previous record.

      mask
         The actual value that the cipher XORs with the plaintext prior
         to encryption of the first cipher block of the record.

      In prior versions of TLS, there was no IV field and the CBC
      residue and mask were one and the same.  See Sections 6.1,
      6.2.3.2, and 6.3, of [TLS1.0] for details of TLS 1.0 IV handling.

      One of the following two algorithms SHOULD be used to generate the
      per-record IV:

      (1) Generate a cryptographically strong random string R of length
          CipherSpec.block_length.  Place R in the IV field.  Set the
          mask to R.  Thus, the first cipher block will be encrypted as
          E(R XOR Data).

      (2) Generate a cryptographically strong random number R of length
          CipherSpec.block_length and prepend it to the plaintext prior
          to encryption.  In this case either:

          (a) The cipher may use a fixed mask such as zero.
          (b) The CBC residue from the previous record may be used as
              the mask.  This preserves maximum code compatibility with
              TLS 1.0 and SSL 3.  It also has the advantage that it does
              not require the ability to quickly reset the IV, which is
              known to be a problem on some systems.

          In either (2)(a) or (2)(b) the data (R || data) is fed into
          the encryption process.  The first cipher block (containing
          E(mask XOR R) is placed in the IV field.  The first block of
          content contains E(IV XOR data).

      The following alternative procedure MAY be used; however, it has
      not been demonstrated to be as cryptographically strong as the
      above procedures.  The sender prepends a fixed block F to the
      plaintext (or, alternatively, a block generated with a weak PRNG).
      He then encrypts as in (2), above, using the CBC residue from the
      previous block as the mask for the prepended block.  Note that in
      this case the mask for the first record transmitted by the
      application (the Finished) MUST be generated using a
      cryptographically strong PRNG.

      The decryption operation for all three alternatives is the same.
      The receiver decrypts the entire GenericBlockCipher structure and
      then discards the first cipher block, corresponding to the IV
      component.

   padding
      Padding that is added to force the length of the plaintext to be
      an integral multiple of the block cipher's block length.  The
      padding MAY be any length up to 255 bytes, as long as it results
      in the TLSCiphertext.length being an integral multiple of the
      block length.  Lengths longer than necessary might be desirable to
      frustrate attacks on a protocol that are based on analysis of the
      lengths of exchanged messages.  Each uint8 in the padding data
      vector MUST be filled with the padding length value.  The receiver

      MUST check this padding and SHOULD use the bad_record_mac alert to
      indicate padding errors.

   padding_length
      The padding length MUST be such that the total size of the
      GenericBlockCipher structure is a multiple of the cipher's block
      length.  Legal values range from zero to 255, inclusive.  This
      length specifies the length of the padding field exclusive of the
      padding_length field itself.

   The encrypted data length (TLSCiphertext.length) is one more than the
   sum of CipherSpec.block_length, TLSCompressed.length,
   CipherSpec.hash_size, and padding_length.

   Example: If the block length is 8 bytes, the content length
            (TLSCompressed.length) is 61 bytes, and the MAC length is 20
            bytes, then the length before padding is 82 bytes (this does
            not include the IV, which may or may not be encrypted, as
            discussed above).  Thus, the padding length modulo 8 must be
            equal to 6 in order to make the total length an even
            multiple of 8 bytes (the block length).  The padding length
            can be 6, 14, 22, and so on, through 254.  If the padding
            length were the minimum necessary, 6, the padding would be 6
            bytes, each containing the value 6.  Thus, the last 8 octets
            of the GenericBlockCipher before block encryption would be
            xx 06 06 06 06 06 06 06, where xx is the last octet of the
            MAC.

   Note: With block ciphers in CBC mode (Cipher Block Chaining), it is
         critical that the entire plaintext of the record be known
         before any ciphertext is transmitted.  Otherwise, it is
         possible for the attacker to mount the attack described in
         [CBCATT].

   Implementation Note: Canvel et al. [CBCTIME] have demonstrated a
                        timing attack on CBC padding based on the time
                        required to compute the MAC.  In order to defend
                        against this attack, implementations MUST ensure
                        that record processing time is essentially the
                        same whether or not the padding is correct.  In
                        general, the best way to do this is to compute
                        the MAC even if the padding is incorrect, and
                        only then reject the packet.  For instance, if
                        the pad appears to be incorrect, the
                        implementation might assume a zero-length pad
                        and then compute the MAC.  This leaves a small
                        timing channel, since MAC performance depends to
                        some extent on the size of the data fragment,

                        but it is not believed to be large enough to be
                        exploitable, due to the large block size of
                        existing MACs and the small size of the timing
                        signal.

6.3. Key Calculation

   The Record Protocol requires an algorithm to generate keys, and MAC
   secrets from the security parameters provided by the handshake
   protocol.

   The master secret is hashed into a sequence of secure bytes, which
   are assigned to the MAC secrets and keys required by the current
   connection state (see Appendix A.6).  CipherSpecs require a client
   write MAC secret, a server write MAC secret, a client write key, and
   a server write key, each of which is generated from the master secret
   in that order.  Unused values are empty.

   When keys and MAC secrets are generated, the master secret is used as
   an entropy source.

   To generate the key material, compute

       key_block = PRF(SecurityParameters.master_secret,
                          "key expansion",
                          SecurityParameters.server_random +
             SecurityParameters.client_random);

   until enough output has been generated.  Then the key_block is
   partitioned as follows:

       client_write_MAC_secret[SecurityParameters.hash_size]
       server_write_MAC_secret[SecurityParameters.hash_size]
       client_write_key[SecurityParameters.key_material_length]
       server_write_key[SecurityParameters.key_material_length]

   Implementation note: The currently defined cipher suite that requires
   the most material is AES_256_CBC_SHA, defined in [TLSAES].  It
   requires 2 x 32 byte keys, 2 x 20 byte MAC secrets, and 2 x 16 byte
   Initialization Vectors, for a total of 136 bytes of key material.

7. The TLS Handshaking Protocols

   TLS has three subprotocols that are used to allow peers to agree upon
   security parameters for the record layer, to authenticate themselves,
   to instantiate negotiated security parameters, and to report error
   conditions to each other.

   The Handshake Protocol is responsible for negotiating a session,
   which consists of the following items:

   session identifier
      An arbitrary byte sequence chosen by the server to identify an
      active or resumable session state.

   peer certificate
      X509v3 [X509] certificate of the peer.  This element of the state
      may be null.

   compression method
      The algorithm used to compress data prior to encryption.

   cipher spec
      Specifies the bulk data encryption algorithm (such as null, DES,
      etc.) and a MAC algorithm (such as MD5 or SHA).  It also defines
      cryptographic attributes such as the hash_size.  (See Appendix A.6
      for formal definition.)

   master secret
      48-byte secret shared between the client and server.

   is resumable
      A flag indicating whether the session can be used to initiate new
      connections.

   These items are then used to create security parameters for use by
   the Record Layer when protecting application data.  Many connections
   can be instantiated using the same session through the resumption
   feature of the TLS Handshake Protocol.

7.1. Change Cipher Spec Protocol

   The change cipher spec protocol exists to signal transitions in
   ciphering strategies.  The protocol consists of a single message,
   which is encrypted and compressed under the current (not the pending)
   connection state.  The message consists of a single byte of value 1.

       struct {
           enum { change_cipher_spec(1), (255) } type;
       } ChangeCipherSpec;

   The change cipher spec message is sent by both the client and the
   server to notify the receiving party that subsequent records will be
   protected under the newly negotiated CipherSpec and keys.  Reception
   of this message causes the receiver to instruct the Record Layer to
   immediately copy the read pending state into the read current state.

   Immediately after sending this message, the sender MUST instruct the
   record layer to make the write pending state the write active state.
   (See Section 6.1.)  The change cipher spec message is sent during the
   handshake after the security parameters have been agreed upon, but
   before the verifying finished message is sent (see Section 7.4.9).

   Note: If a rehandshake occurs while data is flowing on a connection,
         the communicating parties may continue to send data using the
         old CipherSpec.  However, once the ChangeCipherSpec has been
         sent, the new CipherSpec MUST be used.  The first side to send
         the ChangeCipherSpec does not know that the other side has
         finished computing the new keying material (e.g., if it has to
         perform a time consuming public key operation).  Thus, a small
         window of time, during which the recipient must buffer the
         data, MAY exist.  In practice, with modern machines this
         interval is likely to be fairly short.

7.2. Alert Protocol

         One of the content types supported by the TLS Record layer is
         the alert type.  Alert messages convey the severity of the
         message and a description of the alert.  Alert messages with a
         level of fatal result in the immediate termination of the
         connection.  In this case, other connections corresponding to
         the session may continue, but the session identifier MUST be
         invalidated, preventing the failed session from being used to
         establish new connections.  Like other messages, alert messages
         are encrypted and compressed, as specified by the current
         connection state.

             enum { warning(1), fatal(2), (255) } AlertLevel;

             enum {
                 close_notify(0),
                 unexpected_message(10),
                 bad_record_mac(20),
                 decryption_failed(21),
                 record_overflow(22),
                 decompression_failure(30),
                 handshake_failure(40),
                 no_certificate_RESERVED (41),
                 bad_certificate(42),
                 unsupported_certificate(43),
                 certificate_revoked(44),
                 certificate_expired(45),
                 certificate_unknown(46),
                 illegal_parameter(47),
                 unknown_ca(48),

                 access_denied(49),
                 decode_error(50),
                 decrypt_error(51),
                 export_restriction_RESERVED(60),
                 protocol_version(70),
                 insufficient_security(71),
                 internal_error(80),
                 user_canceled(90),
                 no_renegotiation(100),
                 (255)
             } AlertDescription;

             struct {
                 AlertLevel level;
                 AlertDescription description;
             } Alert;

7.2.1. Closure Alerts

   The client and the server must share knowledge that the connection is
   ending in order to avoid a truncation attack.  Either party may
   initiate the exchange of closing messages.

   close_notify
      This message notifies the recipient that the sender will not send
      any more messages on this connection.  Note that as of TLS 1.1,
      failure to properly close a connection no longer requires that a
      session not be resumed.  This is a change from TLS 1.0 to conform
      with widespread implementation practice.

   Either party may initiate a close by sending a close_notify alert.
   Any data received after a closure alert is ignored.

   Unless some other fatal alert has been transmitted, each party is
   required to send a close_notify alert before closing the write side
   of the connection.  The other party MUST respond with a close_notify
   alert of its own and close down the connection immediately,
   discarding any pending writes.  It is not required for the initiator
   of the close to wait for the responding close_notify alert before
   closing the read side of the connection.

   If the application protocol using TLS provides that any data may be
   carried over the underlying transport after the TLS connection is
   closed, the TLS implementation must receive the responding
   close_notify alert before indicating to the application layer that
   the TLS connection has ended.  If the application protocol will not
   transfer any additional data, but will only close the underlying
   transport connection, then the implementation MAY choose to close the

   transport without waiting for the responding close_notify.  No part
   of this standard should be taken to dictate the manner in which a
   usage profile for TLS manages its data transport, including when
   connections are opened or closed.

   Note: It is assumed that closing a connection reliably delivers
         pending data before destroying the transport.

7.2.2. Error Alerts

   Error handling in the TLS Handshake protocol is very simple.  When an
   error is detected, the detecting party sends a message to the other
   party.  Upon transmission or receipt of a fatal alert message, both
   parties immediately close the connection.  Servers and clients MUST
   forget any session-identifiers, keys, and secrets associated with a
   failed connection.  Thus, any connection terminated with a fatal
   alert MUST NOT be resumed.  The following error alerts are defined:

   unexpected_message
      An inappropriate message was received.  This alert is always fatal
      and should never be observed in communication between proper
      implementations.

   bad_record_mac
      This alert is returned if a record is received with an incorrect
      MAC.  This alert also MUST be returned if an alert is sent because
      a TLSCiphertext decrypted in an invalid way: either it wasn't an
      even multiple of the block length, or its padding values, when
      checked, weren't correct.  This message is always fatal.

   decryption_failed
      This alert MAY be returned if a TLSCiphertext decrypted in an
      invalid way: either it wasn't an even multiple of the block
      length, or its padding values, when checked, weren't correct.
      This message is always fatal.

   Note: Differentiating between bad_record_mac and decryption_failed
         alerts may permit certain attacks against CBC mode as used in
         TLS [CBCATT].  It is preferable to uniformly use the
         bad_record_mac alert to hide the specific type of the error.

   record_overflow
         A TLSCiphertext record was received that had a length more than
         2^14+2048 bytes, or a record decrypted to a TLSCompressed
         record with more than 2^14+1024 bytes.  This message is always
         fatal.

   decompression_failure
         The decompression function received improper input (e.g., data
         that would expand to excessive length).  This message is always
         fatal.

   handshake_failure
         Reception of a handshake_failure alert message indicates that
         the sender was unable to negotiate an acceptable set of
         security parameters given the options available.  This is a
         fatal error.

   no_certificate_RESERVED
         This alert was used in SSLv3 but not in TLS.  It should not be
         sent by compliant implementations.

   bad_certificate
         A certificate was corrupt, contained signatures that did not
         verify correctly, etc.

   unsupported_certificate
         A certificate was of an unsupported type.

   certificate_revoked
         A certificate was revoked by its signer.

   certificate_expired
         A certificate has expired or is not currently valid.

   certificate_unknown
         Some other (unspecified) issue arose in processing the
         certificate, rendering it unacceptable.

   illegal_parameter
         A field in the handshake was out of range or inconsistent with
         other fields.  This is always fatal.

   unknown_ca
         A valid certificate chain or partial chain was received, but
         the certificate was not accepted because the CA certificate
         could not be located or couldn't be matched with a known,
         trusted CA.  This message is always fatal.

   access_denied
         A valid certificate was received, but when access control was
         applied, the sender decided not to proceed with negotiation.
         This message is always fatal.

   decode_error
         A message could not be decoded because some field was out of
         the specified range or the length of the message was incorrect.
         This message is always fatal.

   decrypt_error
         A handshake cryptographic operation failed, including being
         unable to correctly verify a signature, decrypt a key exchange,
         or validate a finished message.

   export_restriction_RESERVED
         This alert was used in TLS 1.0 but not TLS 1.1.

   protocol_version
         The protocol version the client has attempted to negotiate is
         recognized but not supported.  (For example, old protocol
         versions might be avoided for security reasons).  This message
         is always fatal.

   insufficient_security
         Returned instead of handshake_failure when a negotiation has
         failed specifically because the server requires ciphers more
         secure than those supported by the client.  This message is
         always fatal.

   internal_error
         An internal error unrelated to the peer or the correctness of
         the protocol (such as a memory allocation failure) makes it
         impossible to continue.  This message is always fatal.

   user_canceled
         This handshake is being canceled for some reason unrelated to a
         protocol failure.  If the user cancels an operation after the
         handshake is complete, just closing the connection by sending a
         close_notify is more appropriate.  This alert should be
         followed by a close_notify.  This message is generally a
         warning.

   no_renegotiation
         Sent by the client in response to a hello request or by the
         server in response to a client hello after initial handshaking.
         Either of these would normally lead to renegotiation; when that
         is not appropriate, the recipient should respond with this
         alert.  At that point, the original requester can decide
         whether to proceed with the connection.  One case where this
         would be appropriate is where a server has spawned a process to
         satisfy a request; the process might receive security
         parameters (key length, authentication, etc.) at startup and it

         might be difficult to communicate changes to these parameters
         after that point.  This message is always a warning.

   For all errors where an alert level is not explicitly specified, the
   sending party MAY determine at its discretion whether this is a fatal
   error or not; if an alert with a level of warning is received, the
   receiving party MAY decide at its discretion whether to treat this as
   a fatal error or not.  However, all messages that are transmitted
   with a level of fatal MUST be treated as fatal messages.

   New alert values MUST be defined by RFC 2434 Standards Action.  See
   Section 11 for IANA Considerations for alert values.

7.3. Handshake Protocol Overview

   The cryptographic parameters of the session state are produced by the
   TLS Handshake Protocol, which operates on top of the TLS Record
   Layer.  When a TLS client and server first start communicating, they
   agree on a protocol version, select cryptographic algorithms,
   optionally authenticate each other, and use public-key encryption
   techniques to generate shared secrets.

   The TLS Handshake Protocol involves the following steps:

   -  Exchange hello messages to agree on algorithms, exchange random
      values, and check for session resumption.

   -  Exchange the necessary cryptographic parameters to allow the
      client and server to agree on a premaster secret.

   -  Exchange certificates and cryptographic information to allow the
      client and server to authenticate themselves.

   -  Generate a master secret from the premaster secret and exchanged
      random values.

   -  Provide security parameters to the record layer.

   -  Allow the client and server to verify that their peer has
      calculated the same security parameters and that the handshake
      occurred without tampering by an attacker.

   Note that higher layers should not be overly reliant on whether TLS
   always negotiates the strongest possible connection between two
   peers.  There are a number of ways in which a man-in-the-middle
   attacker can attempt to make two entities drop down to the least
   secure method they support.  The protocol has been designed to
   minimize this risk, but there are still attacks available.  For

   example, an attacker could block access to the port a secure service
   runs on, or attempt to get the peers to negotiate an unauthenticated
   connection.  The fundamental rule is that higher levels must be
   cognizant of what their security requirements are and never transmit
   information over a channel less secure than what they require.  The
   TLS protocol is secure in that any cipher suite offers its promised
   level of security: if you negotiate 3DES with a 1024 bit RSA key
   exchange with a host whose certificate you have verified, you can
   expect to be that secure.

   However, one SHOULD never send data over a link encrypted with 40-bit
   security unless one feels that data is worth no more than the effort
   required to break that encryption.

   These goals are achieved by the handshake protocol, which can be
   summarized as follows: The client sends a client hello message to
   which the server must respond with a server hello message, or else a
   fatal error will occur and the connection will fail.  The client
   hello and server hello are used to establish security enhancement
   capabilities between client and server.  The client hello and server
   hello establish the following attributes: Protocol Version, Session
   ID, Cipher Suite, and Compression Method.  Additionally, two random
   values are generated and exchanged: ClientHello.random and
   ServerHello.random.

   The actual key exchange uses up to four messages: the server
   certificate, the server key exchange, the client certificate, and the
   client key exchange.  New key exchange methods can be created by
   specifying a format for these messages and by defining the use of the
   messages to allow the client and server to agree upon a shared
   secret.  This secret MUST be quite long; currently defined key
   exchange methods exchange secrets that range from 48 to 128 bytes in
   length.

   Following the hello messages, the server will send its certificate,
   if it is to be authenticated.  Additionally, a server key exchange
   message may be sent, if it is required (e.g., if the server has no
   certificate, or if its certificate is for signing only).  If the
   server is authenticated, it may request a certificate from the
   client, if that is appropriate to the cipher suite selected.  Next,
   the server will send the server hello done message, indicating that
   the hello-message phase of the handshake is complete.  The server
   will then wait for a client response.  If the server has sent a
   certificate request message, the client must send the certificate
   message.  The client key exchange message is now sent, and the
   content of that message will depend on the public key algorithm
   selected between the client hello and the server hello.  If the
   client has sent a certificate with signing ability, a digitally-

   signed certificate verify message is sent to explicitly verify the
   certificate.


   At this point, a change cipher spec message is sent by the client,
   and the client copies the pending Cipher Spec into the current Cipher
   Spec.  The client then immediately sends the finished message under
   the new algorithms, keys, and secrets.  In response, the server will
   send its own change cipher spec message, transfer the pending to the
   current Cipher Spec, and send its finished message under the new
   Cipher Spec.  At this point, the handshake is complete, and the
   client and server may begin to exchange application layer data.  (See
   flow chart below.)  Application data MUST NOT be sent prior to the
   completion of the first handshake (before a cipher suite other
   TLS_NULL_WITH_NULL_NULL is established).

      Client                                               Server

      ClientHello                  -------->
                                                      ServerHello
                                                     Certificate*
                                               ServerKeyExchange*
                                              CertificateRequest*
                                   <--------      ServerHelloDone
      Certificate*
      ClientKeyExchange
      CertificateVerify*
      [ChangeCipherSpec]
      Finished                     -------->
                                               [ChangeCipherSpec]
                                   <--------             Finished
      Application Data             <------->     Application Data

             Fig. 1. Message flow for a full handshake

      * Indicates optional or situation-dependent messages that are not
        always sent.

   Note: To help avoid pipeline stalls, ChangeCipherSpec is an
         independent TLS Protocol content type, and is not actually a
         TLS handshake message.

   When the client and server decide to resume a previous session or
   duplicate an existing session (instead of negotiating new security
   parameters), the message flow is as follows:

   The client sends a ClientHello using the Session ID of the session to
   be resumed.  The server then checks its session cache for a match.

   If a match is found, and the server is willing to re-establish the
   connection under the specified session state, it will send a
   ServerHello with the same Session ID value.  At this point, both
   client and server MUST send change cipher spec messages and proceed
   directly to finished messages.  Once the re-establishment is
   complete, the client and server MAY begin to exchange application
   layer data.  (See flow chart below.)  If a Session ID match is not
   found, the server generates a new session ID and the TLS client and
   server perform a full handshake.

      Client                                                Server

      ClientHello                   -------->
                                                       ServerHello
                                                [ChangeCipherSpec]
                                    <--------             Finished
      [ChangeCipherSpec]
      Finished                      -------->
      Application Data              <------->     Application Data

          Fig. 2. Message flow for an abbreviated handshake

   The contents and significance of each message will be presented in
   detail in the following sections.