g3p-hash-2.0.0.0: Global Password Prehash Protocol
Copyright(c) 2024 Auth Global
LicenseApache2
Safe HaskellSafe-Inferred
LanguageHaskell2010

Crypto.G3P.V1

Description

 
Synopsis

Documentation

data G3PInputBlock Source #

These input parameters are grouped together because the envisioned use for them is that they are constants (or near-constants) specified by a deployment. User-supplied inputs would typically not go here. In this role, all these parameters function as salt.

The seguid parameter acts as a deployment-wide salt. Cryptographically speaking, the most important thing a deployment can do is specify a constant seguid. It is highly recommended that the seguid input be a genuine Self-Documenting Globally Unique Identifier attesting to the parameters, purposes, and public playbook of the protocol for y'all to follow to use the deployment to spec.

The remaining string parameters are all directly-documenting, embedded attributions. A deployment can use these tags to encode a message into the password hash function so that it must be known to whomever can compute it. There are a variety of different parameters because there are different lengths of messages that can be expressed for free, and there are different incremental costs for exceeding that limit.

It is particularly important to include some kind of actionable message in the domainTag, longTag, bcryptTag, and bcryptSaltTag parameters. Specifying an empty string in any of these parameters means that a significant quantity of cryptoacoustic messaging space will be filled with silence.

Especially useful messages include URIs, legal names, and domain names.

Constructors

G3PInputBlock 

Fields

  • g3pInputBlock_seguid :: !ByteString

    HMAC-SHA256 key, usable as a high-repetition indirect tag via self-documenting globally unique identifiers (seguids).

  • g3pInputBlock_domainTag :: !ByteString

    plaintext tag with one repetition per PHKDF round. 0-19 bytes are free, 20-82 bytes cost a additional sha256 block per PHKDF round, with 83-146 and every 64 bytes thereafter incurring a similar cost.

    Tags up to 82 or maybe even 146 bytes long are reasonable in most contexts. In the case of long domain tags, it is strategically advantageous to ensure that the first 32 bytes are highly actionable, as these bytes are commonly used as filler padding.

    This parameter provides domain separation. A suggested value is a ICANN domain name controlled by the deployment. The name is also a bit of an homage to the "realm" parameter of HTTP basic authentication, which in part inspired it.

  • g3pInputBlock_longTag :: !ByteString

    plaintext tag with 1x repetition, then cycled for roughly 8 kilobytes. Constant time on inputs up to nearly 5 kilobytes.

    Overages incur one sha256 block per 64 bytes.

  • g3pInputBlock_tags :: !(Vector ByteString)

    plaintext tags with 3x repetition. Constant-time on 0-63 encoded bytes, which includes the length encoding of each string. Thus 60 of those free bytes are usable if the tags vector is a single string, or less if it contains two or more strings.

    Overages incur three sha256 blocks per 64 bytes.

    This parameter is notable because it is the least expensive purely auxiliary input that is not horn-loaded. Thus if you want a very long salt input that provides a bit of extra collision resistance, (because the collision resistance of HMAC-SHA256 isn't enough?!) this would be a logical candidate input location to consider.

  • g3pInputBlock_phkdfRounds :: !Word32

    How expensive will the PHKDF component be? An optimal implementation computes exactly three SHA256 blocks per round if the domain tag is 19 bytes or less, plus a reasonably large but constant number of additional blocks. I recommend at least 20,000 rounds, if not 40,000. You might consider adjusting that recommendation downward in the case of domain tags that exceed 19 bytes in length: 15,000 rounds of PHKDF with a domain tag that is 83 bytes long should cost about the same number of SHA256 blocks as 20,000 rounds of PHKDF with a domain tag that is 19 bytes long.

  • g3pInputBlock_bcryptRounds :: !Word32

    How expensive will the bcrypt component be? 4000 rounds recommended, give or take a factor of 2 or so. Each bcrypt round is approximately as time consuming as 60 PHKDF rounds. Using the recommendation, the cost should be dominated by bcrypt.

  • g3pInputBlock_bcryptTag :: !ByteString

    Repeated once or twice per bcrypt round, plus once in PHKDF. This tag has exactly two full repetitions per bcrypt round when the tag is up to 56 bytes long. Above 56 bytes, this tag is cyclically extended to 112 bytes and then split into two strings of 56 bytes, each repeated once. Hashed once in PHKDF to avoid any truncation gotchas, and to force recomputation of PHKDF if this tag is varied.

    0-112 bytes can be handled in a constant number of cryptographic operations. Overages incur a cost of one SHA-256 block per 64 bytes.

data G3PInputArgs Source #

The username and password are grouped together because they are normally expected to be supplied by users or other observers of a deployment.

Furthermore, the credentials vector is here because it is an ideal location to include other user input. For example, one could implement a Two-Secret Key Derivation (2SKD) scheme analogous to 1Password's.

A deployment can also specify additional constant tags as part of the credentials vector. As the plaintext of these tags is only ever hashed into the output a single time, this alongside the bcrypt tag and long tag are incrementally the least expensive options for plaintext tagging.

Note that the username and password are subjected to additional length hardening. The G3P operates in a constant number of SHA256 blocks so long as the combined length of the username and password is less than about 3 KiB, or the combined length of the username, password, and long tag is less than about 8 KiB. The actual numbers are somewhat less in both cases, but this is a reasonable approximation. Note that the bcrypt tag can subtract up to 113 bytes from the 8 KiB total, and don't effect the 3 KiB total.

In the case of all of the inputs in this record, longer values incur one SHA256 block per 64 bytes.

Constructors

G3PInputArgs 

Fields

newtype G3PInputRole Source #

These parameters are used to tweak the final output, without redoing any expensive key stretching. A possible use case is including a high entropy secret in the role itself that isn't available until after a successful stage of authentication.

Since these parameters are processed in a context that could conceivably be performance sensitive, we don't apply any length padding or side-channel hardening. Instead we opt for maximizing free tagging space. Thus we want to avoid incurring additional SHA256 block computations, one of the favorite techniques employed by the key-stretching phase of the G3P to harden against timing side-channels.

A deployment could conceivably harden this expansion phase against timing side channels themselves, if the were sufficiently inclined. There are several techniques. For starters, a deployment could ensure that these parameters themselves are constant-length. Alternatively, a deployment could specify an additional variable-length string in the role vector, used to control the ending position relative to the SHA256 buffer.

Constructors

G3PInputRole 

Fields

  • g3pInputRole_roleTags :: Vector ByteString

    This is the least expensive parameter that will vary the secret HMAC key used to generate the final output. Very much analogous to HKDF's initial keying material (ikm) parameter. This is the recommended last call for mixing additional secrets into the output.

newtype G3PInputEcho Source #

Constructors

G3PInputEcho 

Fields

  • g3pInputEcho_echoTag :: ByteString

    the absolute least expensive parameter to vary, if your implementation supports efficent parameter variation. Very much analogous to HKDF's info parameter. 0-19 bytes are free. This incurs a cost of one SHA-256 block per output block at 20 bytes, 83 bytes, and every 64 bytes thereafter.

data G3PSeed Source #

A plain-old-data explicit representation of the intermediate g3pHash computation after the G3PInputBlock and G3PInputArgs have been processed and key stretching has been completed, but before the tweaks have been applied and the final output generated.

If you ever need to serialize or persist a seed, you probably want this.

Intended to be generated by g3pHash_seedInit and consumed by g3pHash_seedFinalize.

Constructors

G3PSeed 

Fields

g3pHash :: G3PInputBlock -> G3PInputArgs -> G3PInputRole -> G3PInputEcho -> Stream ByteString Source #

The Global Password Prehash Protocol (G3P). Note that this function is very intentionally implemented in such a way that the following idiom is efficient. It performs the expensive key stretching phase only once.

 let mySeed = g3pHash block args
     myKey0 = mySeed myRole0
     myKey1 = mySeed myRole1
  in [ myKey0 myEcho, myKey0 altEcho, myKey1 myEcho, myKey1 altEcho ]

Although the savings from sharing the computation of the seed among all four computations is much larger, this expression also shares the computation of the G3PKey among these four output streams, so that only two keys are computed.

In the case that you want or need to persist or serialize the intermediate seed, or change the seguid or domain tag before final output expansion, then the plain-old-datatype G3PSeed and its companion functions g3pHash_keyInit and g3pHash_finalizeGen are needed.

g3pHash_seedInit :: G3PInputBlock -> G3PInputArgs -> G3PSeed Source #

This generates a seed, which encapsulates the expensive key-stretching component of g3pHash into a reusable, tweakable cryptographic value. This function is way slower than it's companions, g3pHash_keyInit and g3pHash_finalizeGen

Broadly comparable to PBKDF2, and incorporates bcrypt.

g3pHash_keyInit :: G3PInputRole -> G3PSeed -> G3PKey Source #

This applies the role vector to a computed G3P seed to arrive at a precomputed G3P key used during the final output expansion. Broadly comparable to HKDF-Extract. This function applies no key stretching, and thus output can be available in as little as 6 or 7 SHA256 block computations when you vary its parameters.

Note that this function ignores g3pSeed_seguid in favor of g3pSeed_seguidKey.

g3pHash_finalizeGen :: G3PInputEcho -> G3PKey -> G3PGen Source #

Apply the echo tag parameter to a precomputed G3P key, obtaining a cryptographically secure output generator. Broadly comparable to HKDF-Expand. Applies no key stretching, so output is available in as little as 2 SHA256 block computations.

g3pGen_read :: G3PGen -> (ByteString, G3PGen) Source #

Read a 32-byte hash from the G3P's output generator.

g3pGen_finalizeStream :: G3PGen -> Stream ByteString Source #

Turn a G3P output generator into an unbounded stream.