nimbus-eth2/beacon_chain/spec/state_transition_epoch.nim

1231 lines
53 KiB
Nim

# beacon_chain
# Copyright (c) 2018-2023 Status Research & Development GmbH
# Licensed and distributed under either of
# * MIT license (license terms in the root directory or at https://opensource.org/licenses/MIT).
# * Apache v2 license (license terms in the root directory or at https://www.apache.org/licenses/LICENSE-2.0).
# at your option. This file may not be copied, modified, or distributed except according to those terms.
# State transition - epoch processing, as described in
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/beacon-chain.md#epoch-processing
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#epoch-processing
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/bellatrix/beacon-chain.md#epoch-processing
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/capella/beacon-chain.md#epoch-processing
#
# The entry point is `process_epoch`, which is at the bottom of this file.
#
# General notes about the code:
# * Weird styling - the sections taken from the spec use python styling while
# the others use NEP-1 - helps grepping identifiers in spec
# * When updating the code, add TODO sections to mark where there are clear
# improvements to be made - other than that, keep things similar to spec unless
# motivated by security or performance considerations
{.push raises: [].}
import
stew/bitops2, chronicles,
../extras,
"."/[beaconstate, eth2_merkleization, validator]
from std/math import sum, `^`
from ./datatypes/capella import
BeaconState, HistoricalSummary, Withdrawal, WithdrawalIndex
export extras, phase0, altair
# Logging utilities
# --------------------------------------------------------
logScope: topics = "consens"
# Accessors that implement the max condition in `get_total_balance`:
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#get_total_balance
template current_epoch*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.current_epoch_raw)
template previous_epoch*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.previous_epoch_raw)
template current_epoch_attesters*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.current_epoch_attesters_raw)
template current_epoch_target_attesters*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.current_epoch_target_attesters_raw)
template previous_epoch_attesters*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.previous_epoch_attesters_raw)
template previous_epoch_target_attesters*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.previous_epoch_target_attesters_raw)
template previous_epoch_head_attesters*(v: TotalBalances): Gwei =
max(EFFECTIVE_BALANCE_INCREMENT, v.previous_epoch_head_attesters_raw)
func init*(info: var phase0.EpochInfo, state: phase0.BeaconState) =
info.balances = TotalBalances()
info.validators.setLen(state.validators.len)
for i in 0..<state.validators.len:
let v = unsafeAddr state.validators[i]
var flags: set[RewardFlags]
if v[].slashed:
flags.incl(RewardFlags.isSlashed)
if state.get_current_epoch() >= v[].withdrawable_epoch:
flags.incl RewardFlags.canWithdrawInCurrentEpoch
if v[].is_active_validator(state.get_current_epoch()):
info.balances.current_epoch_raw += v[].effective_balance
if v[].is_active_validator(state.get_previous_epoch()):
flags.incl RewardFlags.isActiveInPreviousEpoch
info.balances.previous_epoch_raw += v[].effective_balance
info.validators[i] = RewardStatus(
current_epoch_effective_balance: v[].effective_balance,
flags: flags,
)
func add(a: var RewardDelta, b: RewardDelta) =
a.rewards += b.rewards
a.penalties += b.penalties
func process_attestation(
info: var phase0.EpochInfo, state: phase0.BeaconState, a: PendingAttestation,
cache: var StateCache) =
# Collect information about the attestation
var
flags: set[RewardFlags]
is_previous_epoch_attester: Opt[InclusionInfo]
if a.data.target.epoch == state.get_current_epoch():
flags.incl RewardFlags.isCurrentEpochAttester
if a.data.target.root == get_block_root(state, state.get_current_epoch()):
flags.incl RewardFlags.isCurrentEpochTargetAttester
elif a.data.target.epoch == state.get_previous_epoch():
is_previous_epoch_attester = Opt.some(InclusionInfo(
delay: a.inclusion_delay,
proposer_index: a.proposer_index,
))
if a.data.target.root == get_block_root(state, state.get_previous_epoch()):
flags.incl RewardFlags.isPreviousEpochTargetAttester
if a.data.beacon_block_root == get_block_root_at_slot(state, a.data.slot):
flags.incl RewardFlags.isPreviousEpochHeadAttester
# Update the cache for all participants
for validator_index in get_attesting_indices_iter(
state, a.data, a.aggregation_bits, cache):
template v(): untyped = info.validators[validator_index]
v.flags = v.flags + flags
if is_previous_epoch_attester.isSome:
if v.is_previous_epoch_attester.isSome:
if is_previous_epoch_attester.get().delay <
v.is_previous_epoch_attester.get().delay:
v.is_previous_epoch_attester = is_previous_epoch_attester
else:
v.is_previous_epoch_attester = is_previous_epoch_attester
func process_attestations*(
info: var phase0.EpochInfo, state: phase0.BeaconState, cache: var StateCache) =
# Walk state attestations and update the status information
for a in state.previous_epoch_attestations:
process_attestation(info, state, a, cache)
for a in state.current_epoch_attestations:
process_attestation(info, state, a, cache)
for idx, v in info.validators:
if v.flags.contains RewardFlags.isSlashed:
continue
let validator_balance = state.validators[idx].effective_balance
if v.flags.contains RewardFlags.isCurrentEpochAttester:
info.balances.current_epoch_attesters_raw += validator_balance
if v.flags.contains RewardFlags.isCurrentEpochTargetAttester:
info.balances.current_epoch_target_attesters_raw += validator_balance
if v.is_previous_epoch_attester.isSome():
info.balances.previous_epoch_attesters_raw += validator_balance
if v.flags.contains RewardFlags.isPreviousEpochTargetAttester:
info.balances.previous_epoch_target_attesters_raw += validator_balance
if v.flags.contains RewardFlags.isPreviousEpochHeadAttester:
info.balances.previous_epoch_head_attesters_raw += validator_balance
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#helpers
# get_eligible_validator_indices
func is_eligible_validator*(validator: RewardStatus): bool =
validator.flags.contains(RewardFlags.isActiveInPreviousEpoch) or
(validator.flags.contains(RewardFlags.isSlashed) and not
(validator.flags.contains RewardFlags.canWithdrawInCurrentEpoch))
func is_eligible_validator*(validator: Validator, previous_epoch: Epoch): bool =
is_active_validator(validator, previous_epoch) or
(validator.slashed and previous_epoch + 1 < validator.withdrawable_epoch)
func is_eligible_validator*(validator: ParticipationInfo): bool =
validator.flags.contains(ParticipationFlag.eligible)
# Spec
# --------------------------------------------------------
from ./datatypes/deneb import BeaconState
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/altair/beacon-chain.md#get_unslashed_participating_indices
func get_unslashed_participating_balances*(
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState): UnslashedParticipatingBalances =
let
previous_epoch = get_previous_epoch(state)
current_epoch = get_current_epoch(state)
var res: UnslashedParticipatingBalances
for validator_index in 0'u64 ..< state.validators.lenu64:
let
is_active_current_epoch = is_active_validator(
state.validators[validator_index], current_epoch)
validator_effective_balance =
state.validators[validator_index].effective_balance
if is_active_current_epoch:
# Active balance counted also for slashed validators
res.current_epoch += validator_effective_balance
if state.validators[validator_index].slashed:
continue
let
is_active_previous_epoch = is_active_validator(
state.validators[validator_index], previous_epoch)
previous_epoch_participation =
state.previous_epoch_participation[validator_index]
if is_active_previous_epoch:
for flag_index in TimelyFlag:
if has_flag(previous_epoch_participation, flag_index):
res.previous_epoch[flag_index] += validator_effective_balance
# Only TIMELY_TARGET_FLAG_INDEX is used with the current epoch in Altair
# and merge
if is_active_current_epoch and has_flag(
state.current_epoch_participation[validator_index],
TIMELY_TARGET_FLAG_INDEX):
res.current_epoch_TIMELY_TARGET += validator_effective_balance
for flag_index in TimelyFlag:
res.previous_epoch[flag_index] =
max(EFFECTIVE_BALANCE_INCREMENT, res.previous_epoch[flag_index])
res.current_epoch_TIMELY_TARGET =
max(EFFECTIVE_BALANCE_INCREMENT, res.current_epoch_TIMELY_TARGET)
res.current_epoch = max(EFFECTIVE_BALANCE_INCREMENT, res.current_epoch)
res
func is_unslashed_participating_index(
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState,
flag_index: TimelyFlag, epoch: Epoch, validator_index: ValidatorIndex): bool =
doAssert epoch in [get_previous_epoch(state), get_current_epoch(state)]
# TODO hoist this conditional
let epoch_participation =
if epoch == get_current_epoch(state):
unsafeAddr state.current_epoch_participation
else:
unsafeAddr state.previous_epoch_participation
is_active_validator(state.validators[validator_index], epoch) and
has_flag(epoch_participation[].item(validator_index), flag_index) and
not state.validators[validator_index].slashed
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#justification-and-finalization
type FinalityState = object
slot: Slot
current_epoch_ancestor_root: Eth2Digest
previous_epoch_ancestor_root: Eth2Digest
justification_bits: JustificationBits
previous_justified_checkpoint: Checkpoint
current_justified_checkpoint: Checkpoint
finalized_checkpoint: Checkpoint
func toFinalityState(state: ForkyBeaconState): FinalityState =
let
current_epoch = get_current_epoch(state)
previous_epoch = get_previous_epoch(state)
FinalityState(
slot: state.slot,
current_epoch_ancestor_root:
if state.slot > current_epoch.start_slot:
get_block_root(state, current_epoch)
else:
ZERO_HASH,
previous_epoch_ancestor_root:
if state.slot > previous_epoch.start_slot:
get_block_root(state, previous_epoch)
else:
ZERO_HASH,
justification_bits:
state.justification_bits,
previous_justified_checkpoint:
state.previous_justified_checkpoint,
current_justified_checkpoint:
state.current_justified_checkpoint,
finalized_checkpoint:
state.finalized_checkpoint)
func get_current_epoch(state: FinalityState): Epoch =
state.slot.epoch
func get_previous_epoch(state: FinalityState): Epoch =
get_previous_epoch(get_current_epoch(state))
func get_block_root(state: FinalityState, epoch: Epoch): Eth2Digest =
doAssert state.slot > epoch.start_slot
if epoch == get_current_epoch(state):
state.current_epoch_ancestor_root
else:
doAssert epoch == get_previous_epoch(state)
state.previous_epoch_ancestor_root
proc weigh_justification_and_finalization(
state: var (ForkyBeaconState | FinalityState),
total_active_balance: Gwei,
previous_epoch_target_balance: Gwei,
current_epoch_target_balance: Gwei,
flags: UpdateFlags = {}) =
let
previous_epoch = get_previous_epoch(state)
current_epoch = get_current_epoch(state)
old_previous_justified_checkpoint = state.previous_justified_checkpoint
old_current_justified_checkpoint = state.current_justified_checkpoint
# Process justifications
state.previous_justified_checkpoint = state.current_justified_checkpoint
## Spec:
## state.justification_bits[1:] = state.justification_bits[:-1]
## state.justification_bits[0] = 0b0
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#misc
const JUSTIFICATION_BITS_LENGTH = 4
state.justification_bits = JustificationBits(
(uint8(state.justification_bits) shl 1) and
uint8((2^JUSTIFICATION_BITS_LENGTH) - 1))
if previous_epoch_target_balance * 3 >= total_active_balance * 2:
state.current_justified_checkpoint =
Checkpoint(epoch: previous_epoch,
root: get_block_root(state, previous_epoch))
uint8(state.justification_bits).setBit 1
trace "Justified with previous epoch",
current_epoch = current_epoch,
checkpoint = shortLog(state.current_justified_checkpoint)
elif strictVerification in flags:
fatal "Low attestation participation in previous epoch",
total_active_balance,
previous_epoch_target_balance,
current_epoch_target_balance,
epoch = get_current_epoch(state)
quit 1
if current_epoch_target_balance * 3 >= total_active_balance * 2:
state.current_justified_checkpoint =
Checkpoint(epoch: current_epoch,
root: get_block_root(state, current_epoch))
uint8(state.justification_bits).setBit 0
trace "Justified with current epoch",
current_epoch = current_epoch,
checkpoint = shortLog(state.current_justified_checkpoint)
# Process finalizations
let bitfield = uint8(state.justification_bits)
## The 2nd/3rd/4th most recent epochs are justified, the 2nd using the 4th
## as source
if (bitfield and 0b1110) == 0b1110 and
old_previous_justified_checkpoint.epoch + 3 == current_epoch:
state.finalized_checkpoint = old_previous_justified_checkpoint
trace "Finalized with rule 234",
current_epoch = current_epoch,
checkpoint = shortLog(state.finalized_checkpoint)
## The 2nd/3rd most recent epochs are justified, the 2nd using the 3rd as
## source
if (bitfield and 0b110) == 0b110 and
old_previous_justified_checkpoint.epoch + 2 == current_epoch:
state.finalized_checkpoint = old_previous_justified_checkpoint
trace "Finalized with rule 23",
current_epoch = current_epoch,
checkpoint = shortLog(state.finalized_checkpoint)
## The 1st/2nd/3rd most recent epochs are justified, the 1st using the 3rd as
## source
if (bitfield and 0b111) == 0b111 and
old_current_justified_checkpoint.epoch + 2 == current_epoch:
state.finalized_checkpoint = old_current_justified_checkpoint
trace "Finalized with rule 123",
current_epoch = current_epoch,
checkpoint = shortLog(state.finalized_checkpoint)
## The 1st/2nd most recent epochs are justified, the 1st using the 2nd as
## source
if (bitfield and 0b11) == 0b11 and
old_current_justified_checkpoint.epoch + 1 == current_epoch:
state.finalized_checkpoint = old_current_justified_checkpoint
trace "Finalized with rule 12",
current_epoch = current_epoch,
checkpoint = shortLog(state.finalized_checkpoint)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#justification-and-finalization
proc process_justification_and_finalization*(
state: var phase0.BeaconState,
balances: TotalBalances, flags: UpdateFlags = {}) =
# Initial FFG checkpoint values have a `0x00` stub for `root`.
# Skip FFG updates in the first two epochs to avoid corner cases that might
# result in modifying this stub.
if get_current_epoch(state) <= GENESIS_EPOCH + 1:
return
weigh_justification_and_finalization(
state, balances.current_epoch,
balances.previous_epoch_target_attesters,
balances.current_epoch_target_attesters, flags)
proc compute_unrealized_finality*(
state: phase0.BeaconState, cache: var StateCache): FinalityCheckpoints =
if get_current_epoch(state) <= GENESIS_EPOCH + 1:
return FinalityCheckpoints(
justified: state.current_justified_checkpoint,
finalized: state.finalized_checkpoint)
var info: phase0.EpochInfo
info.init(state)
info.process_attestations(state, cache)
template balances(): auto = info.balances
var finalityState = state.toFinalityState()
weigh_justification_and_finalization(
finalityState, balances.current_epoch,
balances.previous_epoch_target_attesters,
balances.current_epoch_target_attesters)
FinalityCheckpoints(
justified: finalityState.current_justified_checkpoint,
finalized: finalityState.finalized_checkpoint)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#justification-and-finalization
proc process_justification_and_finalization*(
state: var (altair.BeaconState | bellatrix.BeaconState |
capella.BeaconState | deneb.BeaconState),
balances: UnslashedParticipatingBalances,
flags: UpdateFlags = {}) =
# Initial FFG checkpoint values have a `0x00` stub for `root`.
# Skip FFG updates in the first two epochs to avoid corner cases that might
# result in modifying this stub.
if get_current_epoch(state) <= GENESIS_EPOCH + 1:
return
weigh_justification_and_finalization(
state, balances.current_epoch,
balances.previous_epoch[TIMELY_TARGET_FLAG_INDEX],
balances.current_epoch_TIMELY_TARGET, flags)
proc compute_unrealized_finality*(
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState): FinalityCheckpoints =
if get_current_epoch(state) <= GENESIS_EPOCH + 1:
return FinalityCheckpoints(
justified: state.current_justified_checkpoint,
finalized: state.finalized_checkpoint)
let balances = get_unslashed_participating_balances(state)
var finalityState = state.toFinalityState()
weigh_justification_and_finalization(
finalityState, balances.current_epoch,
balances.previous_epoch[TIMELY_TARGET_FLAG_INDEX],
balances.current_epoch_TIMELY_TARGET)
FinalityCheckpoints(
justified: finalityState.current_justified_checkpoint,
finalized: finalityState.finalized_checkpoint)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#helpers
func get_base_reward_sqrt*(state: phase0.BeaconState, index: ValidatorIndex,
total_balance_sqrt: auto): Gwei =
# Spec function recalculates total_balance every time, which creates an
# O(n^2) situation.
let effective_balance = state.validators[index].effective_balance
effective_balance * BASE_REWARD_FACTOR div
total_balance_sqrt div BASE_REWARDS_PER_EPOCH
func get_proposer_reward*(base_reward: Gwei): Gwei =
# Spec version recalculates get_total_active_balance(state) quadratically
base_reward div PROPOSER_REWARD_QUOTIENT
func is_in_inactivity_leak(finality_delay: uint64): bool =
finality_delay > MIN_EPOCHS_TO_INACTIVITY_PENALTY
func get_finality_delay*(state: ForkyBeaconState): uint64 =
get_previous_epoch(state) - state.finalized_checkpoint.epoch
func get_attestation_component_reward*(attesting_balance: Gwei,
total_balance: Gwei,
base_reward: uint64,
finality_delay: uint64): Gwei =
if is_in_inactivity_leak(finality_delay):
# Since full base reward will be canceled out by inactivity penalty deltas,
# optimal participation receives full base reward compensation here.
base_reward
else:
let reward_numerator =
base_reward * (attesting_balance div EFFECTIVE_BALANCE_INCREMENT)
reward_numerator div (total_balance div EFFECTIVE_BALANCE_INCREMENT)
func get_attestation_component_delta(is_unslashed_attester: bool,
attesting_balance: Gwei,
total_balance: Gwei,
base_reward: uint64,
finality_delay: uint64): RewardDelta =
# Helper with shared logic for use by get source, target, and head deltas
# functions
if is_unslashed_attester:
RewardDelta(rewards: get_attestation_component_reward(
attesting_balance,
total_balance,
base_reward,
finality_delay))
else:
RewardDelta(penalties: base_reward)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#components-of-attestation-deltas
func get_source_delta*(validator: RewardStatus,
base_reward: uint64,
balances: TotalBalances,
finality_delay: uint64): RewardDelta =
## Return attester micro-rewards/penalties for source-vote for each validator.
get_attestation_component_delta(
validator.is_previous_epoch_attester.isSome() and
not (validator.flags.contains RewardFlags.isSlashed),
balances.previous_epoch_attesters,
balances.current_epoch,
base_reward,
finality_delay)
func get_target_delta*(validator: RewardStatus,
base_reward: uint64,
balances: TotalBalances,
finality_delay: uint64): RewardDelta =
## Return attester micro-rewards/penalties for target-vote for each validator.
get_attestation_component_delta(
validator.flags.contains(RewardFlags.isPreviousEpochTargetAttester) and
not (validator.flags.contains(RewardFlags.isSlashed)),
balances.previous_epoch_target_attesters,
balances.current_epoch,
base_reward,
finality_delay)
func get_head_delta*(validator: RewardStatus,
base_reward: uint64,
balances: TotalBalances,
finality_delay: uint64): RewardDelta =
## Return attester micro-rewards/penalties for head-vote for each validator.
get_attestation_component_delta(
validator.flags.contains(RewardFlags.isPreviousEpochHeadAttester) and
((not validator.flags.contains(RewardFlags.isSlashed))),
balances.previous_epoch_head_attesters,
balances.current_epoch,
base_reward,
finality_delay)
func get_inclusion_delay_delta*(validator: RewardStatus,
base_reward: uint64):
(RewardDelta, Opt[(uint64, RewardDelta)]) =
## Return proposer and inclusion delay micro-rewards/penalties for each validator.
if validator.is_previous_epoch_attester.isSome() and ((not validator.flags.contains(RewardFlags.isSlashed))):
let
inclusion_info = validator.is_previous_epoch_attester.get()
proposer_reward = get_proposer_reward(base_reward)
proposer_delta = RewardDelta(rewards: proposer_reward)
let
max_attester_reward = base_reward - proposer_reward
delta = RewardDelta(rewards: max_attester_reward div inclusion_info.delay)
proposer_index = inclusion_info.proposer_index;
return (delta, Opt.some((proposer_index, proposer_delta)))
func get_inactivity_penalty_delta*(validator: RewardStatus,
base_reward: Gwei,
finality_delay: uint64): RewardDelta =
## Return inactivity reward/penalty deltas for each validator.
var delta: RewardDelta
if is_in_inactivity_leak(finality_delay):
# If validator is performing optimally this cancels all rewards for a neutral balance
delta.penalties +=
BASE_REWARDS_PER_EPOCH * base_reward - get_proposer_reward(base_reward)
# Additionally, all validators whose FFG target didn't match are penalized extra
# This condition is equivalent to this condition from the spec:
# `index not in get_unslashed_attesting_indices(state, matching_target_attestations)`
if (validator.flags.contains(RewardFlags.isSlashed)) or
((not validator.flags.contains(RewardFlags.isPreviousEpochTargetAttester))):
delta.penalties +=
validator.current_epoch_effective_balance * finality_delay div
INACTIVITY_PENALTY_QUOTIENT
delta
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#get_attestation_deltas
func get_attestation_deltas(
state: phase0.BeaconState, info: var phase0.EpochInfo) =
## Update rewards with attestation reward/penalty deltas for each validator.
let
finality_delay = get_finality_delay(state)
total_balance = info.balances.current_epoch
total_balance_sqrt = integer_squareroot(total_balance)
# Filter out ineligible validators. All sub-functions of the spec do this
# except for `get_inclusion_delay_deltas`. It's safe to do so here because
# any validator that is in the unslashed indices of the matching source
# attestations is active, and therefore eligible.
for index, validator in info.validators.mpairs():
if not is_eligible_validator(validator):
continue
let
base_reward = get_base_reward_sqrt(
state, index.ValidatorIndex, total_balance_sqrt)
let
source_delta = get_source_delta(
validator, base_reward, info.balances, finality_delay)
target_delta = get_target_delta(
validator, base_reward, info.balances, finality_delay)
head_delta = get_head_delta(
validator, base_reward, info.balances, finality_delay)
(inclusion_delay_delta, proposer_delta) =
get_inclusion_delay_delta(validator, base_reward)
inactivity_delta = get_inactivity_penalty_delta(
validator, base_reward, finality_delay)
validator.delta.add source_delta
validator.delta.add target_delta
validator.delta.add head_delta
validator.delta.add inclusion_delay_delta
validator.delta.add inactivity_delta
if proposer_delta.isSome:
let proposer_index = proposer_delta.get()[0]
if proposer_index < info.validators.lenu64:
info.validators[proposer_index].delta.add(
proposer_delta.get()[1])
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#get_base_reward
func get_base_reward_increment*(
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState,
index: ValidatorIndex, base_reward_per_increment: Gwei): Gwei =
## Return the base reward for the validator defined by ``index`` with respect
## to the current ``state``.
let increments =
state.validators[index].effective_balance div EFFECTIVE_BALANCE_INCREMENT
increments * base_reward_per_increment
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#get_flag_index_deltas
func get_flag_index_reward*(
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState,
base_reward: Gwei, active_increments: Gwei,
unslashed_participating_increments: Gwei,
weight, finality_delay: uint64): Gwei =
if not is_in_inactivity_leak(finality_delay):
let reward_numerator =
base_reward * weight * unslashed_participating_increments
reward_numerator div (active_increments * WEIGHT_DENOMINATOR)
else:
0.Gwei
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#get_flag_index_deltas
func get_unslashed_participating_increment*(
info: altair.EpochInfo | bellatrix.BeaconState, flag_index: TimelyFlag): Gwei =
info.balances.previous_epoch[flag_index] div EFFECTIVE_BALANCE_INCREMENT
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#get_flag_index_deltas
func get_active_increments*(
info: altair.EpochInfo | bellatrix.BeaconState): Gwei =
info.balances.current_epoch div EFFECTIVE_BALANCE_INCREMENT
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#get_flag_index_deltas
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#modified-get_inactivity_penalty_deltas
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/bellatrix/beacon-chain.md#modified-get_inactivity_penalty_deltas
# Combines get_flag_index_deltas() and get_inactivity_penalty_deltas()
iterator get_flag_and_inactivity_deltas*(
cfg: RuntimeConfig,
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState,
base_reward_per_increment: Gwei, info: var altair.EpochInfo,
finality_delay: uint64):
(ValidatorIndex, Gwei, Gwei, Gwei, Gwei, Gwei, Gwei) =
## Return the deltas for a given ``flag_index`` by scanning through the
## participation flags.
#
# This deviates from spec by processing all flags at once, so does not take a
# flag_index parameter. Fold get_inactivity_penalty_deltas loop into this one
# as well.
const INACTIVITY_PENALTY_QUOTIENT =
when state is altair.BeaconState:
INACTIVITY_PENALTY_QUOTIENT_ALTAIR
else:
INACTIVITY_PENALTY_QUOTIENT_BELLATRIX
static: doAssert ord(high(TimelyFlag)) == 2
let
previous_epoch = get_previous_epoch(state)
active_increments = get_active_increments(info)
penalty_denominator =
cfg.INACTIVITY_SCORE_BIAS * INACTIVITY_PENALTY_QUOTIENT
epoch_participation =
if previous_epoch == get_current_epoch(state):
unsafeAddr state.current_epoch_participation
else:
unsafeAddr state.previous_epoch_participation
participating_increments = [
get_unslashed_participating_increment(info, TIMELY_SOURCE_FLAG_INDEX),
get_unslashed_participating_increment(info, TIMELY_TARGET_FLAG_INDEX),
get_unslashed_participating_increment(info, TIMELY_HEAD_FLAG_INDEX)]
for vidx in state.validators.vindices:
if not is_eligible_validator(info.validators[vidx]):
continue
let
base_reward = get_base_reward_increment(state, vidx, base_reward_per_increment)
pflags =
if is_active_validator(state.validators[vidx], previous_epoch) and
not state.validators[vidx].slashed:
epoch_participation[].item(vidx)
else:
0
if has_flag(pflags, TIMELY_SOURCE_FLAG_INDEX):
info.validators[vidx].flags.incl ParticipationFlag.timelySourceAttester
if has_flag(pflags, TIMELY_TARGET_FLAG_INDEX):
info.validators[vidx].flags.incl ParticipationFlag.timelyTargetAttester
if has_flag(pflags, TIMELY_HEAD_FLAG_INDEX):
info.validators[vidx].flags.incl ParticipationFlag.timelyHeadAttester
template reward(flag: untyped): untyped =
if has_flag(pflags, flag):
get_flag_index_reward(
state, base_reward, active_increments,
participating_increments[ord(flag)],
PARTICIPATION_FLAG_WEIGHTS[flag], finality_delay)
else:
0
template penalty(flag: untyped): untyped =
if not has_flag(pflags, flag):
base_reward * PARTICIPATION_FLAG_WEIGHTS[flag] div WEIGHT_DENOMINATOR
else:
0
let inactivity_penalty =
if has_flag(pflags, TIMELY_TARGET_FLAG_INDEX):
0.Gwei
else:
let penalty_numerator = state.validators[vidx].effective_balance * state.inactivity_scores[vidx]
penalty_numerator div penalty_denominator
# Yielding these as a structure with identifiable names rather than
# multiple-return-value style creates spurious nimZeroMem calls.
yield
(vidx, reward(TIMELY_SOURCE_FLAG_INDEX),
reward(TIMELY_TARGET_FLAG_INDEX), reward(TIMELY_HEAD_FLAG_INDEX),
penalty(TIMELY_SOURCE_FLAG_INDEX), penalty(TIMELY_TARGET_FLAG_INDEX),
inactivity_penalty)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/beacon-chain.md#rewards-and-penalties-1
func process_rewards_and_penalties*(
state: var phase0.BeaconState, info: var phase0.EpochInfo) =
# No rewards are applied at the end of `GENESIS_EPOCH` because rewards are
# for work done in the previous epoch
doAssert info.validators.len == state.validators.len
if get_current_epoch(state) == GENESIS_EPOCH:
return
get_attestation_deltas(state, info)
# Here almost all balances are updated (assuming most validators are active) -
# clearing the cache becomes a bottleneck if done item by item because of the
# recursive nature of cache clearing - instead, we clear the whole cache then
# update the raw list directly
state.balances.clearCache()
for idx, v in info.validators:
var balance = state.balances.item(idx)
increase_balance(balance, v.delta.rewards)
decrease_balance(balance, v.delta.penalties)
state.balances.asSeq()[idx] = balance
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/altair/beacon-chain.md#rewards-and-penalties
func process_rewards_and_penalties*(
cfg: RuntimeConfig,
state: var (altair.BeaconState | bellatrix.BeaconState |
capella.BeaconState | deneb.BeaconState),
info: var altair.EpochInfo) =
if get_current_epoch(state) == GENESIS_EPOCH:
return
let
total_active_balance = info.balances.current_epoch
base_reward_per_increment = get_base_reward_per_increment(
total_active_balance)
finality_delay = get_finality_delay(state)
doAssert state.validators.len() == info.validators.len()
for validator_index, reward0, reward1, reward2, penalty0, penalty1, penalty2 in
get_flag_and_inactivity_deltas(
cfg, state, base_reward_per_increment, info, finality_delay):
info.validators[validator_index].delta.rewards += reward0 + reward1 + reward2
info.validators[validator_index].delta.penalties += penalty0 + penalty1 + penalty2
# Here almost all balances are updated (assuming most validators are active) -
# clearing the cache becomes a bottleneck if done item by item because of the
# recursive nature of cache clearing - instead, we clear the whole cache then
# update the raw list directly
state.balances.clearCache()
for vidx in state.validators.vindices:
var balance = state.balances.item(vidx)
increase_balance(balance, info.validators[vidx].delta.rewards)
decrease_balance(balance, info.validators[vidx].delta.penalties)
state.balances.asSeq()[vidx] = balance
from std/heapqueue import HeapQueue, `[]`, len, push, replace
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#registry-updates
func process_registry_updates*(
cfg: RuntimeConfig, state: var ForkyBeaconState, cache: var StateCache):
Result[void, cstring] =
## Process activation eligibility and ejections
# Make visible, e.g.,
# https://github.com/status-im/nimbus-eth2/pull/608
# https://github.com/sigp/lighthouse/pull/657
let epoch {.used.} = get_current_epoch(state)
trace "process_registry_updates validator balances",
balances=state.balances,
active_validator_indices=get_active_validator_indices(state, epoch),
epoch=epoch
# is_active_validator(...) is activation_epoch <= epoch < exit_epoch,
# and changes here to either activation_epoch or exit_epoch only take
# effect with a compute_activation_exit_epoch(...) delay of, based on
# the current epoch, 1 + MAX_SEED_LOOKAHEAD epochs ahead. Thus caches
# remain valid for this epoch through though this function along with
# the rest of the epoch transition.
#
# This implementation fuses the two loops over all validators in the
# spec code.
## Queue validators eligible for activation and not dequeued for activation
var activation_queue: HeapQueue[(uint64, uint32)]
let churn_limit =
when typeof(state).toFork >= ConsensusFork.Deneb:
get_validator_activation_churn_limit(cfg, state, cache)
else:
get_validator_churn_limit(cfg, state, cache)
for vidx in state.validators.vindices:
if is_eligible_for_activation_queue(state.validators.item(vidx)):
state.validators.mitem(vidx).activation_eligibility_epoch =
get_current_epoch(state) + 1
if is_active_validator(state.validators.item(vidx), get_current_epoch(state)) and
state.validators.item(vidx).effective_balance <= cfg.EJECTION_BALANCE:
? initiate_validator_exit(cfg, state, vidx, cache)
let validator = unsafeAddr state.validators.item(vidx)
if is_eligible_for_activation(state, validator[]):
let val_key =
(FAR_FUTURE_EPOCH - validator[].activation_eligibility_epoch,
high(distinctBase(ValidatorIndex)) - distinctBase(vidx))
if activation_queue.len.uint64 < churn_limit:
activation_queue.push val_key
elif val_key > activation_queue[0]:
discard activation_queue.replace val_key
## Dequeued validators for activation up to activation churn limit
## (without resetting activation epoch)
doAssert activation_queue.len.uint64 <= churn_limit
for i in 0 ..< activation_queue.len:
let (_, vidx_complement) = activation_queue[i]
state.validators.mitem(
high(distinctBase(ValidatorIndex)) - vidx_complement).activation_epoch =
compute_activation_exit_epoch(get_current_epoch(state))
ok()
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/bellatrix/beacon-chain.md#slashings
func get_adjusted_total_slashing_balance*(
state: ForkyBeaconState, total_balance: Gwei): Gwei =
const multiplier =
# tradeoff here about interleaving phase0/altair, but for these
# single-constant changes...
when state is phase0.BeaconState:
PROPORTIONAL_SLASHING_MULTIPLIER
elif state is altair.BeaconState:
PROPORTIONAL_SLASHING_MULTIPLIER_ALTAIR
elif state is bellatrix.BeaconState or state is capella.BeaconState or
state is deneb.BeaconState:
PROPORTIONAL_SLASHING_MULTIPLIER_BELLATRIX
else:
{.fatal: "process_slashings: incorrect BeaconState type".}
min(sum(state.slashings.data) * multiplier, total_balance)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/bellatrix/beacon-chain.md#slashings
func slashing_penalty_applies*(validator: Validator, epoch: Epoch): bool =
validator.slashed and
epoch + EPOCHS_PER_SLASHINGS_VECTOR div 2 == validator.withdrawable_epoch
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/bellatrix/beacon-chain.md#slashings
func get_slashing_penalty*(validator: Validator,
adjusted_total_slashing_balance,
total_balance: Gwei): Gwei =
# Factored out from penalty numerator to avoid uint64 overflow
const increment = EFFECTIVE_BALANCE_INCREMENT
let penalty_numerator = validator.effective_balance div increment *
adjusted_total_slashing_balance
penalty_numerator div total_balance * increment
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#slashings
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/bellatrix/beacon-chain.md#slashings
func process_slashings*(state: var ForkyBeaconState, total_balance: Gwei) =
let
epoch = get_current_epoch(state)
adjusted_total_slashing_balance = get_adjusted_total_slashing_balance(
state, total_balance)
for vidx in state.validators.vindices:
let validator = unsafeAddr state.validators.item(vidx)
if slashing_penalty_applies(validator[], epoch):
let penalty = get_slashing_penalty(
validator[], adjusted_total_slashing_balance, total_balance)
decrease_balance(state, vidx, penalty)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#eth1-data-votes-updates
func process_eth1_data_reset*(state: var ForkyBeaconState) =
let next_epoch = get_current_epoch(state) + 1
# Reset eth1 data votes
if next_epoch mod EPOCHS_PER_ETH1_VOTING_PERIOD == 0:
state.eth1_data_votes = default(type state.eth1_data_votes)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#effective-balances-updates
func process_effective_balance_updates*(state: var ForkyBeaconState) =
# Update effective balances with hysteresis
for vidx in state.validators.vindices:
let balance = state.balances.item(vidx)
const
HYSTERESIS_INCREMENT =
EFFECTIVE_BALANCE_INCREMENT div HYSTERESIS_QUOTIENT
DOWNWARD_THRESHOLD =
HYSTERESIS_INCREMENT * HYSTERESIS_DOWNWARD_MULTIPLIER
UPWARD_THRESHOLD = HYSTERESIS_INCREMENT * HYSTERESIS_UPWARD_MULTIPLIER
let effective_balance = state.validators.item(vidx).effective_balance
if balance + DOWNWARD_THRESHOLD < effective_balance or
effective_balance + UPWARD_THRESHOLD < balance:
let new_effective_balance =
min(
balance - balance mod EFFECTIVE_BALANCE_INCREMENT,
MAX_EFFECTIVE_BALANCE)
# Protect against unnecessary cache invalidation
if new_effective_balance != effective_balance:
state.validators.mitem(vidx).effective_balance = new_effective_balance
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#slashings-balances-updates
func process_slashings_reset*(state: var ForkyBeaconState) =
let next_epoch = get_current_epoch(state) + 1
# Reset slashings
state.slashings[int(next_epoch mod EPOCHS_PER_SLASHINGS_VECTOR)] = 0.Gwei
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#randao-mixes-updates
func process_randao_mixes_reset*(state: var ForkyBeaconState) =
let
current_epoch = get_current_epoch(state)
next_epoch = current_epoch + 1
# Set randao mix
state.randao_mixes[next_epoch mod EPOCHS_PER_HISTORICAL_VECTOR] =
get_randao_mix(state, current_epoch)
func compute_historical_root*(state: var ForkyBeaconState): Eth2Digest =
# Equivalent to hash_tree_root(foo: HistoricalBatch), but without using
# significant additional stack or heap.
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#historicalbatch
# In response to https://github.com/status-im/nimbus-eth2/issues/921
hash_tree_root([
hash_tree_root(state.block_roots), hash_tree_root(state.state_roots)])
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#historical-roots-updates
func process_historical_roots_update*(state: var ForkyBeaconState) =
## Set historical root accumulator
let next_epoch = get_current_epoch(state) + 1
if next_epoch mod (SLOTS_PER_HISTORICAL_ROOT div SLOTS_PER_EPOCH) == 0:
# Equivalent to hash_tree_root(foo: HistoricalBatch), but without using
# significant additional stack or heap.
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#historicalbatch
# In response to https://github.com/status-im/nimbus-eth2/issues/921
if not state.historical_roots.add state.compute_historical_root():
raiseAssert "no more room for historical roots, so long and thanks for the fish!"
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/beacon-chain.md#participation-records-rotation
func process_participation_record_updates*(state: var phase0.BeaconState) =
# Rotate current/previous epoch attestations - using swap avoids copying all
# elements using a slow genericSeqAssign
state.previous_epoch_attestations.clear()
swap(state.previous_epoch_attestations, state.current_epoch_attestations)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#participation-flags-updates
func process_participation_flag_updates*(
state: var (altair.BeaconState | bellatrix.BeaconState |
capella.BeaconState | deneb.BeaconState)) =
state.previous_epoch_participation = state.current_epoch_participation
const zero = 0.ParticipationFlags
for i in 0 ..< state.current_epoch_participation.len:
asList(state.current_epoch_participation)[i] = zero
# Shouldn't be wasted zeroing, because state.current_epoch_participation only
# grows. New elements are automatically initialized to 0, as required.
doAssert state.current_epoch_participation.asList.setLen(state.validators.len)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#sync-committee-updates
func process_sync_committee_updates*(
state: var (altair.BeaconState | bellatrix.BeaconState |
capella.BeaconState | deneb.BeaconState)) =
let next_epoch = get_current_epoch(state) + 1
if next_epoch.is_sync_committee_period():
state.current_sync_committee = state.next_sync_committee
state.next_sync_committee = get_next_sync_committee(state)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#inactivity-scores
func process_inactivity_updates*(
cfg: RuntimeConfig,
state: var (altair.BeaconState | bellatrix.BeaconState |
capella.BeaconState | deneb.BeaconState),
info: altair.EpochInfo) =
# Score updates based on previous epoch participation, skip genesis epoch
if get_current_epoch(state) == GENESIS_EPOCH:
return
let
previous_epoch = get_previous_epoch(state) # get_eligible_validator_indices()
finality_delay = get_finality_delay(state)
not_in_inactivity_leak = not is_in_inactivity_leak(finality_delay)
for index in 0'u64 ..< state.validators.lenu64:
if not is_eligible_validator(info.validators[index]):
continue
# Increase the inactivity score of inactive validators
let pre_inactivity_score = state.inactivity_scores.asSeq()[index]
var inactivity_score = pre_inactivity_score
# TODO activeness already checked; remove redundant checks between
# is_active_validator and is_unslashed_participating_index
if is_unslashed_participating_index(
state, TIMELY_TARGET_FLAG_INDEX, previous_epoch, index.ValidatorIndex):
inactivity_score -= min(1'u64, inactivity_score)
else:
inactivity_score += cfg.INACTIVITY_SCORE_BIAS
# Decrease the inactivity score of all eligible validators during a
# leak-free epoch
if not_in_inactivity_leak:
inactivity_score -= min(INACTIVITY_SCORE_RECOVERY_RATE.uint64, inactivity_score)
# Most inactivity scores remain at 0 - avoid invalidating cache
if pre_inactivity_score != inactivity_score:
state.inactivity_scores[index] = inactivity_score
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/capella/beacon-chain.md#historical-summaries-updates
func process_historical_summaries_update*(
state: var (capella.BeaconState | deneb.BeaconState)):
Result[void, cstring] =
# Set historical block root accumulator.
let next_epoch = get_current_epoch(state) + 1
if next_epoch mod (SLOTS_PER_HISTORICAL_ROOT div SLOTS_PER_EPOCH) == 0:
let historical_summary = HistoricalSummary(
block_summary_root: hash_tree_root(state.block_roots),
state_summary_root: hash_tree_root(state.state_roots),
)
if not state.historical_summaries.add(historical_summary):
return err("process_historical_summaries_update: state.historical_summaries full")
ok()
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#epoch-processing
proc process_epoch*(
cfg: RuntimeConfig, state: var phase0.BeaconState, flags: UpdateFlags,
cache: var StateCache, info: var phase0.EpochInfo): Result[void, cstring] =
let epoch = get_current_epoch(state)
trace "process_epoch", epoch
info.init(state)
info.process_attestations(state, cache)
process_justification_and_finalization(state, info.balances, flags)
# state.slot hasn't been incremented yet.
if strictVerification in flags and epoch >= 2:
doAssert state.current_justified_checkpoint.epoch + 2 >= epoch
if strictVerification in flags and epoch >= 3:
# Rule 2/3/4 finalization results in the most pessimal case. The other
# three finalization rules finalize more quickly as long as the any of
# the finalization rules triggered.
doAssert state.finalized_checkpoint.epoch + 3 >= epoch
process_rewards_and_penalties(state, info)
? process_registry_updates(cfg, state, cache)
process_slashings(state, info.balances.current_epoch)
process_eth1_data_reset(state)
process_effective_balance_updates(state)
process_slashings_reset(state)
process_randao_mixes_reset(state)
process_historical_roots_update(state)
process_participation_record_updates(state)
ok()
func init*(
info: var altair.EpochInfo,
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState) =
# init participation, overwriting the full structure
info.balances = get_unslashed_participating_balances(state)
info.validators.setLen(state.validators.len())
let previous_epoch = get_previous_epoch(state)
for index in 0..<state.validators.len():
var flags: set[ParticipationFlag]
if is_eligible_validator(state.validators[index], previous_epoch):
flags.incl ParticipationFlag.eligible
info.validators[index] = ParticipationInfo(
flags: flags
)
func init*(
T: type altair.EpochInfo,
state: altair.BeaconState | bellatrix.BeaconState | capella.BeaconState |
deneb.BeaconState): T =
init(result, state)
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/altair/beacon-chain.md#epoch-processing
proc process_epoch*(
cfg: RuntimeConfig,
state: var (altair.BeaconState | bellatrix.BeaconState),
flags: UpdateFlags, cache: var StateCache, info: var altair.EpochInfo):
Result[void, cstring] =
let epoch = get_current_epoch(state)
trace "process_epoch", epoch
info.init(state)
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/altair/beacon-chain.md#justification-and-finalization
# [Modified in Altair]
process_justification_and_finalization(state, info.balances, flags)
# state.slot hasn't been incremented yet.
if strictVerification in flags and epoch >= 2:
doAssert state.current_justified_checkpoint.epoch + 2 >= epoch
if strictVerification in flags and epoch >= 3:
# Rule 2/3/4 finalization results in the most pessimal case. The other
# three finalization rules finalize more quickly as long as the any of
# the finalization rules triggered.
doAssert state.finalized_checkpoint.epoch + 3 >= epoch
process_inactivity_updates(cfg, state, info) # [New in Altair]
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/altair/beacon-chain.md#rewards-and-penalties
process_rewards_and_penalties(cfg, state, info) # [Modified in Altair]
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#registry-updates
? process_registry_updates(cfg, state, cache)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#slashings
process_slashings(state, info.balances.current_epoch) # [Modified in Altair]
process_eth1_data_reset(state)
process_effective_balance_updates(state)
process_slashings_reset(state)
process_randao_mixes_reset(state)
process_historical_roots_update(state)
process_participation_flag_updates(state) # [New in Altair]
process_sync_committee_updates(state) # [New in Altair]
ok()
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/capella/beacon-chain.md#epoch-processing
proc process_epoch*(
cfg: RuntimeConfig,
state: var (capella.BeaconState | deneb.BeaconState),
flags: UpdateFlags, cache: var StateCache, info: var altair.EpochInfo):
Result[void, cstring] =
let epoch = get_current_epoch(state)
trace "process_epoch", epoch
info.init(state)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#justification-and-finalization
process_justification_and_finalization(state, info.balances, flags)
# state.slot hasn't been incremented yet.
if strictVerification in flags:
# Rule 2/3/4 finalization results in the most pessimal case. The other
# three finalization rules finalize more quickly as long as the any of
# the finalization rules triggered.
if (epoch >= 2 and state.current_justified_checkpoint.epoch + 2 < epoch) or
(epoch >= 3 and state.finalized_checkpoint.epoch + 3 < epoch):
fatal "The network did not finalize",
epoch, finalizedEpoch = state.finalized_checkpoint.epoch
quit 1
process_inactivity_updates(cfg, state, info)
# https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/altair/beacon-chain.md#rewards-and-penalties
process_rewards_and_penalties(cfg, state, info)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/phase0/beacon-chain.md#registry-updates
? process_registry_updates(cfg, state, cache)
# https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.2/specs/altair/beacon-chain.md#slashings
process_slashings(state, info.balances.current_epoch)
process_eth1_data_reset(state)
process_effective_balance_updates(state)
process_slashings_reset(state)
process_randao_mixes_reset(state)
? process_historical_summaries_update(state) # [Modified in Capella]
process_participation_flag_updates(state)
process_sync_committee_updates(state)
ok()