Note
Network deployment and testing in progress. Subject to minor revisions. See [SPEC] for the official specification.
Overview
This document proposes changes to Tunnel Build message encryption using crypto primitives introduced by [ECIES-X25519]. It is a portion of the overall proposal [Prop156] for converting routers from ElGamal to ECIES-X25519 keys.
For the purposes of transitioning the network from ElGamal + AES256 to ECIES + ChaCha20, tunnels with mixed ElGamal and ECIES routers are necessary. Specifications for handling mixed tunnel hops are provided. No changes will be made to the format, processing, or encryption of ElGamal hops.
ElGamal tunnel creators will need create ephemeral X25519 keypairs per-hop, and follow this spec for creating tunnels containing ECIES hops.
This proposal specifies changes needed for ECIES-X25519 Tunnel Building. For an overview of all changes required for ECIES routers, see proposal 156 [Prop156].
This proposal maintains the same size for tunnel build records, as required for compatibility. Smaller build records and messages will be implemented later - see [Prop157].
Cryptographic Primitives
No new cryptographic primitives are introduced. The primitives required to implement this proposal are:
- AES-256-CBC as in [Cryptography]
- STREAM ChaCha20/Poly1305 functions: ENCRYPT(k, n, plaintext, ad) and DECRYPT(k, n, ciphertext, ad) - as in [NTCP2] [ECIES-X25519] and [RFC-7539]
- X25519 DH functions - as in [NTCP2] and [ECIES-X25519]
- HKDF(salt, ikm, info, n) - as in [NTCP2] and [ECIES-X25519]
Other Noise functions defined elsewhere:
- MixHash(d) - as in [NTCP2] and [ECIES-X25519]
- MixKey(d) - as in [NTCP2] and [ECIES-X25519]
Goals
- Increase speed of crypto operations
- Replace ElGamal + AES256/CBC with ECIES primitives for tunnel BuildRequestRecords and BuildReplyRecords.
- No change to size of encrypted BuildRequestRecords and BuildReplyRecords (528 bytes) for compatibility
- No new I2NP messages
- Maintain encrypted build record size for compatibility
- Add forward secrecy for Tunnel Build Messages.
- Add authenticated encryption
- Detect hops reordering BuildRequestRecords
- Increase resolution of timestamp so that Bloom filter size may be reduced
- Add field for tunnel expiration so that varying tunnel lifetimes will be possible (all-ECIES tunnels only)
- Add extensible options field for future features
- Reuse existing cryptographic primitives
- Improve tunnel build message security where possible while maintaining compatibility
- Support tunnels with mixed ElGamal/ECIES peers
- Improve defenses against "tagging" attacks on build messages
- Hops do not need to know the encryption type of the next hop before processing the build message, as they may not have the next hop's RI at that time
- Maximize compatibility with current network
- No change to tunnel build AES request/reply encryption for ElGamal routers
- No change to tunnel AES "layer" encryption, for that see [Prop153]
- Continue to support both 8-record TBM/TBRM and variable-size VTBM/VTBRM
- Do not require "flag day" upgrade to entire network
Non-Goals
- Complete redesign of tunnel build messages requiring a "flag day".
- Shrinking tunnel build messages (requires all-ECIES hops and a new proposal)
- Use of tunnel build options as defined in [Prop143], only required for small messages
- Bidirectional tunnels - for that see [Prop119]
- Smaller tunnel build messages - for that see [Prop157]
Threat Model
Design Goals
- No hops are able to determine the originator of the tunnel.
- Middle hops must not be able to determine the direction of the tunnel or their position in the tunnel.
- No hops can read any contents of other request or reply records, except for truncated router hash and ephemeral key for next hop
- No member of reply tunnel for outbound build can read any reply records.
- No member of outbound tunnel for inbound build can read any request records, except that OBEP can see truncated router hash and ephemeral key for IBGW
Tagging Attacks
A major goal of the tunnel building design is to make it harder for colluding routers X and Y to know that they are in a single tunnel. If router X is at hop m and router Y is at hop m+1, they obviously will know. But if router X is at hop m and router Y is at hop m+n for n>1, this should be much harder.
Tagging attacks are where middle-hop router X alters the tunnel build message in such a way that router Y can detect the alteration when the build message gets there. The goal is for any altered message is dropped by a router between X and Y before it gets to router Y. For modifications that are not dropped before router Y, the tunnel creator should detect the corruption in the reply and discard the tunnel.
Possible attacks:
- Alter a build record
- Replace a build record
- Add or remove a build record
- Reorder the build records
TODO: Does the current design prevent all these attacks?
Design
Noise Protocol Framework
This proposal provides the requirements based on the Noise Protocol Framework [NOISE] (Revision 34, 2018-07-11). In Noise parlance, Alice is the initiator, and Bob is the responder.
This proposal is based on the Noise protocol Noise_N_25519_ChaChaPoly_SHA256. This Noise protocol uses the following primitives:
- One-Way Handshake Pattern: N Alice does not transmit her static key to Bob (N)
- DH Function: X25519 X25519 DH with a key length of 32 bytes as specified in [RFC-7748].
- Cipher Function: ChaChaPoly AEAD_CHACHA20_POLY1305 as specified in [RFC-7539] section 2.8. 12 byte nonce, with the first 4 bytes set to zero. Identical to that in [NTCP2].
- Hash Function: SHA256 Standard 32-byte hash, already used extensively in I2P.
Additions to the Framework
None.
Handshake Patterns
Handshakes use [Noise] handshake patterns.
The following letter mapping is used:
- e = one-time ephemeral key
- s = static key
- p = message payload
The build request is identical to the Noise N pattern. This is also identical to the first (Session Request) message in the XK pattern used in [NTCP2].
<- s
...
e es p ->
Request encryption
Build request records are created by the tunnel creator and asymmetrically encrypted to the individual hop. This asymmetric encryption of request records is currently ElGamal as defined in [Cryptography] and contains a SHA-256 checksum. This design is not forward-secret.
The new design will use the one-way Noise pattern "N" with ECIES-X25519 ephemeral-static DH, with an HKDF, and ChaCha20/Poly1305 AEAD for forward secrecy, integrity, and authentication. Alice is the tunnel build requestor. Each hop in the tunnel is a Bob.
(Payload Security Properties)
N: Authentication Confidentiality
-> e, es 0 2
Authentication: None (0).
This payload may have been sent by any party, including an active attacker.
Confidentiality: 2.
Encryption to a known recipient, forward secrecy for sender compromise
only, vulnerable to replay. This payload is encrypted based only on DHs
involving the recipient's static key pair. If the recipient's static
private key is compromised, even at a later date, this payload can be
decrypted. This message can also be replayed, since there's no ephemeral
contribution from the recipient.
"e": Alice generates a new ephemeral key pair and stores it in the e
variable, writes the ephemeral public key as cleartext into the
message buffer, and hashes the public key along with the old h to
derive a new h.
"es": A DH is performed between the Alice's ephemeral key pair and the
Bob's static key pair. The result is hashed along with the old ck to
derive a new ck and k, and n is set to zero.
Reply encryption
Build reply records are created by the hops creator and symmetrically encrypted to the creator. This symmetric encryption of reply records is currently AES with a prepended SHA-256 checksum. and contains a SHA-256 checksum. This design is not forward-secret.
The new design will use ChaCha20/Poly1305 AEAD for integrity, and authentication.
Justification
The ephemeral public key in the request does not need to be obfuscated with AES or Elligator2. The previous hop is the only one that can see it, and that hop knows that the next hop is ECIES.
Reply records do not need full asymmetric encryption with another DH.
Specification
Build Request Records
Encrypted BuildRequestRecords are 528 bytes for both ElGamal and ECIES, for compatibility.
Request Record Unencrypted (ElGamal)
For reference, this is the current specification of the tunnel BuildRequestRecord for ElGamal routers, taken from [I2NP]. The unencrypted data is prepended with a nonzero byte and the SHA-256 hash of the data before encryption, as defined in [Cryptography].
All fields are big-endian.
Unencrypted size: 222 bytes
bytes 0-3: tunnel ID to receive messages as, nonzero
bytes 4-35: local router identity hash
bytes 36-39: next tunnel ID, nonzero
bytes 40-71: next router identity hash
bytes 72-103: AES-256 tunnel layer key
bytes 104-135: AES-256 tunnel IV key
bytes 136-167: AES-256 reply key
bytes 168-183: AES-256 reply IV
byte 184: flags
bytes 185-188: request time (in hours since the epoch, rounded down)
bytes 189-192: next message ID
bytes 193-221: uninterpreted / random padding
Request Record Encrypted (ElGamal)
For reference, this is the current specification of the tunnel BuildRequestRecord for ElGamal routers, taken from [I2NP].
Encrypted size: 528 bytes
bytes 0-15: Hop's truncated identity hash
bytes 16-528: ElGamal encrypted BuildRequestRecord
Request Record Unencrypted (ECIES)
This is the proposed specification of the tunnel BuildRequestRecord for ECIES-X25519 routers. Summary of changes:
- Remove unused 32-byte router hash
- Change request time from hours to minutes
- Add expiration field for future variable tunnel time
- Add more space for flags
- Add Mapping for additional build options
- AES-256 reply key and IV are not used for the hop's own reply record
- Unencrypted record is longer because there is less encryption overhead
The request record does not contain any ChaCha reply keys. Those keys are derived from a KDF. See below.
All fields are big-endian.
Unencrypted size: 464 bytes
bytes 0-3: tunnel ID to receive messages as, nonzero
bytes 4-7: next tunnel ID, nonzero
bytes 8-39: next router identity hash
bytes 40-71: AES-256 tunnel layer key
bytes 72-103: AES-256 tunnel IV key
bytes 104-135: AES-256 reply key
bytes 136-151: AES-256 reply IV
byte 152: flags
bytes 153-155: more flags, unused, set to 0 for compatibility
bytes 156-159: request time (in minutes since the epoch, rounded down)
bytes 160-163: request expiration (in seconds since creation)
bytes 164-167: next message ID
bytes 168-x: tunnel build options (Mapping)
bytes x-x: other data as implied by flags or options
bytes x-463: random padding
The flags field is the same as defined in [Tunnel-Creation] and contains the following:
Bit order: 76543210 (bit 7 is MSB) bit 7: if set, allow messages from anyone bit 6: if set, allow messages to anyone, and send the reply to the specified next hop in a Tunnel Build Reply Message bits 5-0: Undefined, must set to 0 for compatibility with future options
Bit 7 indicates that the hop will be an inbound gateway (IBGW). Bit 6 indicates that the hop will be an outbound endpoint (OBEP). If neither bit is set, the hop will be an intermediate participant. Both cannot be set at once.
The request exipration is for future variable tunnel duration. For now, the only supported value is 600 (10 minutes).
The tunnel build options is a Mapping structure as defined in [Common]. This is for future use. No options are currently defined. If the Mapping structure is empty, this is two bytes 0x00 0x00. The maximum size of the Mapping (including the length field) is 296 bytes, and the maximum value of the Mapping length field is 294.
Request Record Encrypted (ECIES)
All fields are big-endian except for the ephemeral public key which is little-endian.
Encrypted size: 528 bytes
bytes 0-15: Hop's truncated identity hash
bytes 16-47: Sender's ephemeral X25519 public key
bytes 48-511: ChaCha20 encrypted BuildRequestRecord
bytes 512-527: Poly1305 MAC
Build Reply Records
Encrypted BuildReplyRecords are 528 bytes for both ElGamal and ECIES, for compatibility.
Reply Record Unencrypted (ElGamal)
ElGamal replies are encrypted with AES.
All fields are big-endian.
Unencrypted size: 528 bytes
bytes 0-31: SHA-256 Hash of bytes 32-527
bytes 32-526: random data
byte 527: reply
total length: 528
Reply Record Unencrypted (ECIES)
This is the proposed specification of the tunnel BuildReplyRecord for ECIES-X25519 routers. Summary of changes:
- Add Mapping for build reply options
- Unencrypted record is longer because there is less encryption overhead
ECIES replies are encrypted with ChaCha20/Poly1305.
All fields are big-endian.
Unencrypted size: 512 bytes
bytes 0-x: Tunnel Build Reply Options (Mapping)
bytes x-x: other data as implied by options
bytes x-510: Random padding
byte 511: Reply byte
The tunnel build reply options is a Mapping structure as defined in [Common]. This is for future use. No options are currently defined. If the Mapping structure is empty, this is two bytes 0x00 0x00. The maximum size of the Mapping (including the length field) is 511 bytes, and the maximum value of the Mapping length field is 509.
The reply byte is one of the following values as defined in [Tunnel-Creation] to avoid fingerprinting:
- 0x00 (accept)
- 30 (TUNNEL_REJECT_BANDWIDTH)
Reply Record Encrypted (ECIES)
Encrypted size: 528 bytes
bytes 0-511: ChaCha20 encrypted BuildReplyRecord
bytes 512-527: Poly1305 MAC
After full transition to ECIES records, ranged padding rules are the same as for request records.
Symmetric Encryption of Records
Mixed tunnels are allowed, and necessary, for the transition from ElGamal to ECIES. During the transitionary period, an increasing number of routers will be keyed under ECIES keys.
Symmetric cryptography preprocessing will run in the same way:
- "encryption":
- cipher run in decryption mode
- request records preemptively decrypted in preprocessing (concealing encrypted request records)
- "decryption":
- cipher run in encryption mode
- request records encrypted (revealing next plaintext request record) by participant hops
- ChaCha20 does not have "modes", so it is simply run three times:
- once in preprocessing
- once by the hop
- once on final reply processing
When mixed tunnels are used, tunnel creators will need to base the symmetric encryption of BuildRequestRecord on the current and previous hop's encryption type.
Each hop will use its own encryption type for encrypting BuildReplyRecords, and the other records in the VariableTunnelBuildMessage (VTBM).
On the reply path, the endpoint (sender) will need to undo the [Multiple-Encryption], using each hop's reply key.
As a clarifying example, let's look at an outbound tunnel w/ ECIES surrounded by ElGamal:
- Sender (OBGW) -> ElGamal (H1) -> ECIES (H2) -> ElGamal (H3)
All BuildRequestRecords are in their encrypted state (using ElGamal or ECIES).
AES256/CBC cipher, when used, is still used for each record, without chaining across multiple records.
Likewise, ChaCha20 will be used to encrypt each record, not streaming across the entire VTBM.
The request records are preprocessed by the Sender (OBGW):
- H3's record is "encrypted" using:
- H2's reply key (ChaCha20)
- H1's reply key (AES256/CBC)
- H2's record is "encrypted" using:
- H1's reply key (AES256/CBC)
- H1's record goes out without symmetric encryption
Only H2 checks the reply encryption flag, and sees its followed by AES256/CBC.
After being processed by each hop, the records are in a "decrypted" state:
- H3's record is "decrypted" using:
- H3's reply key (AES256/CBC)
- H2's record is "decrypted" using:
- H3's reply key (AES256/CBC)
- H2's reply key (ChaCha20-Poly1305)
- H1's record is "decrypted" using:
- H3's reply key (AES256/CBC)
- H2's reply key (ChaCha20)
- H1's reply key (AES256/CBC)
The tunnel creator, a.k.a. Inbound Endpoint (IBEP), postprocesses the reply:
- H3's record is "encrypted" using:
- H3's reply key (AES256/CBC)
- H2's record is "encrypted" using:
- H3's reply key (AES256/CBC)
- H2's reply key (ChaCha20-Poly1305)
- H1's record is "encrypted" using:
- H3's reply key (AES256/CBC)
- H2's reply key (ChaCha20)
- H1's reply key (AES256/CBC)
Request Record Keys (ECIES)
These keys are explicitly included in ElGamal BuildRequestRecords. For ECIES BuildRequestRecords, the tunnel keys and AES reply keys are included, but the ChaCha reply keys are derived from the DH exchange. See [Prop156] for details of the router static ECIES keys.
Below is a description of how to derive the keys previously transmitted in request records.
KDF for Initial ck and h
This is standard [NOISE] for pattern "N" with a standard protocol name.
This is the "e" message pattern:
// Define protocol_name.
Set protocol_name = "Noise_N_25519_ChaChaPoly_SHA256"
(31 bytes, US-ASCII encoded, no NULL termination).
// Define Hash h = 32 bytes
// Pad to 32 bytes. Do NOT hash it, because it is not more than 32 bytes.
h = protocol_name || 0
Define ck = 32 byte chaining key. Copy the h data to ck.
Set chainKey = h
// MixHash(null prologue)
h = SHA256(h);
// up until here, can all be precalculated by all routers.
KDF for Request Record
ElGamal tunnel creators generate an ephemeral X25519 keypair for each ECIES hop in the tunnel, and use scheme above for encrypting their BuildRequestRecord. ElGamal tunnel creators will use the scheme prior to this spec for encrypting to ElGamal hops.
ECIES tunnel creators will need to encrypt to each of the ElGamal hop's public key using the scheme defined in [Tunnel-Creation]. ECIES tunnel creators will use the above scheme for encrypting to ECIES hops.
This means that tunnel hops will only see encrypted records from their same encryption type.
For ElGamal and ECIES tunnel creators, they will generate unique ephemeral X25519 keypairs per-hop for encrypting to ECIES hops.
IMPORTANT: Ephemeral keys must be unique per ECIES hop, and per build record. Failing to use unique keys opens an attack vector for colluding hops to confirm they are in the same tunnel.
// Each hop's X25519 static keypair (hesk, hepk) from the Router Identity
hesk = GENERATE_PRIVATE()
hepk = DERIVE_PUBLIC(hesk)
// MixHash(hepk)
// || below means append
h = SHA256(h || hepk);
// up until here, can all be precalculated by each router
// for all incoming build requests
// Sender generates an X25519 ephemeral keypair per ECIES hop in the VTBM (sesk, sepk)
sesk = GENERATE_PRIVATE()
sepk = DERIVE_PUBLIC(sesk)
// MixHash(sepk)
h = SHA256(h || sepk);
End of "e" message pattern.
This is the "es" message pattern:
// Noise es
// Sender performs an X25519 DH with Hop's static public key.
// Each Hop, finds the record w/ their truncated identity hash,
// and extracts the Sender's ephemeral key preceding the encrypted record.
sharedSecret = DH(sesk, hepk) = DH(hesk, sepk)
// MixKey(DH())
//[chainKey, k] = MixKey(sharedSecret)
// ChaChaPoly parameters to encrypt/decrypt
keydata = HKDF(chainKey, sharedSecret, "", 64)
// Save for Reply Record KDF
chainKey = keydata[0:31]
// AEAD parameters
k = keydata[32:64]
n = 0
plaintext = 464 byte build request record
ad = h
ciphertext = ENCRYPT(k, n, plaintext, ad)
End of "es" message pattern.
// MixHash(ciphertext)
// Save for Reply Record KDF
h = SHA256(h || ciphertext)
replyKey, layerKey and layerIV must still be included inside ElGamal records, and can be generated randomly.
Request Record Encryption (ElGamal)
As defined in [Tunnel-Creation]. There are no changes to encryption for ElGamal hops.
Reply Record Encryption (ECIES)
The reply record is ChaCha20/Poly1305 encrypted.
// AEAD parameters
k = chainkey from build request
n = 0
plaintext = 512 byte build reply record
ad = h from build request
ciphertext = ENCRYPT(k, n, plaintext, ad)
Reply Record Encryption (ElGamal)
As defined in [Tunnel-Creation]. There are no changes to encryption for ElGamal hops.
Security Analysis
ElGamal does not provide forward secrecy for Tunnel Build Messages.
AES256/CBC is in slightly better standing, only being vulnerable to a theoretical weakening from a known plaintext biclique attack.
The only known practical attack against AES256/CBC is a padding oracle attack, when the IV is known to the attacker.
An attacker would need to break the next hop's ElGamal encryption to gain the AES256/CBC key info (reply key and IV).
ElGamal is significantly more CPU-intensive than ECIES, leading to potential resource exhaustion.
ECIES, used with new ephemeral keys per-BuildRequestRecord or VariableTunnelBuildMessage, provides forward-secrecy.
ChaCha20Poly1305 provides AEAD encryption, allowing the recipient to verify message integrity before attempting decryption.
Justification
This design maximizes reuse of existing cryptographic primitives, protocols, and code. This design minimizes risk.
Implementation Notes
- Older routers do not check the encryption type of the hop and will send ElGamal-encrypted records. Some recent routers are buggy and will send various types of malformed records. Implementers should detect and reject these records prior to the DH operation if possible, to reduce CPU usage.
Issues
Migration
See [Prop156].
References
[Common] | (1, 2) https://geti2p.net/spec/common-structures |
[Cryptography] | (1, 2, 3) https://geti2p.net/spec/cryptography |
[ECIES-X25519] | (1, 2, 3, 4, 5, 6) https://geti2p.net/spec/ecies |
[I2NP] | (1, 2) https://geti2p.net/spec/i2np |
[NOISE] | (1, 2, 3) https://noiseprotocol.org/noise.html |
[NTCP2] | (1, 2, 3, 4, 5, 6, 7) https://geti2p.net/spec/ntcp2 |
[Prop119] | https://geti2p.net/spec/proposals/119-bidirectional-tunnels |
[Prop143] | https://geti2p.net/spec/proposals/143-build-message-options |
[Prop153] | https://geti2p.net/spec/proposals/153-chacha20-layer-encryption |
[Prop156] | (1, 2, 3, 4) https://geti2p.net/spec/proposals/156-ecies-routers |
[Prop157] | (1, 2) https://geti2p.net/spec/proposals/157-new-tbm |
[SPEC] | https://geti2p.net/spec/tunnel-creation-ecies |
[Tunnel-Creation] | (1, 2, 3, 4, 5) https://geti2p.net/spec/tunnel-creation |
[Multiple-Encryption] | https://en.wikipedia.org/wiki/Multiple_encryption |
[RFC-7539] | (1, 2) https://tools.ietf.org/html/rfc7539 |
[RFC-7748] | https://tools.ietf.org/html/rfc7748 |