1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253
//! Traits and implementations for prime-order groups in which
//! the [decisional Diffie–Hellman][DDH] (DDH), [computational Diffie–Hellman][CDH] (CDH)
//! and [discrete log][DLP] (DL) problems are believed to be hard.
//!
//! (Decisional Diffie–Hellman assumption is considered stronger than both CDH and DL,
//! so if DDH is believed to hold for a certain group, it should be good to go.)
//!
//! Such groups can be applied for ElGamal encryption and other cryptographic protocols
//! from this crate.
//!
//! [DDH]: https://en.wikipedia.org/wiki/Decisional_Diffie%E2%80%93Hellman_assumption
//! [CDH]: https://en.wikipedia.org/wiki/Diffie%E2%80%93Hellman_problem
//! [DLP]: https://en.wikipedia.org/wiki/Discrete_logarithm
use merlin::Transcript;
use rand_chacha::ChaChaRng;
use rand_core::{CryptoRng, RngCore, SeedableRng};
use zeroize::Zeroize;
use core::{fmt, ops, str};
#[cfg(any(feature = "curve25519-dalek", feature = "curve25519-dalek-ng"))]
mod curve25519;
mod generic;
#[cfg(any(feature = "curve25519-dalek", feature = "curve25519-dalek-ng"))]
mod ristretto;
pub use self::generic::Generic;
#[cfg(any(feature = "curve25519-dalek", feature = "curve25519-dalek-ng"))]
pub use self::{curve25519::Curve25519Subgroup, ristretto::Ristretto};
/// Provides an arbitrary number of random bytes.
///
/// Unlike [`RngCore::fill_bytes()`], a single provider can only be used once.
pub struct RandomBytesProvider<'a> {
transcript: &'a mut Transcript,
label: &'static [u8],
}
impl fmt::Debug for RandomBytesProvider<'_> {
fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
let label = str::from_utf8(self.label).unwrap_or("(non-utf8 label)");
formatter
.debug_struct("RandomBytesProvider")
.field("label", &label)
.finish()
}
}
impl<'a> RandomBytesProvider<'a> {
pub(crate) fn new(transcript: &'a mut Transcript, label: &'static [u8]) -> Self {
Self { transcript, label }
}
/// Writes random bytes into the specified buffer. As follows from the signature, this method
/// can only be called once for a provider instance.
pub fn fill_bytes(self, dest: &mut [u8]) {
self.transcript.challenge_bytes(self.label, dest);
}
}
/// Helper trait for [`Group`] that describes operations on group scalars.
pub trait ScalarOps {
/// Scalar type. As per [`Group`] contract, scalars must form a prime field.
/// Arithmetic operations on scalars requested here must be constant-time.
type Scalar: Copy
+ Default
+ From<u64>
+ From<Self::Scalar> // `PublicKey::encrypt()` doesn't work without this
+ ops::Neg<Output = Self::Scalar>
+ ops::Add<Output = Self::Scalar>
+ for<'a> ops::Add<&'a Self::Scalar, Output = Self::Scalar>
+ ops::Sub<Output = Self::Scalar>
+ ops::Mul<Output = Self::Scalar>
+ for<'a> ops::Mul<&'a Self::Scalar, Output = Self::Scalar>
+ PartialEq
+ Zeroize
+ fmt::Debug;
/// Byte size of a serialized [`Self::Scalar`].
const SCALAR_SIZE: usize;
/// Generates a random scalar based on the provided CSPRNG. This operation
/// must be constant-time.
fn generate_scalar<R: CryptoRng + RngCore>(rng: &mut R) -> Self::Scalar;
/// Generates a scalar from a `source` of random bytes. This operation must be constant-time.
/// The `source` is guaranteed to return any necessary number of bytes.
///
/// # Default implementation
///
/// 1. Create a [ChaCha RNG] using 32 bytes read from `source` as the seed.
/// 2. Call [`Self::generate_scalar()`] with the created RNG.
///
/// [ChaCha RNG]: https://docs.rs/rand_chacha/
fn scalar_from_random_bytes(source: RandomBytesProvider<'_>) -> Self::Scalar {
let mut rng_seed = <ChaChaRng as SeedableRng>::Seed::default();
source.fill_bytes(&mut rng_seed);
let mut rng = ChaChaRng::from_seed(rng_seed);
Self::generate_scalar(&mut rng)
}
/// Inverts the `scalar`, which is guaranteed to be non-zero. This operation does not
/// need to be constant-time.
fn invert_scalar(scalar: Self::Scalar) -> Self::Scalar;
/// Inverts scalars in a batch. This operation does not need to be constant-time.
///
/// # Default implementation
///
/// Inverts every scalar successively.
fn invert_scalars(scalars: &mut [Self::Scalar]) {
for scalar in scalars {
*scalar = Self::invert_scalar(*scalar);
}
}
/// Serializes the scalar into the provided `buffer`, which is guaranteed to have length
/// [`Self::SCALAR_SIZE`].
fn serialize_scalar(scalar: &Self::Scalar, buffer: &mut [u8]);
/// Deserializes the scalar from `buffer`, which is guaranteed to have length
/// [`Self::SCALAR_SIZE`]. This method returns `None` if the buffer
/// does not correspond to a representation of a valid scalar.
fn deserialize_scalar(buffer: &[u8]) -> Option<Self::Scalar>;
}
/// Helper trait for [`Group`] that describes operations on group elements (i.e., EC points
/// for elliptic curve groups).
pub trait ElementOps: ScalarOps {
/// Element of the group. Arithmetic operations requested here (addition among
/// elements and multiplication by a `Scalar`) must be constant-time.
type Element: Copy
+ ops::Add<Output = Self::Element>
+ ops::Sub<Output = Self::Element>
+ ops::Neg<Output = Self::Element>
+ for<'a> ops::Mul<&'a Self::Scalar, Output = Self::Element>
+ PartialEq
+ fmt::Debug;
/// Byte size of a serialized [`Self::Element`].
const ELEMENT_SIZE: usize;
/// Returns the identity of the group (aka point at infinity for EC groups).
fn identity() -> Self::Element;
/// Checks if the specified element is the identity.
fn is_identity(element: &Self::Element) -> bool;
/// Returns the agreed-upon generator of the group.
fn generator() -> Self::Element;
/// Serializes `element` into the provided `buffer`, which is guaranteed to have length
/// [`Self::ELEMENT_SIZE`].
fn serialize_element(element: &Self::Element, output: &mut [u8]);
/// Deserializes an element from `buffer`, which is guaranteed to have length
/// [`Self::ELEMENT_SIZE`]. This method returns `None` if the buffer
/// does not correspond to a representation of a valid scalar.
fn deserialize_element(buffer: &[u8]) -> Option<Self::Element>;
}
/// Prime-order group in which the discrete log problem and decisional / computational
/// Diffie–Hellman problems are believed to be hard.
///
/// Groups conforming to this trait can be used for ElGamal encryption and other
/// cryptographic protocols defined in this crate.
///
/// This crate provides the following implementations of this trait:
///
/// - [`Curve25519Subgroup`], representation of a prime-order subgroup of Curve25519
/// with the conventionally chosen generator.
/// - [`Ristretto`], a transform of Curve25519 which eliminates its co-factor 8 with the help
/// of the [eponymous technique][ristretto].
/// - [`Generic`] implementation defined in terms of traits from the [`elliptic-curve`] crate.
/// (For example, this means secp256k1 support via the [`k256`] crate.)
///
/// [ristretto]: https://ristretto.group/
/// [`elliptic-curve`]: https://docs.rs/elliptic-curve/
/// [`k256`]: https://docs.rs/k256/
pub trait Group: Copy + ScalarOps + ElementOps + 'static {
/// Multiplies the provided scalar by [`ElementOps::generator()`]. This operation must be
/// constant-time.
///
/// # Default implementation
///
/// Implemented using [`Mul`](ops::Mul) (which is constant-time as per the [`ElementOps`]
/// contract).
fn mul_generator(k: &Self::Scalar) -> Self::Element {
Self::generator() * k
}
/// Multiplies the provided scalar by [`ElementOps::generator()`].
/// Unlike [`Self::mul_generator()`], this operation does not need to be constant-time;
/// thus, it may employ additional optimizations.
///
/// # Default implementation
///
/// Implemented by calling [`Self::mul_generator()`].
#[inline]
fn vartime_mul_generator(k: &Self::Scalar) -> Self::Element {
Self::mul_generator(k)
}
/// Multiplies provided `scalars` by `elements`. This operation must be constant-time
/// w.r.t. the given length of elements.
///
/// # Default implementation
///
/// Implemented by straightforward computations, which are constant-time as per
/// the [`ElementOps`] contract.
fn multi_mul<'a, I, J>(scalars: I, elements: J) -> Self::Element
where
I: IntoIterator<Item = &'a Self::Scalar>,
J: IntoIterator<Item = Self::Element>,
{
let mut output = Self::identity();
for (scalar, element) in scalars.into_iter().zip(elements) {
output = output + element * scalar;
}
output
}
/// Calculates `k * k_element + r * G`, where `G` is the group generator. This operation
/// does not need to be constant-time.
///
/// # Default implementation
///
/// Implemented by straightforward arithmetic.
fn vartime_double_mul_generator(
k: &Self::Scalar,
k_element: Self::Element,
r: &Self::Scalar,
) -> Self::Element {
k_element * k + Self::generator() * r
}
/// Multiplies provided `scalars` by `elements`. Unlike [`Self::multi_mul()`],
/// this operation does not need to be constant-time; thus, it may employ
/// additional optimizations.
///
/// # Default implementation
///
/// Implemented by calling [`Self::multi_mul()`].
#[inline]
fn vartime_multi_mul<'a, I, J>(scalars: I, elements: J) -> Self::Element
where
I: IntoIterator<Item = &'a Self::Scalar>,
J: IntoIterator<Item = Self::Element>,
{
Self::multi_mul(scalars, elements)
}
}