Using Transport Layer Security (TLS) to Secure QUIC

TLS provides two endpoints with a way to establish a means of communication over an untrusted medium (that is, the Internet) that ensures that messages they exchange cannot be observed, modified, or forged.¶

TLS features can be separated into two basic functions: an authenticated key exchange and record protection. QUIC primarily uses the authenticated key exchange provided by TLS but provides its own packet protection.¶

The TLS authenticated key exchange occurs between two entities: client and server. The client initiates the exchange and the server responds. If the key exchange completes successfully, both client and server will agree on a secret. TLS supports both pre-shared key (PSK) and Diffie-Hellman (DH) key exchanges. PSK is the basis for 0-RTT; the latter provides perfect forward secrecy (PFS) when the DH keys are destroyed.¶

After completing the TLS handshake, the client will have learned and authenticated an identity for the server and the server is optionally able to learn and authenticate an identity for the client. TLS supports X.509 [RFC5280] certificate-based authentication for both server and client.¶

The TLS key exchange is resistent to tampering by attackers and it produces shared secrets that cannot be controlled by either participating peer.¶

• A full 1-RTT handshake in which the client is able to send application data after one round trip and the server immediately after receiving the first handshake message from the client.
• A 0-RTT handshake in which the client uses information it has previously learned about the server to send application data immediately. This application data can be replayed by an attacker so it MUST NOT carry a self-contained trigger for any non-idempotent action.

A simplified TLS 1.3 handshake with 0-RTT application data is shown in Figure 2, see [I-D.ietf-tls-tls13] for more options and details.¶

This 0-RTT handshake is only possible if the client and server have previously communicated. In the 1-RTT handshake, the client is unable to send protected application data until it has received all of the handshake messages sent by the server.¶

• The server can respond to a ClientHello with a HelloRetryRequest, which adds an additional round trip prior to the basic exchange. This is needed if the server wishes to request a different key exchange key from the client. HelloRetryRequest is also used to verify that the client is correctly able to receive packets on the address it claims to have (see [QUIC-TRANSPORT]).
• A pre-shared key mode can be used for subsequent handshakes to reduce the number of public key operations. This is the basis for 0-RTT data, even if the remainder of the connection is protected by a new Diffie-Hellman exchange.

The integration of QUIC with a TLS handshake is shown in more detail in Figure 3. QUIC STREAM frames on stream 0 carry the TLS handshake. QUIC performs loss recovery [QUIC-RECOVERY] for this stream and ensures that TLS handshake messages are delivered in the correct order.¶

• ”<” and “>” enclose stream numbers.
• ”@” indicates the keys that are used for protecting the QUIC packet (C = cleartext, with integrity only; 0 = 0-RTT keys; 1 = 1-RTT keys).
• ”(“ and “)” enclose messages that are protected with TLS 0-RTT handshake or application keys.
• ”{“ and “}” enclose messages that are protected by the TLS Handshake keys.

If 0-RTT is not attempted, then the client does not send packets protected by the 0-RTT key (@0). In that case, the only key transition on the client is from cleartext packets (@C) to 1-RTT protection (@1), which happens after it sends its final set of TLS handshake messages.¶

Note: the client uses two different types of cleartext packet during the handshake. The Client Initial packet carries a TLS ClientHello message; the remainder of the TLS handshake is carried in Client Cleartext packets.¶

The server sends TLS handshake messages without protection (@C). The server transitions from no protection (@C) to full 1-RTT protection (@1) after it sends the last of its handshake messages.¶

Some TLS handshake messages are protected by the TLS handshake record protection. These keys are not exported from the TLS connection for use in QUIC. QUIC packets from the server are sent in the clear until the final transition to 1-RTT keys.¶

The client transitions from cleartext (@C) to 0-RTT keys (@0) when sending 0-RTT data, and subsequently to to 1-RTT keys (@1) after its second flight of TLS handshake messages. This creates the potential for unprotected packets to be received by a server in close proximity to packets that are protected with 1-RTT keys.¶

More information on key transitions is included in Section 7.1.¶

As shown in Figure 1, the interface from QUIC to TLS consists of four primary functions: Handshake, Source Address Validation, Key Ready Events, and Secret Export.¶

Additional functions might be needed to configure TLS.¶

This document describes how TLS 1.3 [I-D.ietf-tls-tls13] is used with QUIC.¶

In practice, the TLS handshake will negotiate a version of TLS to use. This could result in a newer version of TLS than 1.3 being negotiated if both endpoints support that version. This is acceptable provided that the features of TLS 1.3 that are used by QUIC are supported by the newer version.¶

A badly configured TLS implementation could negotiate TLS 1.2 or another older version of TLS. An endpoint MUST terminate the connection if a version of TLS older than 1.3 is negotiated.¶

QUIC requires that the initial handshake packet from a client fit within the payload of a single packet. The size limits on QUIC packets mean that a record containing a ClientHello needs to fit within 1197 octets.¶

A TLS ClientHello can fit within this limit with ample space remaining. However, there are several variables that could cause this limit to be exceeded. Implementations are reminded that large session tickets or HelloRetryRequest cookies, multiple or large key shares, and long lists of supported ciphers, signature algorithms, versions, QUIC transport parameters, and other negotiable parameters and extensions could cause this message to grow.¶

For servers, the size of the session tickets and HelloRetryRequest cookie extension can have an effect on a client’s ability to connect. Choosing a small value increases the probability that these values can be successfully used by a client.¶

The TLS implementation does not need to ensure that the ClientHello is sufficiently large. QUIC PADDING frames are added to increase the size of the packet as necessary.¶

The requirements for authentication depend on the application protocol that is in use. TLS provides server authentication and permits the server to request client authentication.¶

A client MUST authenticate the identity of the server. This typically involves verification that the identity of the server is included in a certificate and that the certificate is issued by a trusted entity (see for example [RFC2818]).¶

A server MAY request that the client authenticate during the handshake. A server MAY refuse a connection if the client is unable to authenticate when requested. The requirements for client authentication vary based on application protocol and deployment.¶

A server MUST NOT use post-handshake client authentication (see Section 4.6.2 of [I-D.ietf-tls-tls13]).¶

Errors in the TLS connection SHOULD be signaled using TLS alerts on stream 0. A failure in the handshake MUST be treated as a QUIC connection error of type TLS_HANDSHAKE_FAILED. Once the handshake is complete, an error in the TLS connection that causes a TLS alert to be sent or received MUST be treated as a QUIC connection error of type TLS_FATAL_ALERT_GENERATED or TLS_FATAL_ALERT_RECEIVED respectively.¶

As TLS reports the availability of keying material, the packet protection keys and initialization vectors (IVs) are updated (see Section 5.2). The selection of AEAD function is also updated to match the AEAD negotiated by TLS.¶

For packets other than any unprotected handshake packets (see Section 7.1), once a change of keys has been made, packets with higher packet numbers MUST be sent with the new keying material. The KEY_PHASE bit on these packets is inverted each time new keys are installed to signal the use of the new keys to the recipient (see Section 7 for details).¶

An endpoint retransmits stream data in a new packet. New packets have new packet numbers and use the latest packet protection keys. This simplifies key management when there are key updates (see Section 7.2).¶

QUIC uses a system of packet protection secrets, keys and IVs that are modelled on the system used in TLS [I-D.ietf-tls-tls13]. The secrets that QUIC uses as the basis of its key schedule are obtained using TLS exporters (see Section 7.5 of [I-D.ietf-tls-tls13]).¶

QUIC uses HKDF with the same hash function negotiated by TLS for key derivation. For example, if TLS is using the TLS_AES_128_GCM_SHA256, the SHA-256 hash function is used.¶

The Authentication Encryption with Associated Data (AEAD) [RFC5116] function used for QUIC packet protection is AEAD that is negotiated for use with the TLS connection. For example, if TLS is using the TLS_AES_128_GCM_SHA256, the AEAD_AES_128_GCM function is used.¶

Regular QUIC packets are protected by an AEAD algorithm [RFC5116]. Version negotiation and public reset packets are not protected.¶

Once TLS has provided a key, the contents of regular QUIC packets immediately after any TLS messages have been sent are protected by the AEAD selected by TLS.¶

The key, K, is either the client packet protection key (client_pp_key_n) or the server packet protection key (server_pp_key_n), derived as defined in Section 5.2.¶

The nonce, N, is formed by combining the packet protection IV (either client_pp_iv_n or server_pp_iv_n) with the packet number. The 64 bits of the reconstructed QUIC packet number in network byte order is left-padded with zeros to the size of the IV. The exclusive OR of the padded packet number and the IV forms the AEAD nonce.¶

The associated data, A, for the AEAD is the contents of the QUIC header, starting from the flags octet in the common header.¶

The input plaintext, P, for the AEAD is the contents of the QUIC frame following the packet number, as described in [QUIC-TRANSPORT].¶

The output ciphertext, C, of the AEAD is transmitted in place of P.¶

Prior to TLS providing keys, no record protection is performed and the plaintext, P, is transmitted unmodified.¶

QUIC has a single, contiguous packet number space. In comparison, TLS restarts its sequence number each time that record protection keys are changed. The sequence number restart in TLS ensures that a compromise of the current traffic keys does not allow an attacker to truncate the data that is sent after a key update by sending additional packets under the old key (causing new packets to be discarded).¶

QUIC does not assume a reliable transport and is required to handle attacks where packets are dropped in other ways. QUIC is therefore not affected by this form of truncation.¶

The QUIC packet number is not reset and it is not permitted to go higher than its maximum value of 2^64-1. This establishes a hard limit on the number of packets that can be sent.¶

Some AEAD functions have limits for how many packets can be encrypted under the same key and IV (see for example [AEBounds]). This might be lower than the packet number limit. An endpoint MUST initiate a key update (Section 7.2) prior to exceeding any limit set for the AEAD that is in use.¶

TLS maintains a separate sequence number that is used for record protection on the connection that is hosted on stream 0. This sequence number is not visible to QUIC.¶

Once an endpoint successfully receives a packet with a given packet number, it MUST discard all packets with higher packet numbers if they cannot be successfully unprotected with either the same key, or - if there is a key update - the next packet protection key (see Section 7.2). Similarly, a packet that appears to trigger a key update, but cannot be unprotected successfully MUST be discarded.¶

Failure to unprotect a packet does not necessarily indicate the existence of a protocol error in a peer or an attack. The truncated packet number encoding used in QUIC can cause packet numbers to be decoded incorrectly if they are delayed significantly.¶

An endpoint sending a packet that has a long header and a type that does not indicate that the packet will be protected (that is, 0-RTT Encrypted (0x05), 1-RTT Encrypted (key phase 0) (0x06), or 1-RTT Encrypted (key phase 1) (0x07)) first constructs the packet that it sends without the integrity check.¶

The sender then calculates the integrity check over the entire packet, starting from the type field. The output of the hash is appended to the packet.¶

A receiver that receives an unprotected packet first checks that the version is correct, then removes the trailing 8 octets. It calculates the integrity check over the remainder of the packet. Unprotected packets that do not contain a valid integrity check MUST be discarded.¶

QUIC uses the 64-bit version of the alternative Fowler/Noll/Vo hash (FNV-1a) [FNV].¶

hash := offset basis for each input octet: hash := hash XOR input octet hash := hash * prime¶

That is, a 64-bit unsigned integer is initialized with an offset basis. Then, for each octet of the input, the exclusive binary OR of the value is taken, then multiplied by a prime. Any overflow from multiplication is discarded.¶

The offset basis for the 64-bit FNV-1a is the decimal value 14695981039346656037 (in hex, 0xcbf29ce484222325). The prime is 1099511628211 (in hex, 0x100000001b3; or as an expression 2^40 + 2^8 + 0xb3).¶

Once all octets have been processed in this fashion, the final integer value is encoded as 8 octets in network byte order.¶

The initial exchange of packets are sent without protection. These packets use a cleartext packet type.¶

TLS handshake messages MUST NOT be protected using QUIC packet protection. All TLS handshake messages up to the TLS Finished message sent by either endpoint use cleartext packets.¶

Any TLS handshake messages that are sent after completing the TLS handshake do not need special packet protection rules. Packets containing these messages use the packet protection keys that are current at the time of sending (or retransmission).¶

Like the client, a server MUST send retransmissions of its unprotected handshake messages or acknowledgments for unprotected handshake messages sent by the client in cleartext packets.¶

Once the TLS handshake is complete, the KEY_PHASE bit allows for refreshes of keying material by either peer. Endpoints start using updated keys immediately without additional signaling; the change in the KEY_PHASE bit indicates that a new key is in use.¶

An endpoint MUST NOT initiate more than one key update at a time. A new key cannot be used until the endpoint has received and successfully decrypted a packet with a matching KEY_PHASE. Note that when 0-RTT is attempted the value of the KEY_PHASE bit will be different on packets sent by either peer.¶

A receiving endpoint detects an update when the KEY_PHASE bit doesn’t match what it is expecting. It creates a new secret (see Section 5.2) and the corresponding read key and IV. If the packet can be decrypted and authenticated using these values, then the keys it uses for packet protection are also updated. The next packet sent by the endpoint will then use the new keys.¶

An endpoint doesn’t need to send packets immediately when it detects that its peer has updated keys. The next packet that it sends will simply use the new keys. If an endpoint detects a second update before it has sent any packets with updated keys it indicates that its peer has updated keys twice without awaiting a reciprocal update. An endpoint MUST treat consecutive key updates as a fatal error and abort the connection.¶

An endpoint SHOULD retain old keys for a short period to allow it to decrypt packets with smaller packet numbers than the packet that triggered the key update. This allows an endpoint to consume packets that are reordered around the transition between keys. Packets with higher packet numbers always use the updated keys and MUST NOT be decrypted with old keys.¶

Keys and their corresponding secrets SHOULD be discarded when an endpoint has received all packets with sequence numbers lower than the lowest sequence number used for the new key. An endpoint might discard keys if it determines that the length of the delay to affected packets is excessive.¶

This ensures that once the handshake is complete, packets with the same KEY_PHASE will have the same packet protection keys, unless there are multiple key updates in a short time frame succession and significant packet reordering.¶

As shown in Figure 3 and Figure 12, there is never a situation where there are more than two different sets of keying material that might be received by a peer. Once both sending and receiving keys have been updated,¶

A server cannot initiate a key update until it has received the client’s Finished message. Otherwise, packets protected by the updated keys could be confused for retransmissions of handshake messages. A client cannot initiate a key update until all of its handshake messages have been acknowledged by the server.¶

A packet that triggers a key update could arrive after successfully processing a packet with a higher packet number. This is only possible if there is a key compromise and an attack, or if the peer is incorrectly reverting to use of old keys. Because the latter cannot be differentiated from an attack, an endpoint MUST immediately terminate the connection if it detects this condition.¶

The cookie extension in the TLS HelloRetryRequest message allows a server to perform source address validation during the handshake.¶

When QUIC requests address validation during the processing of the first ClientHello, the token it provides is included in the cookie extension of a HelloRetryRequest. As long as the cookie cannot be successfully guessed by a client, the server can be assured that the client received the HelloRetryRequest if it includes the value in a second ClientHello.¶

An initial ClientHello never includes a cookie extension. Thus, if a server constructs a cookie that contains all the information necessary to reconstruct state, it can discard local state after sending a HelloRetryRequest. Presence of a valid cookie in a ClientHello indicates that the ClientHello is a second attempt from the client.¶

An address validation token can be extracted from a second ClientHello and passed to the transport for further validation. If that validation fails, the server MUST fail the TLS handshake and send an illegal_parameter alert.¶

Combining address validation with the other uses of HelloRetryRequest ensures that there are fewer ways in which an additional round-trip can be added to the handshake. In particular, this makes it possible to combine a request for address validation with a request for a different client key share.¶

If TLS needs to send a HelloRetryRequest for other reasons, it needs to ensure that it can correctly identify the reason that the HelloRetryRequest was generated. During the processing of a second ClientHello, TLS does not need to consult the transport protocol regarding address validation if address validation was not requested originally. In such cases, the cookie extension could either be absent or it could indicate that an address validation token is not present.¶

The ticket in the TLS NewSessionTicket message allows a server to provide a client with a similar sort of token. When a client resumes a TLS connection - whether or not 0-RTT is attempted - it includes the ticket in the handshake message. As with the HelloRetryRequest cookie, the server includes the address validation token in the ticket. TLS provides the token it extracts from the session ticket to the transport when it asks whether source address validation is needed.¶

If both a HelloRetryRequest cookie and a session ticket are present in the ClientHello, only the token from the cookie is passed to the transport. The presence of a cookie indicates that this is a second ClientHello - the token from the session ticket will have been provided to the transport when it appeared in the first ClientHello.¶

A server can send a NewSessionTicket message at any time. This allows it to update the state - and the address validation token - that is included in the ticket. This might be done to refresh the ticket or token, or it might be generated in response to changes in the state of the connection. QUIC can request that a NewSessionTicket be sent by providing a new address validation token.¶

A server that intends to support 0-RTT SHOULD provide an address validation token immediately after completing the TLS handshake.¶

TLS MUST provide integrity protection for address validation token unless the transport guarantees integrity protection by other means. For a NewSessionTicket that includes confidential information - such as the resumption secret - including the token under authenticated encryption ensures that the token gains both confidentiality and integrity protection without duplicating the overheads of that protection.¶

This section describes the handling of messages that are sent and received prior to the completion of the TLS handshake.¶

Sending and receiving unprotected messages is hazardous. Unless expressly permitted, receipt of an unprotected message of any kind MUST be treated as a fatal error.¶

If 0-RTT keys are available, the lack of replay protection means that restrictions on their use are necessary to avoid replay attacks on the protocol.¶

A client MUST only use 0-RTT keys to protect data that is idempotent. A client MAY wish to apply additional restrictions on what data it sends prior to the completion of the TLS handshake. A client otherwise treats 0-RTT keys as equivalent to 1-RTT keys.¶

A client that receives an indication that its 0-RTT data has been accepted by a server can send 0-RTT data until it receives all of the server’s handshake messages. A client SHOULD stop sending 0-RTT data if it receives an indication that 0-RTT data has been rejected.¶

A server MUST NOT use 0-RTT keys to protect packets.¶

Due to reordering and loss, protected packets might be received by an endpoint before the final TLS handshake messages are received. A client will be unable to decrypt 1-RTT packets from the server, whereas a server will be able to decrypt 1-RTT packets from the client.¶

Packets protected with 1-RTT keys MAY be stored and later decrypted and used once the handshake is complete. A server MUST NOT use 1-RTT protected packets before verifying either the client Finished message or - in the case that the server has chosen to use a pre-shared key - the pre-shared key binder (see Section 4.2.8 of [I-D.ietf-tls-tls13]). Verifying these values provides the server with an assurance that the ClientHello has not been modified.¶

A server could receive packets protected with 0-RTT keys prior to receiving a TLS ClientHello. The server MAY retain these packets for later decryption in anticipation of receiving a ClientHello.¶

Receiving and verifying the TLS Finished message is critical in ensuring the integrity of the TLS handshake. A server MUST NOT use protected packets from the client prior to verifying the client Finished message if its response depends on client authentication.¶

The QUIC version negotiation mechanism is used to negotiate the version of QUIC that is used prior to the completion of the handshake. However, this packet is not authenticated, enabling an active attacker to force a version downgrade.¶

To ensure that a QUIC version downgrade is not forced by an attacker, version information is copied into the TLS handshake, which provides integrity protection for the QUIC negotiation. This does not prevent version downgrade prior to the completion of the handshake, though it means that a downgrade causes a handshake failure.¶

TLS uses Application Layer Protocol Negotiation (ALPN) [RFC7301] to select an application protocol. The application-layer protocol MAY restrict the QUIC versions that it can operate over. Servers MUST select an application protocol compatible with the QUIC version that the client has selected.¶

If the server cannot select a compatible combination of application protocol and QUIC version, it MUST abort the connection. A client MUST abort a connection if the server picks an incompatible combination of QUIC version and ALPN identifier.¶

QUIC transport parameters are carried in a TLS extension. Different versions of QUIC might define a different format for this struct.¶

Including transport parameters in the TLS handshake provides integrity protection for these values.¶

The extension_data field of the quic_transport_parameters extension contains a value that is defined by the version of QUIC that is in use. The quic_transport_parameters extension carries a TransportParameters when the version of QUIC defined in [QUIC-TRANSPORT] is used.¶

The quic_transport_parameters extension is carried in the ClientHello and the EncryptedExtensions messages during the handshake. The extension MAY be included in a NewSessionTicket message.¶

QUIC uses TLS without modification. Therefore, it is possible to use a pre-shared key that was established in a TLS handshake over TCP to enable 0-RTT in QUIC. Similarly, QUIC can provide a pre-shared key that can be used to enable 0-RTT in TCP.¶

All the restrictions on the use of 0-RTT apply, with the exception of the ALPN label, which MUST only change to a label that is explicitly designated as being compatible. The client indicates which ALPN label it has chosen by placing that ALPN label first in the ALPN extension.¶

The certificate that the server uses MUST be considered valid for both connections, which will use different protocol stacks and could use different port numbers. For instance, HTTP/1.1 and HTTP/2 operate over TLS and TCP, whereas QUIC operates over UDP.¶

Source address validation is not completely portable between different protocol stacks. Even if the source IP address remains constant, the port number is likely to be different. Packet reflection attacks are still possible in this situation, though the set of hosts that can initiate these attacks is greatly reduced. A server might choose to avoid source address validation for such a connection, or allow an increase to the amount of data that it sends toward the client without source validation.¶

A small ClientHello that results in a large block of handshake messages from a server can be used in packet reflection attacks to amplify the traffic generated by an attacker.¶

Certificate caching [RFC7924] can reduce the size of the server’s handshake messages significantly.¶

QUIC requires that the packet containing a ClientHello be padded to a minimum size. A server is less likely to generate a packet reflection attack if the data it sends is a small multiple of this size. A server SHOULD use a HelloRetryRequest if the size of the handshake messages it sends is likely to significantly exceed the size of the packet containing the ClientHello.¶

QUIC, TLS and HTTP/2 all contain a messages that have legitimate uses in some contexts, but that can be abused to cause a peer to expend processing resources without having any observable impact on the state of the connection. If processing is disproportionately large in comparison to the observable effects on bandwidth or state, then this could allow a malicious peer to exhaust processing capacity without consequence.¶

QUIC prohibits the sending of empty STREAM frames unless they are marked with the FIN bit. This prevents STREAM frames from being sent that only waste effort.¶

TLS records SHOULD always contain at least one octet of a handshake messages or alert. Records containing only padding are permitted during the handshake, but an excessive number might be used to generate unnecessary work. Once the TLS handshake is complete, endpoints SHOULD NOT send TLS application data records unless it is to hide the length of QUIC records. QUIC packet protection does not include any allowance for padding; padded TLS application data records can be used to mask the length of QUIC frames.¶

While there are legitimate uses for some redundant packets, implementations SHOULD track redundant packets and treat excessive volumes of any non-productive packets as indicative of an attack.¶

• Updates to match changes in transport draft

• Use TLS alerts to signal TLS errors (#272, #374)
• Require ClientHello to fit in a single packet (#338)
• The second client handshake flight is now sent in the clear (#262, #337)
• The QUIC header is included as AEAD Associated Data (#226, #243, #302)
• Add interface necessary for client address validation (#275)
• Define peer authentication (#140)
• Require at least TLS 1.3 (#138)
• Define transport parameters as a TLS extension (#122)
• Define handling for protected packets before the handshake completes (#39)
• Decouple QUIC version and ALPN (#12)

• Changed bit used to signal key phase
• Updated key phase markings during the handshake
• Added TLS interface requirements section
• Moved to use of TLS exporters for key derivation
• Moved TLS error code definitions into this document

• Adopted as base for draft-ietf-quic-tls
• Updated authors/editors list
• Added status note