Merge pull request #317 from ethereum/vbuterin-patch-15
Put validator balances into a separate list in the state
This commit is contained in:
commit
8548f03d5a
|
@ -454,6 +454,7 @@ Unless otherwise indicated, code appearing in `this style` is to be interpreted
|
|||
|
||||
# Validator registry
|
||||
'validator_registry': [ValidatorRecord],
|
||||
'validator_balances': ['uint64'],
|
||||
'validator_registry_latest_change_slot': 'uint64',
|
||||
'validator_registry_exit_count': 'uint64',
|
||||
'validator_registry_delta_chain_tip': 'hash32', # For light clients to track deltas
|
||||
|
@ -496,8 +497,6 @@ Unless otherwise indicated, code appearing in `this style` is to be interpreted
|
|||
'randao_commitment': 'hash32',
|
||||
# Slots the proposer has skipped (i.e. layers of RANDAO expected)
|
||||
'randao_layers': 'uint64',
|
||||
# Balance in Gwei
|
||||
'balance': 'uint64',
|
||||
# Status code
|
||||
'status': 'uint64',
|
||||
# Slot when validator last changed status (or 0)
|
||||
|
@ -963,11 +962,11 @@ def get_attestation_participants(state: BeaconState,
|
|||
#### `get_effective_balance`
|
||||
|
||||
```python
|
||||
def get_effective_balance(validator: ValidatorRecord) -> int:
|
||||
def get_effective_balance(state: State, index: int) -> int:
|
||||
"""
|
||||
Returns the effective balance (also known as "balance at stake") for the ``validator``.
|
||||
Returns the effective balance (also known as "balance at stake") for a ``validator`` with the given ``index``.
|
||||
"""
|
||||
return min(validator.balance, MAX_DEPOSIT * GWEI_PER_ETH)
|
||||
return min(state.validator_balances[index], MAX_DEPOSIT * GWEI_PER_ETH)
|
||||
```
|
||||
|
||||
#### `get_new_validator_registry_delta_chain_tip`
|
||||
|
@ -1140,6 +1139,7 @@ def get_initial_beacon_state(initial_validator_deposits: List[Deposit],
|
|||
|
||||
# Validator registry
|
||||
validator_registry=[],
|
||||
validator_balances=[],
|
||||
validator_registry_latest_change_slot=INITIAL_SLOT_NUMBER,
|
||||
validator_registry_exit_count=0,
|
||||
validator_registry_delta_chain_tip=ZERO_HASH,
|
||||
|
@ -1179,7 +1179,7 @@ def get_initial_beacon_state(initial_validator_deposits: List[Deposit],
|
|||
withdrawal_credentials=deposit.deposit_data.deposit_input.withdrawal_credentials,
|
||||
randao_commitment=deposit.deposit_data.deposit_input.randao_commitment
|
||||
)
|
||||
if state.validator_registry[index].balance >= MAX_DEPOSIT * GWEI_PER_ETH:
|
||||
if get_effective_balance(state, validator_index) == MAX_DEPOSIT * GWEI_PER_ETH:
|
||||
update_validator_status(state, index, ACTIVE)
|
||||
|
||||
# set initial committee shuffling
|
||||
|
@ -1198,9 +1198,11 @@ def get_initial_beacon_state(initial_validator_deposits: List[Deposit],
|
|||
First, a helper function:
|
||||
|
||||
```python
|
||||
def min_empty_validator_index(validators: List[ValidatorRecord], current_slot: int) -> int:
|
||||
for i, v in enumerate(validators):
|
||||
if v.balance == 0 and v.latest_status_change_slot + ZERO_BALANCE_VALIDATOR_TTL <= current_slot:
|
||||
def min_empty_validator_index(validators: List[ValidatorRecord],
|
||||
validator_balances: List[int],
|
||||
current_slot: int) -> int:
|
||||
for i, (v, vbal) in enumerate(zip(validators, validator_balances)):
|
||||
if vbal == 0 and v.latest_status_change_slot + ZERO_BALANCE_VALIDATOR_TTL <= current_slot:
|
||||
return i
|
||||
return None
|
||||
```
|
||||
|
@ -1244,25 +1246,25 @@ def process_deposit(state: BeaconState,
|
|||
withdrawal_credentials=withdrawal_credentials,
|
||||
randao_commitment=randao_commitment,
|
||||
randao_layers=0,
|
||||
balance=deposit,
|
||||
status=PENDING_ACTIVATION,
|
||||
latest_status_change_slot=state.slot,
|
||||
exit_count=0
|
||||
)
|
||||
|
||||
index = min_empty_validator_index(validators_copy)
|
||||
index = min_empty_validator_index(state.validator_registry, state.validator_balances, state.slot)
|
||||
if index is None:
|
||||
state.validator_registry.append(validator)
|
||||
index = len(validators_copy) - 1
|
||||
state.validator_balances.append(deposit)
|
||||
index = len(state.validator_registry) - 1
|
||||
else:
|
||||
state.validator_registry[index] = validator
|
||||
state.validator_balances[index] = deposit
|
||||
else:
|
||||
# Increase balance by deposit
|
||||
index = validator_pubkeys.index(pubkey)
|
||||
validator = state.validator_registry[index]
|
||||
assert validator.withdrawal_credentials == withdrawal_credentials
|
||||
assert state.validator_registry[index].withdrawal_credentials == withdrawal_credentials
|
||||
|
||||
validator.balance += deposit
|
||||
state.validator_balances[index] += deposit
|
||||
|
||||
return index
|
||||
```
|
||||
|
@ -1344,12 +1346,12 @@ def exit_validator(state: BeaconState,
|
|||
validator.latest_status_change_slot = state.slot
|
||||
|
||||
if new_status == EXITED_WITH_PENALTY:
|
||||
state.latest_penalized_exit_balances[state.slot // COLLECTIVE_PENALTY_CALCULATION_PERIOD] += get_effective_balance(validator)
|
||||
state.latest_penalized_exit_balances[state.slot // COLLECTIVE_PENALTY_CALCULATION_PERIOD] += get_effective_balance(state, index)
|
||||
|
||||
whistleblower = state.validator_registry[get_beacon_proposer_index(state, state.slot)]
|
||||
whistleblower_reward = validator.balance // WHISTLEBLOWER_REWARD_QUOTIENT
|
||||
whistleblower.balance += whistleblower_reward
|
||||
validator.balance -= whistleblower_reward
|
||||
whistleblower_index = get_beacon_proposer_index(state, state.slot)
|
||||
whistleblower_reward = get_effective_balance(state, index) // WHISTLEBLOWER_REWARD_QUOTIENT
|
||||
state.validator_balances[whistleblower_index] += whistleblower_reward
|
||||
state.validator_balances[index] -= whistleblower_reward
|
||||
|
||||
if prev_status == EXITED_WITHOUT_PENALTY
|
||||
return
|
||||
|
@ -1383,8 +1385,8 @@ Below are the processing steps that happen at every slot.
|
|||
|
||||
### Block roots
|
||||
|
||||
* Let `previous_block_root` be the `hash_tree_root` of the previous beacon block processed in the chain.
|
||||
* Set `state.latest_block_roots = state.latest_block_roots[1:] + [previous_block_root]`.
|
||||
* Let `previous_block_root` be the `tree_hash_root` of the previous beacon block processed in the chain.
|
||||
* Set `state.latest_block_roots[(state.slot - 1) % LATEST_BLOCK_ROOTS_LENGTH] = previous_block_root`.
|
||||
* If `state.slot % LATEST_BLOCK_ROOTS_LENGTH == 0` append `merkle_root(state.latest_block_roots)` to `state.batched_block_roots`.
|
||||
|
||||
## Per-block processing
|
||||
|
@ -1524,8 +1526,8 @@ The steps below happen when `state.slot % EPOCH_LENGTH == 0`.
|
|||
|
||||
All [validators](#dfn-validator):
|
||||
|
||||
* Let `active_validators = [state.validator_registry[i] for i in get_active_validator_indices(state.validator_registry)]`.
|
||||
* Let `total_balance = sum([get_effective_balance(v) for v in active_validators])`.
|
||||
* Let `active_validator_indices = get_active_validator_indices(state.validator_registry)`.
|
||||
* Let `total_balance = sum([get_effective_balance(state, i) for i in active_validator_indices])`.
|
||||
|
||||
[Validators](#dfn-Validator) attesting during the current epoch:
|
||||
|
||||
|
@ -1533,42 +1535,37 @@ All [validators](#dfn-validator):
|
|||
* Validators justifying the epoch boundary block at the start of the current epoch:
|
||||
* Let `this_epoch_boundary_attestations = [a for a in this_epoch_attestations if a.data.epoch_boundary_root == get_block_root(state, state.slot-EPOCH_LENGTH) and a.justified_slot == state.justified_slot]`.
|
||||
* Let `this_epoch_boundary_attester_indices` be the union of the [validator](#dfn-validator) index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in this_epoch_boundary_attestations]`.
|
||||
* Let `this_epoch_boundary_attesters = [state.validator_registry[i] for indices in this_epoch_boundary_attester_indices for i in indices]`.
|
||||
* Let `this_epoch_boundary_attesting_balance = sum([get_effective_balance(v) for v in this_epoch_boundary_attesters])`.
|
||||
* Let `this_epoch_boundary_attesting_balance = sum([get_effective_balance(state, i) for i in this_epoch_boundary_attester_indices])`.
|
||||
|
||||
[Validators](#dfn-Validator) attesting during the previous epoch:
|
||||
|
||||
* Validators that made an attestation during the previous epoch:
|
||||
* Let `previous_epoch_attestations = [a for a in state.latest_attestations if state.slot - 2 * EPOCH_LENGTH <= a.slot < state.slot - EPOCH_LENGTH]`.
|
||||
* Let `previous_epoch_attester_indices` be the union of the validator index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in previous_epoch_attestations]`.
|
||||
* Let `previous_epoch_attesters = [state.validator_registry[i] for indices in previous_epoch_attester_indices for i in indices]`.
|
||||
* Validators targeting the previous justified hash:
|
||||
* Let `previous_epoch_justified_attestations = [a for a in this_epoch_attestations + previous_epoch_attestations if a.justified_slot == state.previous_justified_slot]`.
|
||||
* Let `previous_epoch_justified_attester_indices` be the union of the validator index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in previous_epoch_justified_attestations]`.
|
||||
* Let `previous_epoch_justified_attesters = [state.validator_registry[i] for indices in previous_epoch_justified_attester_indices for i in indices]`.
|
||||
* Let `previous_epoch_justified_attesting_balance = sum([get_effective_balance(v) for v in previous_epoch_justified_attesters])`.
|
||||
* Let `previous_epoch_justified_attesting_balance = sum([get_effective_balance(state, i) for i in previous_epoch_justified_attester_indices])`.
|
||||
* Validators justifying the epoch boundary block at the start of the previous epoch:
|
||||
* Let `previous_epoch_boundary_attestations = [a for a in previous_epoch_justified_attestations if a.epoch_boundary_root == get_block_root(state, state.slot - 2 * EPOCH_LENGTH)]`.
|
||||
* Let `previous_epoch_boundary_attester_indices` be the union of the validator index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in previous_epoch_boundary_attestations]`.
|
||||
* Let `previous_epoch_boundary_attesters = [state.validator_registry[i] for indices in previous_epoch_boundary_attester_indices for i in indices]`.
|
||||
* Let `previous_epoch_boundary_attesting_balance = sum([get_effective_balance(v) for v in previous_epoch_boundary_attesters])`.
|
||||
* Let `previous_epoch_boundary_attesting_balance = sum([get_effective_balance(state, i) for i in previous_epoch_boundary_attester_indices])`.
|
||||
* Validators attesting to the expected beacon chain head during the previous epoch:
|
||||
* Let `previous_epoch_head_attestations = [a for a in previous_epoch_attestations if a.beacon_block_root == get_block_root(state, a.slot)]`.
|
||||
* Let `previous_epoch_head_attester_indices` be the union of the validator index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in previous_epoch_head_attestations]`.
|
||||
* Let `previous_epoch_head_attesters = [state.validator_registry[i] for indices in previous_epoch_head_attester_indices for i in indices]`.
|
||||
* Let `previous_epoch_head_attesting_balance = sum([get_effective_balance(v) for v in previous_epoch_head_attesters])`.
|
||||
* Let `previous_epoch_head_attesting_balance = sum([get_effective_balance(state, i) for i in previous_epoch_head_attester_indices])`.
|
||||
|
||||
**Note**: `previous_epoch_boundary_attesting_balance` balance might be marginally different than `this_epoch_boundary_attesting_balance` during the previous epoch transition. Due to the tight bound on validator churn each epoch and small per-epoch rewards/penalties, the potential balance difference is very low and only marginally affects consensus safety.
|
||||
|
||||
For every `shard_committee` in `state.shard_committees_at_slots`:
|
||||
|
||||
* Let `attesting_validators(shard_committee, shard_block_root)` be the union of the [validator](#dfn-validator) index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in this_epoch_attestations + previous_epoch_attestations if a.shard == shard_committee.shard and a.shard_block_root == shard_block_root]`.
|
||||
* Let `winning_root(shard_committee)` be equal to the value of `shard_block_root` such that `sum([get_effective_balance(v) for v in attesting_validators(shard_committee, shard_block_root)])` is maximized (ties broken by favoring lower `shard_block_root` values).
|
||||
* Let `attesting_validator_indices(shard_committee, shard_block_root)` be the union of the [validator](#dfn-validator) index sets given by `[get_attestation_participants(state, a.data, a.participation_bitfield) for a in this_epoch_attestations + previous_epoch_attestations if a.shard == shard_committee.shard and a.shard_block_root == shard_block_root]`.
|
||||
* Let `winning_root(shard_committee)` be equal to the value of `shard_block_root` such that `sum([get_effective_balance(state, i) for i in attesting_validator_indices(shard_committee, shard_block_root)])` is maximized (ties broken by favoring lower `shard_block_root` values).
|
||||
* Let `attesting_validators(shard_committee)` be equal to `attesting_validators(shard_committee, winning_root(shard_committee))` for convenience.
|
||||
* Let `total_attesting_balance(shard_committee)` be the sum of the balances-at-stake of `attesting_validators(shard_committee)`.
|
||||
* Let `total_balance(shard_committee) = sum([get_effective_balance(v) for v in shard_committee.committee])`.
|
||||
* Let `inclusion_slot(v) = a.slot_included` for the attestation `a` where `v` is in `get_attestation_participants(state, a.data, a.participation_bitfield)`.
|
||||
* Let `inclusion_distance(v) = a.slot_included - a.data.slot` where `a` is the above attestation.
|
||||
* Let `total_balance(shard_committee) = sum([get_effective_balance(state, i) for i in shard_committee.committee])`.
|
||||
* Let `inclusion_slot(state, index) = a.slot_included` for the attestation `a` where `index` is in `get_attestation_participants(state, a.data, a.participation_bitfield)`.
|
||||
* Let `inclusion_distance(state, index) = a.slot_included - a.data.slot` where `a` is the above attestation.
|
||||
* Let `adjust_for_inclusion_distance(magnitude, distance)` be the function below.
|
||||
|
||||
```python
|
||||
|
@ -1613,8 +1610,8 @@ For every `shard_committee` in `state.shard_committees_at_slots`:
|
|||
First, we define some additional helpers:
|
||||
|
||||
* Let `base_reward_quotient = BASE_REWARD_QUOTIENT * integer_squareroot(total_balance // GWEI_PER_ETH)`.
|
||||
* Let `base_reward(v) = get_effective_balance(v) // base_reward_quotient // 4` for any validator `v`.
|
||||
* Let `inactivity_penalty(v, slots_since_finality) = base_reward(v) + get_effective_balance(v) * slots_since_finality // INACTIVITY_PENALTY_QUOTIENT` for any validator `v`.
|
||||
* Let `base_reward(state, index) = get_effective_balance(state, index) // base_reward_quotient // 4` for any validator with the given `index`.
|
||||
* Let `inactivity_penalty(state, index, slots_since_finality) = base_reward(state, index) + get_effective_balance(state, index) * slots_since_finality // INACTIVITY_PENALTY_QUOTIENT` for any validator with the given `index`.
|
||||
|
||||
#### Justification and finalization
|
||||
|
||||
|
@ -1625,32 +1622,32 @@ Note: When applying penalties in the following balance recalculations implemente
|
|||
Case 1: `slots_since_finality <= 4 * EPOCH_LENGTH`:
|
||||
|
||||
* Expected FFG source:
|
||||
* Any [validator](#dfn-validator) `v` in `previous_epoch_justified_attesters` gains `adjust_for_inclusion_distance(base_reward(v) * previous_epoch_justified_attesting_balance // total_balance, inclusion_distance(v))`.
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_justified_attesters` loses `base_reward(v)`.
|
||||
* Any [validator](#dfn-validator) `index` in `previous_epoch_justified_attester_indices` gains `adjust_for_inclusion_distance(base_reward(state, index) * previous_epoch_justified_attesting_balance // total_balance, inclusion_distance(state, index))`.
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_justified_attester_indices` loses `base_reward(state, index)`.
|
||||
* Expected FFG target:
|
||||
* Any [validator](#dfn-validator) `v` in `previous_epoch_boundary_attesters` gains `adjust_for_inclusion_distance(base_reward(v) * previous_epoch_boundary_attesting_balance // total_balance, inclusion_distance(v))`.
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_boundary_attesters` loses `base_reward(v)`.
|
||||
* Any [validator](#dfn-validator) `index` in `previous_epoch_boundary_attester_indices` gains `adjust_for_inclusion_distance(base_reward(state, index) * previous_epoch_boundary_attesting_balance // total_balance, inclusion_distance(state, index))`.
|
||||
* Any [active validator](#dfn-active-validator) `index` not in `previous_epoch_boundary_attester_indices` loses `base_reward(state, index)`.
|
||||
* Expected beacon chain head:
|
||||
* Any [validator](#dfn-validator) `v` in `previous_epoch_head_attesters` gains `adjust_for_inclusion_distance(base_reward(v) * previous_epoch_head_attesting_balance // total_balance, inclusion_distance(v))`.
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_head_attesters` loses `base_reward(v)`.
|
||||
* Any [validator](#dfn-validator) `index` in `previous_epoch_head_attester_indices` gains `adjust_for_inclusion_distance(base_reward(state, index) * previous_epoch_head_attesting_balance // total_balance, inclusion_distance(state, index))`.
|
||||
* Any [active validator](#dfn-active-validator) `index` not in `previous_epoch_head_attester_indices` loses `base_reward(state, index)`.
|
||||
|
||||
Case 2: `slots_since_finality > 4 * EPOCH_LENGTH`:
|
||||
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_justified_attesters`, loses `inactivity_penalty(v, slots_since_finality)`.
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_boundary_attesters`, loses `inactivity_penalty(v, slots_since_finality)`.
|
||||
* Any [active validator](#dfn-active-validator) `v` not in `previous_epoch_head_attesters`, loses `inactivity_penalty(v, slots_since_finality)`.
|
||||
* Any [validator](#dfn-validator) with `status == EXITED_WITH_PENALTY`, loses `3 * inactivity_penalty(v, slots_since_finality)`.
|
||||
* Any [active validator](#dfn-active-validator) `index` not in `previous_epoch_justified_attester_indices`, loses `inactivity_penalty(state, index, slots_since_finality)`.
|
||||
* Any [active validator](#dfn-active-validator) `index` not in `previous_epoch_boundary_attester_indices`, loses `inactivity_penalty(state, index, slots_since_finality)`.
|
||||
* Any [active validator](#dfn-active-validator) `index` not in `previous_epoch_head_attester_indices`, loses `inactivity_penalty(state, index, slots_since_finality)`.
|
||||
* Any [validator](#dfn-validator) `index` with `status == EXITED_WITH_PENALTY`, loses `3 * inactivity_penalty(state, index, slots_since_finality)`.
|
||||
|
||||
#### Attestation inclusion
|
||||
|
||||
For each `v` in `previous_epoch_attesters`, we determine the proposer `proposer_index = get_beacon_proposer_index(state, inclusion_slot(v))` and set `state.validator_registry[proposer_index].balance += base_reward(v) // INCLUDER_REWARD_QUOTIENT`.
|
||||
For each `index` in `previous_epoch_attester_indices`, we determine the proposer `proposer_index = get_beacon_proposer_index(state, inclusion_slot(state, index))` and set `state.validator_balances[proposer_index] += base_reward(state, index) // INCLUDER_REWARD_QUOTIENT`.
|
||||
|
||||
#### Crosslinks
|
||||
|
||||
For every `shard_committee` in `state.shard_committees_at_slots[:EPOCH_LENGTH]` (i.e. the objects corresponding to the epoch before the current one), for each `v` in `[state.validator_registry[index] for index in shard_committee.committee]`, adjust balances as follows:
|
||||
For every `shard_committee` in `state.shard_committees_at_slots[:EPOCH_LENGTH]` (i.e. the objects corresponding to the epoch before the current one), for each `index` in `shard_committee.committee`, adjust balances as follows:
|
||||
|
||||
* If `v in attesting_validators(shard_committee)`, `v.balance += adjust_for_inclusion_distance(base_reward(v) * total_attesting_balance(shard_committee) // total_balance(shard_committee)), inclusion_distance(v))`.
|
||||
* If `v not in attesting_validators(shard_committee)`, `v.balance -= base_reward(v)`.
|
||||
* If `index in attesting_validators(shard_committee)`, `state.validator_balances[index] += adjust_for_inclusion_distance(base_reward(state, index) * total_attesting_balance(shard_committee) // total_balance(shard_committee)), inclusion_distance(state, index))`.
|
||||
* If `index not in attesting_validators(shard_committee)`, `state.validator_balances[index] -= base_reward(state, index)`.
|
||||
|
||||
### Ejections
|
||||
|
||||
|
@ -1663,7 +1660,7 @@ def process_ejections(state: BeaconState) -> None:
|
|||
and eject active validators with balance below ``EJECTION_BALANCE``.
|
||||
"""
|
||||
for index in active_validator_indices(state.validator_registry):
|
||||
if state.validator_registry[index].balance < EJECTION_BALANCE:
|
||||
if state.validator_balances[index] < EJECTION_BALANCE:
|
||||
update_validator_status(state, index, new_status=EXITED_WITHOUT_PENALTY)
|
||||
```
|
||||
|
||||
|
@ -1685,7 +1682,7 @@ def update_validator_registry(state: BeaconState) -> None:
|
|||
# The active validators
|
||||
active_validator_indices = get_active_validator_indices(state.validator_registry)
|
||||
# The total effective balance of active validators
|
||||
total_balance = sum([get_effective_balance(v) for i, v in enumerate(validator_registry) if i in active_validator_indices])
|
||||
total_balance = sum([get_effective_balance(state, i) for i in active_validator_indices])
|
||||
|
||||
# The maximum balance churn in Gwei (for deposits and exits separately)
|
||||
max_balance_churn = max(
|
||||
|
@ -1696,9 +1693,9 @@ def update_validator_registry(state: BeaconState) -> None:
|
|||
# Activate validators within the allowable balance churn
|
||||
balance_churn = 0
|
||||
for index, validator in enumerate(state.validator_registry):
|
||||
if validator.status == PENDING_ACTIVATION and validator.balance >= MAX_DEPOSIT * GWEI_PER_ETH:
|
||||
if validator.status == PENDING_ACTIVATION and state.validator_balances[index] >= MAX_DEPOSIT * GWEI_PER_ETH:
|
||||
# Check the balance churn would be within the allowance
|
||||
balance_churn += get_effective_balance(validator)
|
||||
balance_churn += get_effective_balance(state, index)
|
||||
if balance_churn > max_balance_churn:
|
||||
break
|
||||
|
||||
|
@ -1710,7 +1707,7 @@ def update_validator_registry(state: BeaconState) -> None:
|
|||
for index, validator in enumerate(state.validator_registry):
|
||||
if validator.status == ACTIVE_PENDING_EXIT:
|
||||
# Check the balance churn would be within the allowance
|
||||
balance_churn += get_effective_balance(validator)
|
||||
balance_churn += get_effective_balance(state, index)
|
||||
if balance_churn > max_balance_churn:
|
||||
break
|
||||
|
||||
|
@ -1727,11 +1724,11 @@ def update_validator_registry(state: BeaconState) -> None:
|
|||
)
|
||||
|
||||
# Calculate penalties for slashed validators
|
||||
def to_penalize(v):
|
||||
return v.status == EXITED_WITH_PENALTY
|
||||
validators_to_penalize = filter(to_penalize, validator_registry)
|
||||
for v in validators_to_penalize:
|
||||
v.balance -= get_effective_balance(v) * min(total_penalties * 3, total_balance) // total_balance
|
||||
def to_penalize(index):
|
||||
return state.validator_registry[index].status == EXITED_WITH_PENALTY
|
||||
validators_to_penalize = filter(to_penalize, range(len(validator_registry)))
|
||||
for index in validators_to_penalize:
|
||||
state.validator_balances[index] -= get_effective_balance(state, index) * min(total_penalties * 3, total_balance) // total_balance
|
||||
|
||||
return validator_registry, latest_penalized_exit_balances, validator_registry_delta_chain_tip
|
||||
```
|
||||
|
|
Loading…
Reference in New Issue