Files
serai/coordinator/src/tributary/signing_protocol.rs
Luke Parker e4e4245ee3 One Round DKG (#589)
* Upstream GBP, divisor, circuit abstraction, and EC gadgets from FCMP++

* Initial eVRF implementation

Not quite done yet. It needs to communicate the resulting points and proofs to
extract them from the Pedersen Commitments in order to return those, and then
be tested.

* Add the openings of the PCs to the eVRF as necessary

* Add implementation of secq256k1

* Make DKG Encryption a bit more flexible

No longer requires the use of an EncryptionKeyMessage, and allows pre-defined
keys for encryption.

* Make NUM_BITS an argument for the field macro

* Have the eVRF take a Zeroizing private key

* Initial eVRF-based DKG

* Add embedwards25519 curve

* Inline the eVRF into the DKG library

Due to how we're handling share encryption, we'd either need two circuits or to
dedicate this circuit to the DKG. The latter makes sense at this time.

* Add documentation to the eVRF-based DKG

* Add paragraph claiming robustness

* Update to the new eVRF proof

* Finish routing the eVRF functionality

Still needs errors and serialization, along with a few other TODOs.

* Add initial eVRF DKG test

* Improve eVRF DKG

Updates how we calculcate verification shares, improves performance when
extracting multiple sets of keys, and adds more to the test for it.

* Start using a proper error for the eVRF DKG

* Resolve various TODOs

Supports recovering multiple key shares from the eVRF DKG.

Inlines two loops to save 2**16 iterations.

Adds support for creating a constant time representation of scalars < NUM_BITS.

* Ban zero ECDH keys, document non-zero requirements

* Implement eVRF traits, all the way up to the DKG, for secp256k1/ed25519

* Add Ristretto eVRF trait impls

* Support participating multiple times in the eVRF DKG

* Only participate once per key, not once per key share

* Rewrite processor key-gen around the eVRF DKG

Still a WIP.

* Finish routing the new key gen in the processor

Doesn't touch the tests, coordinator, nor Substrate yet.
`cargo +nightly fmt && cargo +nightly-2024-07-01 clippy --all-features -p serai-processor`
does pass.

* Deduplicate and better document in processor key_gen

* Update serai-processor tests to the new key gen

* Correct amount of yx coefficients, get processor key gen test to pass

* Add embedded elliptic curve keys to Substrate

* Update processor key gen tests to the eVRF DKG

* Have set_keys take signature_participants, not removed_participants

Now no one is removed from the DKG. Only `t` people publish the key however.

Uses a BitVec for an efficient encoding of the participants.

* Update the coordinator binary for the new DKG

This does not yet update any tests.

* Add sensible Debug to key_gen::[Processor, Coordinator]Message

* Have the DKG explicitly declare how to interpolate its shares

Removes the hack for MuSig where we multiply keys by the inverse of their
lagrange interpolation factor.

* Replace Interpolation::None with Interpolation::Constant

Allows the MuSig DKG to keep the secret share as the original private key,
enabling deriving FROST nonces consistently regardless of the MuSig context.

* Get coordinator tests to pass

* Update spec to the new DKG

* Get clippy to pass across the repo

* cargo machete

* Add an extra sleep to ensure expected ordering of `Participation`s

* Update orchestration

* Remove bad panic in coordinator

It expected ConfirmationShare to be n-of-n, not t-of-n.

* Improve documentation on  functions

* Update TX size limit

We now no longer have to support the ridiculous case of having 49 DKG
participations within a 101-of-150 DKG. It does remain quite high due to
needing to _sign_ so many times. It'd may be optimal for parties with multiple
key shares to independently send their preprocesses/shares (despite the
overhead that'll cause with signatures and the transaction structure).

* Correct error in the Processor spec document

* Update a few comments in the validator-sets pallet

* Send/Recv Participation one at a time

Sending all, then attempting to receive all in an expected order, wasn't working
even with notable delays between sending messages. This points to the mempool
not working as expected...

* Correct ThresholdKeys serialization in modular-frost test

* Updating existing TX size limit test for the new DKG parameters

* Increase time allowed for the DKG on the GH CI

* Correct construction of signature_participants in serai-client tests

Fault identified by akil.

* Further contextualize DkgConfirmer by ValidatorSet

Caught by a safety check we wouldn't reuse preprocesses across messages. That
raises the question of we were prior reusing preprocesses (reusing keys)?
Except that'd have caused a variety of signing failures (suggesting we had some
staggered timing avoiding it in practice but yes, this was possible in theory).

* Add necessary calls to set_embedded_elliptic_curve_key in coordinator set rotation tests

* Correct shimmed setting of a secq256k1 key

* cargo fmt

* Don't use `[0; 32]` for the embedded keys in the coordinator rotation test

The key_gen function expects the random values already decided.

* Big-endian secq256k1 scalars

Also restores the prior, safer, Encryption::register function.
2024-09-19 21:43:26 -04:00

362 lines
14 KiB
Rust

/*
A MuSig-based signing protocol executed with the validators' keys.
This is used for confirming the results of a DKG on-chain, an operation requiring all validators
which aren't specified as removed while still satisfying a supermajority.
Since we're using the validator's keys, as needed for their being the root of trust, the
coordinator must perform the signing. This is distinct from all other group-signing operations,
as they're all done by the processor.
The MuSig-aggregation achieves on-chain efficiency and enables a more secure design pattern.
While we could individually tack votes, that'd require logic to prevent voting multiple times and
tracking the accumulated votes. MuSig-aggregation simply requires checking the list is sorted and
the list's weight exceeds the threshold.
Instead of maintaining state in memory, a combination of the DB and re-execution are used. This
is deemed acceptable re: performance as:
1) This is only done prior to a DKG being confirmed on Substrate and is assumed infrequent.
2) This is an O(n) algorithm.
3) The size of the validator set is bounded by MAX_KEY_SHARES_PER_SET.
Accordingly, this should be tolerable.
As for safety, it is explicitly unsafe to reuse nonces across signing sessions. This raises
concerns regarding our re-execution which is dependent on fixed nonces. Safety is derived from
the nonces being context-bound under a BFT protocol. The flow is as follows:
1) Decide the nonce.
2) Publish the nonces' commitments, receiving everyone elses *and potentially the message to be
signed*.
3) Sign and publish the signature share.
In order for nonce re-use to occur, the received nonce commitments (or the message to be signed)
would have to be distinct and sign would have to be called again.
Before we act on any received messages, they're ordered and finalized by a BFT algorithm. The
only way to operate on distinct received messages would be if:
1) A logical flaw exists, letting new messages over write prior messages
2) A reorganization occurred from chain A to chain B, and with it, different messages
Reorganizations are not supported, as BFT is assumed by the presence of a BFT algorithm. While
a significant amount of processes may be byzantine, leading to BFT being broken, that still will
not trigger a reorganization. The only way to move to a distinct chain, with distinct messages,
would be by rebuilding the local process (this time following chain B). Upon any complete
rebuild, we'd re-decide nonces, achieving safety. This does set a bound preventing partial
rebuilds which is accepted.
Additionally, to ensure a rebuilt service isn't flagged as malicious, we have to check the
commitments generated from the decided nonces are in fact its commitments on-chain (TODO).
TODO: We also need to review how we're handling Processor preprocesses and likely implement the
same on-chain-preprocess-matches-presumed-preprocess check before publishing shares.
*/
use core::ops::Deref;
use std::collections::{HashSet, HashMap};
use zeroize::{Zeroize, Zeroizing};
use rand_core::OsRng;
use blake2::{Digest, Blake2s256};
use ciphersuite::{group::ff::PrimeField, Ciphersuite, Ristretto};
use frost::{
FrostError,
dkg::{Participant, musig::musig},
ThresholdKeys,
sign::*,
};
use frost_schnorrkel::Schnorrkel;
use scale::Encode;
#[rustfmt::skip]
use serai_client::validator_sets::primitives::{ValidatorSet, KeyPair, musig_context, set_keys_message};
use serai_db::*;
use crate::tributary::TributarySpec;
create_db!(
SigningProtocolDb {
CachedPreprocesses: (context: &impl Encode) -> [u8; 32]
DataSignedWith: (context: &impl Encode) -> (Vec<u8>, HashMap<Participant, Vec<u8>>),
}
);
struct SigningProtocol<'a, T: DbTxn, C: Encode> {
pub(crate) key: &'a Zeroizing<<Ristretto as Ciphersuite>::F>,
pub(crate) spec: &'a TributarySpec,
pub(crate) txn: &'a mut T,
pub(crate) context: C,
}
impl<T: DbTxn, C: Encode> SigningProtocol<'_, T, C> {
fn preprocess_internal(
&mut self,
participants: &[<Ristretto as Ciphersuite>::G],
) -> (AlgorithmSignMachine<Ristretto, Schnorrkel>, [u8; 64]) {
// Encrypt the cached preprocess as recovery of it will enable recovering the private key
// While the DB isn't expected to be arbitrarily readable, it isn't a proper secret store and
// shouldn't be trusted as one
let mut encryption_key = {
let mut encryption_key_preimage =
Zeroizing::new(b"Cached Preprocess Encryption Key".to_vec());
encryption_key_preimage.extend(self.context.encode());
let repr = Zeroizing::new(self.key.to_repr());
encryption_key_preimage.extend(repr.deref());
Blake2s256::digest(&encryption_key_preimage)
};
let encryption_key_slice: &mut [u8] = encryption_key.as_mut();
// Create the MuSig keys
let keys: ThresholdKeys<Ristretto> =
musig(&musig_context(self.spec.set()), self.key, participants)
.expect("signing for a set we aren't in/validator present multiple times")
.into();
// Define the algorithm
let algorithm = Schnorrkel::new(b"substrate");
// Check if we've prior preprocessed
if CachedPreprocesses::get(self.txn, &self.context).is_none() {
// If we haven't, we create a machine solely to obtain the preprocess with
let (machine, _) =
AlgorithmMachine::new(algorithm.clone(), keys.clone()).preprocess(&mut OsRng);
// Cache and save the preprocess to disk
let mut cache = machine.cache();
assert_eq!(cache.0.len(), 32);
#[allow(clippy::needless_range_loop)]
for b in 0 .. 32 {
cache.0[b] ^= encryption_key_slice[b];
}
CachedPreprocesses::set(self.txn, &self.context, &cache.0);
}
// We're now guaranteed to have the preprocess, hence why this `unwrap` is safe
let cached = CachedPreprocesses::get(self.txn, &self.context).unwrap();
let mut cached = Zeroizing::new(cached);
#[allow(clippy::needless_range_loop)]
for b in 0 .. 32 {
cached[b] ^= encryption_key_slice[b];
}
encryption_key_slice.zeroize();
// Create the machine from the cached preprocess
let (machine, preprocess) =
AlgorithmSignMachine::from_cache(algorithm, keys, CachedPreprocess(cached));
(machine, preprocess.serialize().try_into().unwrap())
}
fn share_internal(
&mut self,
participants: &[<Ristretto as Ciphersuite>::G],
mut serialized_preprocesses: HashMap<Participant, Vec<u8>>,
msg: &[u8],
) -> Result<(AlgorithmSignatureMachine<Ristretto, Schnorrkel>, [u8; 32]), Participant> {
// We can't clear the preprocess as we sitll need it to accumulate all of the shares
// We do save the message we signed so any future calls with distinct messages panic
// This assumes the txn deciding this data is committed before the share is broaadcast
if let Some((existing_msg, existing_preprocesses)) =
DataSignedWith::get(self.txn, &self.context)
{
assert_eq!(msg, &existing_msg, "obtaining a signature share for a distinct message");
assert_eq!(
&serialized_preprocesses, &existing_preprocesses,
"obtaining a signature share with a distinct set of preprocesses"
);
} else {
DataSignedWith::set(
self.txn,
&self.context,
&(msg.to_vec(), serialized_preprocesses.clone()),
);
}
// Get the preprocessed machine
let (machine, _) = self.preprocess_internal(participants);
// Deserialize all the preprocesses
let mut participants = serialized_preprocesses.keys().copied().collect::<Vec<_>>();
participants.sort();
let mut preprocesses = HashMap::new();
for participant in participants {
preprocesses.insert(
participant,
machine
.read_preprocess(&mut serialized_preprocesses.remove(&participant).unwrap().as_slice())
.map_err(|_| participant)?,
);
}
// Sign the share
let (machine, share) = machine.sign(preprocesses, msg).map_err(|e| match e {
FrostError::InternalError(e) => unreachable!("FrostError::InternalError {e}"),
FrostError::InvalidParticipant(_, _) |
FrostError::InvalidSigningSet(_) |
FrostError::InvalidParticipantQuantity(_, _) |
FrostError::DuplicatedParticipant(_) |
FrostError::MissingParticipant(_) => panic!("unexpected error during sign: {e:?}"),
FrostError::InvalidPreprocess(p) | FrostError::InvalidShare(p) => p,
})?;
Ok((machine, share.serialize().try_into().unwrap()))
}
fn complete_internal(
machine: AlgorithmSignatureMachine<Ristretto, Schnorrkel>,
shares: HashMap<Participant, Vec<u8>>,
) -> Result<[u8; 64], Participant> {
let shares = shares
.into_iter()
.map(|(p, share)| {
machine.read_share(&mut share.as_slice()).map(|share| (p, share)).map_err(|_| p)
})
.collect::<Result<HashMap<_, _>, _>>()?;
let signature = machine.complete(shares).map_err(|e| match e {
FrostError::InternalError(e) => unreachable!("FrostError::InternalError {e}"),
FrostError::InvalidParticipant(_, _) |
FrostError::InvalidSigningSet(_) |
FrostError::InvalidParticipantQuantity(_, _) |
FrostError::DuplicatedParticipant(_) |
FrostError::MissingParticipant(_) => unreachable!("{e:?}"),
FrostError::InvalidPreprocess(p) | FrostError::InvalidShare(p) => p,
})?;
Ok(signature.to_bytes())
}
}
// Get the keys of the participants, noted by their threshold is, and return a new map indexed by
// their MuSig is.
fn threshold_i_map_to_keys_and_musig_i_map(
spec: &TributarySpec,
our_key: &Zeroizing<<Ristretto as Ciphersuite>::F>,
mut map: HashMap<Participant, Vec<u8>>,
) -> (Vec<<Ristretto as Ciphersuite>::G>, HashMap<Participant, Vec<u8>>) {
// Insert our own index so calculations aren't offset
let our_threshold_i = spec
.i(<Ristretto as Ciphersuite>::generator() * our_key.deref())
.expect("not in a set we're signing for")
.start;
// Asserts we weren't unexpectedly already present
assert!(map.insert(our_threshold_i, vec![]).is_none());
let spec_validators = spec.validators();
let key_from_threshold_i = |threshold_i| {
for (key, _) in &spec_validators {
if threshold_i == spec.i(*key).expect("validator wasn't in a set they're in").start {
return *key;
}
}
panic!("requested info for threshold i which doesn't exist")
};
let mut sorted = vec![];
let mut threshold_is = map.keys().copied().collect::<Vec<_>>();
threshold_is.sort();
for threshold_i in threshold_is {
sorted.push((
threshold_i,
key_from_threshold_i(threshold_i),
map.remove(&threshold_i).unwrap(),
));
}
// Now that signers are sorted, with their shares, create a map with the is needed for MuSig
let mut participants = vec![];
let mut map = HashMap::new();
let mut our_musig_i = None;
for (raw_i, (threshold_i, key, share)) in sorted.into_iter().enumerate() {
let musig_i = Participant::new(u16::try_from(raw_i).unwrap() + 1).unwrap();
if threshold_i == our_threshold_i {
our_musig_i = Some(musig_i);
}
participants.push(key);
map.insert(musig_i, share);
}
map.remove(&our_musig_i.unwrap()).unwrap();
(participants, map)
}
type DkgConfirmerSigningProtocol<'a, T> =
SigningProtocol<'a, T, (&'static [u8; 12], ValidatorSet, u32)>;
pub(crate) struct DkgConfirmer<'a, T: DbTxn> {
key: &'a Zeroizing<<Ristretto as Ciphersuite>::F>,
spec: &'a TributarySpec,
txn: &'a mut T,
attempt: u32,
}
impl<T: DbTxn> DkgConfirmer<'_, T> {
pub(crate) fn new<'a>(
key: &'a Zeroizing<<Ristretto as Ciphersuite>::F>,
spec: &'a TributarySpec,
txn: &'a mut T,
attempt: u32,
) -> DkgConfirmer<'a, T> {
DkgConfirmer { key, spec, txn, attempt }
}
fn signing_protocol(&mut self) -> DkgConfirmerSigningProtocol<'_, T> {
let context = (b"DkgConfirmer", self.spec.set(), self.attempt);
SigningProtocol { key: self.key, spec: self.spec, txn: self.txn, context }
}
fn preprocess_internal(&mut self) -> (AlgorithmSignMachine<Ristretto, Schnorrkel>, [u8; 64]) {
// This preprocesses with just us as we only decide the participants after obtaining
// preprocesses
let participants = vec![<Ristretto as Ciphersuite>::generator() * self.key.deref()];
self.signing_protocol().preprocess_internal(&participants)
}
// Get the preprocess for this confirmation.
pub(crate) fn preprocess(&mut self) -> [u8; 64] {
self.preprocess_internal().1
}
fn share_internal(
&mut self,
preprocesses: HashMap<Participant, Vec<u8>>,
key_pair: &KeyPair,
) -> Result<(AlgorithmSignatureMachine<Ristretto, Schnorrkel>, [u8; 32]), Participant> {
let (participants, preprocesses) =
threshold_i_map_to_keys_and_musig_i_map(self.spec, self.key, preprocesses);
let msg = set_keys_message(&self.spec.set(), key_pair);
self.signing_protocol().share_internal(&participants, preprocesses, &msg)
}
// Get the share for this confirmation, if the preprocesses are valid.
pub(crate) fn share(
&mut self,
preprocesses: HashMap<Participant, Vec<u8>>,
key_pair: &KeyPair,
) -> Result<[u8; 32], Participant> {
self.share_internal(preprocesses, key_pair).map(|(_, share)| share)
}
pub(crate) fn complete(
&mut self,
preprocesses: HashMap<Participant, Vec<u8>>,
key_pair: &KeyPair,
shares: HashMap<Participant, Vec<u8>>,
) -> Result<[u8; 64], Participant> {
assert_eq!(preprocesses.keys().collect::<HashSet<_>>(), shares.keys().collect::<HashSet<_>>());
let shares = threshold_i_map_to_keys_and_musig_i_map(self.spec, self.key, shares).1;
let machine = self
.share_internal(preprocesses, key_pair)
.expect("trying to complete a machine which failed to preprocess")
.0;
DkgConfirmerSigningProtocol::<'_, T>::complete_internal(machine, shares)
}
}