943 lines
31 KiB
Nim
Raw Normal View History

# Nimbus
# Copyright (c) 2023-2025 Status Research & Development GmbH
# Licensed under either of
# * Apache License, version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or
# http://www.apache.org/licenses/LICENSE-2.0)
# * MIT license ([LICENSE-MIT](LICENSE-MIT) or
# http://opensource.org/licenses/MIT)
# at your option. This file may not be copied, modified, or distributed except
# according to those terms.
{.push raises: [].}
import
std/[tables, hashes, sets, typetraits],
chronicles,
eth/common/eth_types,
results,
minilru,
../utils/mergeutils,
../evm/code_bytes,
../stateless/multi_keys,
../core/eip7702,
"/.."/[constants, utils/utils],
./access_list as ac_access_list,
"."/[core_db, storage_types, transient_storage],
./aristo/aristo_blobify
export
code_bytes
const
debugLedgerRef = false
codeLruSize = 16*1024
# An LRU cache of 16K items gives roughly 90% hit rate anecdotally on a
# small range of test blocks - this number could be studied in more detail
# Per EIP-170, a the code of a contract can be up to `MAX_CODE_SIZE` = 24kb,
# which would cause a worst case of 386MB memory usage though in reality
# code sizes are much smaller - it would make sense to study these numbers
# in greater detail.
slotsLruSize = 16 * 1024
type
AccountFlag = enum
Alive
IsNew
Dirty
Touched
CodeChanged
StorageChanged
NewlyCreated # EIP-6780: self destruct only in same transaction
AccountFlags = set[AccountFlag]
AccountRef = ref object
statement: CoreDbAccount
accPath: Hash32
flags: AccountFlags
code: CodeBytesRef
originalStorage: TableRef[UInt256, UInt256]
overlayStorage: Table[UInt256, UInt256]
WitnessData* = object
storageKeys*: HashSet[UInt256]
codeTouched*: bool
LedgerRef* = ref object
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
txFrame*: CoreDbTxRef
savePoint: LedgerSpRef
witnessCache: Table[Address, WitnessData]
isDirty: bool
ripemdSpecial: bool
storeSlotHash*: bool
cache: Table[Address, AccountRef]
# Second-level cache for the ledger save point, which is cleared on every
# persist
code: LruCache[Hash32, CodeBytesRef]
## The code cache provides two main benefits:
##
## * duplicate code is shared in memory beween accounts
## * the jump destination table does not have to be recomputed for every
## execution, for commonly called called contracts
##
## The former feature is specially important in the 2.3-2.7M block range
## when underpriced code opcodes are being run en masse - both advantages
## help performance broadly as well.
slots: LruCache[UInt256, Hash32]
## Because the same slots often reappear, we want to avoid writing them
## over and over again to the database to avoid the WAL and compation
## write amplification that ensues
ReadOnlyLedger* = distinct LedgerRef
TransactionState = enum
Pending
Committed
RolledBack
LedgerSpRef* = ref object
parentSavepoint: LedgerSpRef
cache: Table[Address, AccountRef]
dirty: Table[Address, AccountRef]
selfDestruct: HashSet[Address]
logEntries: seq[Log]
accessList: ac_access_list.AccessList
transientStorage: TransientStorage
state: TransactionState
when debugLedgerRef:
depth: int
const
emptyEthAccount = Account.init()
resetFlags = {
Dirty,
IsNew,
Touched,
CodeChanged,
StorageChanged,
NewlyCreated
}
when debugLedgerRef:
import
stew/byteutils
proc inspectSavePoint(name: string, x: LedgerSpRef) =
debugEcho "*** ", name, ": ", x.depth, " ***"
var sp = x
while sp != nil:
for address, acc in sp.cache:
debugEcho address.toHex, " ", acc.flags
sp = sp.parentSavepoint
template logTxt(info: static[string]): static[string] =
"LedgerRef " & info
template toAccountKey(acc: AccountRef): Hash32 =
acc.accPath
template toAccountKey(eAddr: Address): Hash32 =
eAddr.data.keccak256
proc beginSavepoint*(ac: LedgerRef): LedgerSpRef {.gcsafe.}
proc resetCoreDbAccount(ac: LedgerRef, acc: AccountRef) =
const info = "resetCoreDbAccount(): "
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.clearStorage(acc.toAccountKey).isOkOr:
raiseAssert info & $$error
acc.statement.nonce = emptyEthAccount.nonce
acc.statement.balance = emptyEthAccount.balance
acc.statement.codeHash = emptyEthAccount.codeHash
proc getAccount(
ac: LedgerRef;
address: Address;
shouldCreate = true;
): AccountRef =
# search account from layers of cache
var sp = ac.savePoint
while sp != nil:
result = sp.cache.getOrDefault(address)
if not result.isNil:
return
sp = sp.parentSavepoint
if ac.cache.pop(address, result):
# Check second-level cache
ac.savePoint.cache[address] = result
return
# not found in cache, look into state trie
let
accPath = address.toAccountKey
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
rc = ac.txFrame.fetch accPath
if rc.isOk:
result = AccountRef(
statement: rc.value,
accPath: accPath,
flags: {Alive})
elif shouldCreate:
result = AccountRef(
statement: CoreDbAccount(
nonce: emptyEthAccount.nonce,
balance: emptyEthAccount.balance,
codeHash: emptyEthAccount.codeHash),
accPath: accPath,
flags: {Alive, IsNew})
else:
return # ignore, don't cache
# cache the account
ac.savePoint.cache[address] = result
ac.savePoint.dirty[address] = result
proc clone(acc: AccountRef, cloneStorage: bool): AccountRef =
result = AccountRef(
statement: acc.statement,
accPath: acc.accPath,
flags: acc.flags,
code: acc.code)
if cloneStorage:
result.originalStorage = acc.originalStorage
# it's ok to clone a table this way
result.overlayStorage = acc.overlayStorage
proc isEmpty(acc: AccountRef): bool =
acc.statement.nonce == 0 and
acc.statement.balance.isZero and
acc.statement.codeHash == EMPTY_CODE_HASH
template exists(acc: AccountRef): bool =
Alive in acc.flags
proc originalStorageValue(
acc: AccountRef;
slot: UInt256;
ac: LedgerRef;
): UInt256 =
# share the same original storage between multiple
# versions of account
if acc.originalStorage.isNil:
acc.originalStorage = newTable[UInt256, UInt256]()
else:
acc.originalStorage[].withValue(slot, val) do:
return val[]
# Not in the original values cache - go to the DB.
let
slotKey = ac.slots.get(slot).valueOr:
slot.toBytesBE.keccak256
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
rc = ac.txFrame.slotFetch(acc.toAccountKey, slotKey)
if rc.isOk:
result = rc.value
acc.originalStorage[slot] = result
proc storageValue(
acc: AccountRef;
slot: UInt256;
ac: LedgerRef;
): UInt256 =
acc.overlayStorage.withValue(slot, val) do:
return val[]
do:
result = acc.originalStorageValue(slot, ac)
proc kill(ac: LedgerRef, acc: AccountRef) =
acc.flags.excl Alive
acc.overlayStorage.clear()
acc.originalStorage = nil
ac.resetCoreDbAccount acc
acc.code.reset()
type
PersistMode = enum
DoNothing
Update
Remove
proc persistMode(acc: AccountRef): PersistMode =
result = DoNothing
if Alive in acc.flags:
if IsNew in acc.flags or Dirty in acc.flags:
result = Update
else:
if IsNew notin acc.flags:
result = Remove
proc persistCode(acc: AccountRef, ac: LedgerRef) =
if acc.code.len != 0 and not acc.code.persisted:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
let rc = ac.txFrame.put(
contractHashKey(acc.statement.codeHash).toOpenArray, acc.code.bytes())
if rc.isErr:
warn logTxt "persistCode()",
codeHash=acc.statement.codeHash, error=($$rc.error)
else:
# If the ledger changes rolled back entirely from the database, the ledger
# code cache must also be cleared!
acc.code.persisted = true
proc persistStorage(acc: AccountRef, ac: LedgerRef) =
const info = "persistStorage(): "
if acc.overlayStorage.len == 0:
# TODO: remove the storage too if we figure out
# how to create 'virtual' storage room for each account
return
if acc.originalStorage.isNil:
acc.originalStorage = newTable[UInt256, UInt256]()
# Make sure that there is an account entry on the database. This is needed by
# `Aristo` for updating the account's storage area reference. As a side effect,
# this action also updates the latest statement data.
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.merge(acc.toAccountKey, acc.statement).isOkOr:
raiseAssert info & $$error
# Save `overlayStorage[]` on database
for slot, value in acc.overlayStorage:
acc.originalStorage[].withValue(slot, v):
if v[] == value:
continue # Avoid writing A-B-A updates
var cached = true
let slotKey = ac.slots.get(slot).valueOr:
cached = false
let hash = slot.toBytesBE.keccak256
ac.slots.put(slot, hash)
hash
if value > 0:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.slotMerge(acc.toAccountKey, slotKey, value).isOkOr:
raiseAssert info & $$error
# move the overlayStorage to originalStorage, related to EIP2200, EIP1283
acc.originalStorage[slot] = value
else:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.slotDelete(acc.toAccountKey, slotKey).isOkOr:
if error.error != StoNotFound:
raiseAssert info & $$error
discard
acc.originalStorage.del(slot)
if ac.storeSlotHash and not cached:
# Write only if it was not cached to avoid writing the same data over and
# over..
let
key = slotKey.data.slotHashToSlotKey
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
rc = ac.txFrame.put(key.toOpenArray, blobify(slot).data)
if rc.isErr:
warn logTxt "persistStorage()", slot, error=($$rc.error)
acc.overlayStorage.clear()
proc makeDirty(ac: LedgerRef, address: Address, cloneStorage = true): AccountRef =
ac.isDirty = true
result = ac.getAccount(address)
if address in ac.savePoint.cache:
# it's already in latest savepoint
result.flags.incl Dirty
ac.savePoint.dirty[address] = result
return
# put a copy into latest savepoint
result = result.clone(cloneStorage)
result.flags.incl Dirty
ac.savePoint.cache[address] = result
ac.savePoint.dirty[address] = result
# ------------------------------------------------------------------------------
# Public methods
# ------------------------------------------------------------------------------
# The LedgerRef is modeled after TrieDatabase for it's transaction style
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
proc init*(x: typedesc[LedgerRef], db: CoreDbTxRef, storeSlotHash: bool): LedgerRef =
new result
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
result.txFrame = db
result.witnessCache = Table[Address, WitnessData]()
result.storeSlotHash = storeSlotHash
result.code = typeof(result.code).init(codeLruSize)
result.slots = typeof(result.slots).init(slotsLruSize)
discard result.beginSavepoint
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
proc init*(x: typedesc[LedgerRef], db: CoreDbTxRef): LedgerRef =
init(x, db, false)
proc getStateRoot*(ac: LedgerRef): Hash32 =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
# make sure all cache already committed
doAssert(ac.isDirty == false)
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.getStateRoot().expect("working database")
proc isTopLevelClean*(ac: LedgerRef): bool =
## Getter, returns `true` if all pending data have been commited.
not ac.isDirty and ac.savePoint.parentSavepoint.isNil
proc beginSavepoint*(ac: LedgerRef): LedgerSpRef =
new result
result.cache = Table[Address, AccountRef]()
result.accessList.init()
result.transientStorage.init()
result.state = Pending
result.parentSavepoint = ac.savePoint
ac.savePoint = result
when debugLedgerRef:
if not result.parentSavePoint.isNil:
result.depth = result.parentSavePoint.depth + 1
inspectSavePoint("snapshot", result)
proc rollback*(ac: LedgerRef, sp: LedgerSpRef) =
# Transactions should be handled in a strictly nested fashion.
# Any child transaction must be committed or rolled-back before
# its parent transactions:
doAssert ac.savePoint == sp and sp.state == Pending
ac.savePoint = sp.parentSavepoint
sp.state = RolledBack
when debugLedgerRef:
inspectSavePoint("rollback", ac.savePoint)
proc commit*(ac: LedgerRef, sp: LedgerSpRef) =
# Transactions should be handled in a strictly nested fashion.
# Any child transaction must be committed or rolled-back before
# its parent transactions:
doAssert ac.savePoint == sp and sp.state == Pending
# cannot commit most inner savepoint
doAssert not sp.parentSavepoint.isNil
ac.savePoint = sp.parentSavepoint
ac.savePoint.cache.mergeAndReset(sp.cache)
ac.savePoint.dirty.mergeAndReset(sp.dirty)
ac.savePoint.transientStorage.mergeAndReset(sp.transientStorage)
ac.savePoint.accessList.mergeAndReset(sp.accessList)
ac.savePoint.selfDestruct.mergeAndReset(sp.selfDestruct)
ac.savePoint.logEntries.mergeAndReset(sp.logEntries)
sp.state = Committed
when debugLedgerRef:
inspectSavePoint("commit", ac.savePoint)
proc dispose*(ac: LedgerRef, sp: LedgerSpRef) =
if sp.state == Pending:
ac.rollback(sp)
proc safeDispose*(ac: LedgerRef, sp: LedgerSpRef) =
if (not isNil(sp)) and (sp.state == Pending):
ac.rollback(sp)
proc getCodeHash*(ac: LedgerRef, address: Address): Hash32 =
let acc = ac.getAccount(address, false)
if acc.isNil: emptyEthAccount.codeHash
else: acc.statement.codeHash
proc getBalance*(ac: LedgerRef, address: Address): UInt256 =
let acc = ac.getAccount(address, false)
if acc.isNil: emptyEthAccount.balance
else: acc.statement.balance
proc getNonce*(ac: LedgerRef, address: Address): AccountNonce =
let acc = ac.getAccount(address, false)
if acc.isNil: emptyEthAccount.nonce
else: acc.statement.nonce
proc getCode*(ac: LedgerRef,
address: Address,
returnHash: static[bool] = false): auto =
let acc = ac.getAccount(address, false)
if acc.isNil:
when returnHash:
return (EMPTY_CODE_HASH, CodeBytesRef())
else:
return CodeBytesRef()
if acc.code == nil:
acc.code =
if acc.statement.codeHash != EMPTY_CODE_HASH:
ac.code.get(acc.statement.codeHash).valueOr:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
var rc = ac.txFrame.get(contractHashKey(acc.statement.codeHash).toOpenArray)
if rc.isErr:
warn logTxt "getCode()", codeHash=acc.statement.codeHash, error=($$rc.error)
CodeBytesRef()
else:
let newCode = CodeBytesRef.init(move(rc.value), persisted = true)
ac.code.put(acc.statement.codeHash, newCode)
newCode
else:
CodeBytesRef()
when returnHash:
(acc.statement.codeHash, acc.code)
else:
acc.code
proc getCodeSize*(ac: LedgerRef, address: Address): int =
let acc = ac.getAccount(address, false)
if acc.isNil:
return 0
if acc.code == nil:
if acc.statement.codeHash == EMPTY_CODE_HASH:
return 0
acc.code = ac.code.get(acc.statement.codeHash).valueOr:
# On a cache miss, we don't fetch the code - instead, we fetch just the
# length - should the code itself be needed, it will typically remain
# cached and easily accessible in the database layer - this is to prevent
# EXTCODESIZE calls from messing up the code cache and thus causing
# recomputation of the jump destination table
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
var rc = ac.txFrame.len(contractHashKey(acc.statement.codeHash).toOpenArray)
return rc.valueOr:
warn logTxt "getCodeSize()", codeHash=acc.statement.codeHash, error=($$rc.error)
0
acc.code.len()
proc resolveCode*(ac: LedgerRef, address: Address): CodeBytesRef =
let code = ac.getCode(address)
let delegateTo = parseDelegationAddress(code).valueOr:
return code
ac.getCode(delegateTo)
proc getCommittedStorage*(ac: LedgerRef, address: Address, slot: UInt256): UInt256 =
let acc = ac.getAccount(address, false)
if acc.isNil:
return
acc.originalStorageValue(slot, ac)
proc getStorage*(ac: LedgerRef, address: Address, slot: UInt256): UInt256 =
let acc = ac.getAccount(address, false)
if acc.isNil:
return
acc.storageValue(slot, ac)
proc contractCollision*(ac: LedgerRef, address: Address): bool =
let acc = ac.getAccount(address, false)
if acc.isNil:
return
acc.statement.nonce != 0 or
acc.statement.codeHash != EMPTY_CODE_HASH or
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
not ac.txFrame.slotStorageEmptyOrVoid(acc.toAccountKey)
proc accountExists*(ac: LedgerRef, address: Address): bool =
let acc = ac.getAccount(address, false)
if acc.isNil:
return
acc.exists()
proc isEmptyAccount*(ac: LedgerRef, address: Address): bool =
let acc = ac.getAccount(address, false)
doAssert not acc.isNil
doAssert acc.exists()
acc.isEmpty()
proc isDeadAccount*(ac: LedgerRef, address: Address): bool =
let acc = ac.getAccount(address, false)
if acc.isNil:
return true
if not acc.exists():
return true
acc.isEmpty()
proc setBalance*(ac: LedgerRef, address: Address, balance: UInt256) =
let acc = ac.getAccount(address)
acc.flags.incl {Alive}
if acc.statement.balance != balance:
ac.makeDirty(address).statement.balance = balance
proc addBalance*(ac: LedgerRef, address: Address, delta: UInt256) =
# EIP161: We must check emptiness for the objects such that the account
# clearing (0,0,0 objects) can take effect.
if delta.isZero:
let acc = ac.getAccount(address)
if acc.isEmpty:
ac.makeDirty(address).flags.incl Touched
return
ac.setBalance(address, ac.getBalance(address) + delta)
proc subBalance*(ac: LedgerRef, address: Address, delta: UInt256) =
if delta.isZero:
# This zero delta early exit is important as shown in EIP-4788.
# If the account is created, it will change the state.
# But early exit will prevent the account creation.
# In this case, the SYSTEM_ADDRESS
return
ac.setBalance(address, ac.getBalance(address) - delta)
proc setNonce*(ac: LedgerRef, address: Address, nonce: AccountNonce) =
let acc = ac.getAccount(address)
acc.flags.incl {Alive}
if acc.statement.nonce != nonce:
ac.makeDirty(address).statement.nonce = nonce
proc incNonce*(ac: LedgerRef, address: Address) =
ac.setNonce(address, ac.getNonce(address) + 1)
proc setCode*(ac: LedgerRef, address: Address, code: seq[byte]) =
let acc = ac.getAccount(address)
acc.flags.incl {Alive}
let codeHash = keccak256(code)
if acc.statement.codeHash != codeHash:
var acc = ac.makeDirty(address)
acc.statement.codeHash = codeHash
# Try to reuse cache entry if it exists, but don't save the code - it's not
# a given that it will be executed within LRU range
acc.code = ac.code.get(codeHash).valueOr(CodeBytesRef.init(code))
acc.flags.incl CodeChanged
proc setStorage*(ac: LedgerRef, address: Address, slot, value: UInt256) =
let acc = ac.getAccount(address)
acc.flags.incl {Alive}
let oldValue = acc.storageValue(slot, ac)
if oldValue != value:
var acc = ac.makeDirty(address)
acc.overlayStorage[slot] = value
acc.flags.incl StorageChanged
proc clearStorage*(ac: LedgerRef, address: Address) =
const info = "clearStorage(): "
# a.k.a createStateObject. If there is an existing account with
# the given address, it is overwritten.
let acc = ac.getAccount(address)
acc.flags.incl {Alive, NewlyCreated}
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
let empty = ac.txFrame.slotStorageEmpty(acc.toAccountKey).valueOr: return
if not empty:
# need to clear the storage from the database first
let acc = ac.makeDirty(address, cloneStorage = false)
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.clearStorage(acc.toAccountKey).isOkOr:
raiseAssert info & $$error
# update caches
if acc.originalStorage.isNil.not:
# also clear originalStorage cache, otherwise
# both getStorage and getCommittedStorage will
# return wrong value
acc.originalStorage.clear()
proc deleteAccount*(ac: LedgerRef, address: Address) =
# make sure all savepoints already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
let acc = ac.getAccount(address)
ac.savePoint.dirty[address] = acc
ac.kill acc
proc selfDestruct*(ac: LedgerRef, address: Address) =
ac.setBalance(address, 0.u256)
ac.savePoint.selfDestruct.incl address
proc selfDestruct6780*(ac: LedgerRef, address: Address) =
let acc = ac.getAccount(address, false)
if acc.isNil:
return
if NewlyCreated in acc.flags:
ac.selfDestruct(address)
proc selfDestructLen*(ac: LedgerRef): int =
ac.savePoint.selfDestruct.len
proc addLogEntry*(ac: LedgerRef, log: Log) =
ac.savePoint.logEntries.add log
proc getAndClearLogEntries*(ac: LedgerRef): seq[Log] =
swap(result, ac.savePoint.logEntries)
proc ripemdSpecial*(ac: LedgerRef) =
ac.ripemdSpecial = true
proc deleteEmptyAccount(ac: LedgerRef, address: Address) =
let acc = ac.getAccount(address, false)
if acc.isNil:
return
if not acc.isEmpty:
return
if not acc.exists:
return
ac.savePoint.dirty[address] = acc
ac.kill acc
proc clearEmptyAccounts(ac: LedgerRef) =
# https://github.com/ethereum/EIPs/blob/master/EIPS/eip-161.md
for acc in ac.savePoint.dirty.values():
if Touched in acc.flags and
acc.isEmpty and acc.exists:
ac.kill acc
# https://github.com/ethereum/EIPs/issues/716
if ac.ripemdSpecial:
ac.deleteEmptyAccount(RIPEMD_ADDR)
ac.ripemdSpecial = false
proc persist*(ac: LedgerRef,
clearEmptyAccount: bool = false,
clearCache = false) =
const info = "persist(): "
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
if clearEmptyAccount:
ac.clearEmptyAccounts()
for address in ac.savePoint.selfDestruct:
ac.deleteAccount(address)
for (eAddr,acc) in ac.savePoint.dirty.pairs(): # This is a hotspot in block processing
case acc.persistMode()
of Update:
if CodeChanged in acc.flags:
acc.persistCode(ac)
if StorageChanged in acc.flags:
acc.persistStorage(ac)
else:
# This one is only necessary unless `persistStorage()` is run which needs
# to `merge()` the latest statement as well.
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.merge(acc.toAccountKey, acc.statement).isOkOr:
raiseAssert info & $$error
of Remove:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
ac.txFrame.delete(acc.toAccountKey).isOkOr:
if error.error != AccNotFound:
raiseAssert info & $$error
ac.savePoint.cache.del eAddr
of DoNothing:
# dead man tell no tales
# remove touched dead account from cache
if Alive notin acc.flags:
ac.savePoint.cache.del eAddr
acc.flags = acc.flags - resetFlags
ac.savePoint.dirty.clear()
if clearCache:
# This overwrites the cache from the previous persist, providing a crude LRU
# scheme with little overhead
# TODO https://github.com/nim-lang/Nim/issues/23759
swap(ac.cache, ac.savePoint.cache)
ac.savePoint.cache.reset()
ac.savePoint.selfDestruct.clear()
# EIP2929
ac.savePoint.accessList.clear()
ac.isDirty = false
iterator addresses*(ac: LedgerRef): Address =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
for address, _ in ac.savePoint.cache:
yield address
iterator accounts*(ac: LedgerRef): Account =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
for _, acc in ac.savePoint.cache:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
yield ac.txFrame.recast(
acc.toAccountKey, acc.statement).value
iterator pairs*(ac: LedgerRef): (Address, Account) =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
for address, acc in ac.savePoint.cache:
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
yield (address, ac.txFrame.recast(
acc.toAccountKey, acc.statement).value)
iterator storage*(
ac: LedgerRef;
eAddr: Address;
): (UInt256, UInt256) =
# beware that if the account not persisted,
# the storage root will not be updated
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
for (slotHash, value) in ac.txFrame.slotPairs eAddr.toAccountKey:
let rc = ac.txFrame.get(slotHashToSlotKey(slotHash).toOpenArray)
if rc.isErr:
warn logTxt "storage()", slotHash, error=($$rc.error)
continue
let r = deblobify(rc.value, UInt256)
if r.isErr:
warn logTxt "storage.deblobify", slotHash, msg=r.error
continue
yield (r.value, value)
iterator cachedStorage*(ac: LedgerRef, address: Address): (UInt256, UInt256) =
let acc = ac.getAccount(address, false)
if not acc.isNil:
if not acc.originalStorage.isNil:
for k, v in acc.originalStorage:
yield (k, v)
proc getStorageRoot*(ac: LedgerRef, address: Address): Hash32 =
# beware that if the account not persisted,
# the storage root will not be updated
let acc = ac.getAccount(address, false)
if acc.isNil: EMPTY_ROOT_HASH
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
else: ac.txFrame.slotStorageRoot(acc.toAccountKey).valueOr: EMPTY_ROOT_HASH
proc update(wd: var WitnessData, acc: AccountRef) =
# once the code is touched make sure it doesn't get reset back to false in another update
if not wd.codeTouched:
wd.codeTouched = CodeChanged in acc.flags or acc.code != nil
if not acc.originalStorage.isNil:
for k, v in acc.originalStorage:
if v.isZero: continue
wd.storageKeys.incl k
for k, v in acc.overlayStorage:
wd.storageKeys.incl k
proc witnessData(acc: AccountRef): WitnessData =
result.storageKeys = HashSet[UInt256]()
update(result, acc)
proc collectWitnessData*(ac: LedgerRef) =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
# usually witness data is collected before we call persist()
for address, acc in ac.savePoint.cache:
ac.witnessCache.withValue(address, val) do:
update(val[], acc)
do:
ac.witnessCache[address] = witnessData(acc)
func multiKeys(slots: HashSet[UInt256]): MultiKeysRef =
if slots.len == 0: return
new result
for x in slots:
result.add x.toBytesBE
result.sort()
proc makeMultiKeys*(ac: LedgerRef): MultiKeysRef =
# this proc is called after we done executing a block
new result
for k, v in ac.witnessCache:
result.add(k, v.codeTouched, multiKeys(v.storageKeys))
result.sort()
proc accessList*(ac: LedgerRef, address: Address) =
ac.savePoint.accessList.add(address)
proc accessList*(ac: LedgerRef, address: Address, slot: UInt256) =
ac.savePoint.accessList.add(address, slot)
func inAccessList*(ac: LedgerRef, address: Address): bool =
var sp = ac.savePoint
while sp != nil:
result = sp.accessList.contains(address)
if result:
return
sp = sp.parentSavepoint
func inAccessList*(ac: LedgerRef, address: Address, slot: UInt256): bool =
var sp = ac.savePoint
while sp != nil:
result = sp.accessList.contains(address, slot)
if result:
return
sp = sp.parentSavepoint
func getTransientStorage*(ac: LedgerRef,
address: Address, slot: UInt256): UInt256 =
var sp = ac.savePoint
while sp != nil:
let (ok, res) = sp.transientStorage.getStorage(address, slot)
if ok:
return res
sp = sp.parentSavepoint
proc setTransientStorage*(ac: LedgerRef,
address: Address, slot, val: UInt256) =
ac.savePoint.transientStorage.setStorage(address, slot, val)
proc clearTransientStorage*(ac: LedgerRef) =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
ac.savePoint.transientStorage.clear()
func getAccessList*(ac: LedgerRef): transactions.AccessList =
# make sure all savepoint already committed
doAssert(ac.savePoint.parentSavepoint.isNil)
ac.savePoint.accessList.getAccessList()
proc getEthAccount*(ac: LedgerRef, address: Address): Account =
let acc = ac.getAccount(address, false)
if acc.isNil:
return emptyEthAccount
## Convert to legacy object, will throw an assert if that fails
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
let rc = ac.txFrame.recast(acc.toAccountKey, acc.statement)
if rc.isErr:
raiseAssert "getAccount(): cannot convert account: " & $$rc.error
rc.value
proc getAccountProof*(ac: LedgerRef, address: Address): seq[seq[byte]] =
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
let accProof = ac.txFrame.proof(address.toAccountKey).valueOr:
raiseAssert "Failed to get account proof: " & $$error
accProof[0]
proc getStorageProof*(ac: LedgerRef, address: Address, slots: openArray[UInt256]): seq[seq[seq[byte]]] =
var storageProof = newSeqOfCap[seq[seq[byte]]](slots.len)
let
addressHash = address.toAccountKey
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
accountExists = ac.txFrame.hasPath(addressHash).valueOr:
raiseAssert "Call to hasPath failed: " & $$error
for slot in slots:
if not accountExists:
storageProof.add(@[])
continue
let
slotKey = ac.slots.get(slot).valueOr:
slot.toBytesBE.keccak256
aristo: fork support via layers/txframes (#2960) * aristo: fork support via layers/txframes This change reorganises how the database is accessed: instead holding a "current frame" in the database object, a dag of frames is created based on the "base frame" held in `AristoDbRef` and all database access happens through this frame, which can be thought of as a consistent point-in-time snapshot of the database based on a particular fork of the chain. In the code, "frame", "transaction" and "layer" is used to denote more or less the same thing: a dag of stacked changes backed by the on-disk database. Although this is not a requirement, in practice each frame holds the change set of a single block - as such, the frame and its ancestors leading up to the on-disk state represents the state of the database after that block has been applied. "committing" means merging the changes to its parent frame so that the difference between them is lost and only the cumulative changes remain - this facility enables frames to be combined arbitrarily wherever they are in the dag. In particular, it becomes possible to consolidate a set of changes near the base of the dag and commit those to disk without having to re-do the in-memory frames built on top of them - this is useful for "flattening" a set of changes during a base update and sending those to storage without having to perform a block replay on top. Looking at abstractions, a side effect of this change is that the KVT and Aristo are brought closer together by considering them to be part of the "same" atomic transaction set - the way the code gets organised, applying a block and saving it to the kvt happens in the same "logical" frame - therefore, discarding the frame discards both the aristo and kvt changes at the same time - likewise, they are persisted to disk together - this makes reasoning about the database somewhat easier but has the downside of increased memory usage, something that perhaps will need addressing in the future. Because the code reasons more strictly about frames and the state of the persisted database, it also makes it more visible where ForkedChain should be used and where it is still missing - in particular, frames represent a single branch of history while forkedchain manages multiple parallel forks - user-facing services such as the RPC should use the latter, ie until it has been finalized, a getBlock request should consider all forks and not just the blocks in the canonical head branch. Another advantage of this approach is that `AristoDbRef` conceptually becomes more simple - removing its tracking of the "current" transaction stack simplifies reasoning about what can go wrong since this state now has to be passed around in the form of `AristoTxRef` - as such, many of the tests and facilities in the code that were dealing with "stack inconsistency" are now structurally prevented from happening. The test suite will need significant refactoring after this change. Once this change has been merged, there are several follow-ups to do: * there's no mechanism for keeping frames up to date as they get committed or rolled back - TODO * naming is confused - many names for the same thing for legacy reason * forkedchain support is still missing in lots of code * clean up redundant logic based on previous designs - in particular the debug and introspection code no longer makes sense * the way change sets are stored will probably need revisiting - because it's a stack of changes where each frame must be interrogated to find an on-disk value, with a base distance of 128 we'll at minimum have to perform 128 frame lookups for *every* database interaction - regardless, the "dag-like" nature will stay * dispose and commit are poorly defined and perhaps redundant - in theory, one could simply let the GC collect abandoned frames etc, though it's likely an explicit mechanism will remain useful, so they stay for now More about the changes: * `AristoDbRef` gains a `txRef` field (todo: rename) that "more or less" corresponds to the old `balancer` field * `AristoDbRef.stack` is gone - instead, there's a chain of `AristoTxRef` objects that hold their respective "layer" which has the actual changes * No more reasoning about "top" and "stack" - instead, each `AristoTxRef` can be a "head" that "more or less" corresponds to the old single-history `top` notion and its stack * `level` still represents "distance to base" - it's computed from the parent chain instead of being stored * one has to be careful not to use frames where forkedchain was intended - layers are only for a single branch of history! * fix layer vtop after rollback * engine fix * Fix test_txpool * Fix test_rpc * Fix copyright year * fix simulator * Fix copyright year * Fix copyright year * Fix tracer * Fix infinite recursion bug * Remove aristo and kvt empty files * Fic copyright year * Fix fc chain_kvt * ForkedChain refactoring * Fix merge master conflict * Fix copyright year * Reparent txFrame * Fix test * Fix txFrame reparent again * Cleanup and fix test * UpdateBase bugfix and fix test * Fixe newPayload bug discovered by hive * Fix engine api fcu * Clean up call template, chain_kvt, andn txguid * Fix copyright year * work around base block loading issue * Add test * Fix updateHead bug * Fix updateBase bug * Change func commitBase to proc commitBase * Touch up and fix debug mode crash --------- Co-authored-by: jangko <jangko128@gmail.com>
2025-02-06 08:04:50 +01:00
slotProof = ac.txFrame.slotProof(addressHash, slotKey).valueOr:
if error.aErr == FetchPathNotFound:
storageProof.add(@[])
continue
else:
raiseAssert "Failed to get slot proof: " & $$error
storageProof.add(slotProof[0])
Core db aristo hasher profiling and timing improvement (#1938) * Explicitly use shared `Kvt` table on `Ledger` and `Clique` lookup. why: Speeds up lookup time with `Aristo` backend. For writing `Clique` data, the `Companion` model allows to write `Clique` data past the database locked by evm transactions. * Implement `CoreDb` profiling with API tracking why: Chasing time spent per APT procs ... * Implement `Ledger` profiling with API tracking why: Chasing time spent per APT procs ... * Always hashify when commiting or storing why: A dirty cache makes no sense when committing * Make sure that a zero key is created when adding/updating vertices why: This is an error fix mainly for edge cases. A typical error was that the root key got deleted when there were only a few vertices left on the DB. * Need all created and changed vertices zero-keyed on the cache why: A zero key (i.e. empty Merkle hash) indicates that a vertex key needs to be updated. This would not be needed immediately after a merge as there is an actual leaf path on the cache layer. But after subsequent merge and delete operations this information might get blurred. * Re-org hashing algorithm why: Apart from errors, the previous implementation was too slow for two reasons: + some control hashes were calculated for debugging (now all verification is done in `aristo_check` module) + the leaf paths stored on the cache are used to build the labelling (aka hashing) schedule; there paths were accumulated over successive hash sessions although it is clear that all keys were generated, already
2023-12-12 17:47:41 +00:00
storageProof
# ------------------------------------------------------------------------------
# Public virtual read-only methods
# ------------------------------------------------------------------------------
proc getStateRoot*(db: ReadOnlyLedger): Hash32 {.borrow.}
proc getCodeHash*(db: ReadOnlyLedger, address: Address): Hash32 = getCodeHash(distinctBase db, address)
proc getStorageRoot*(db: ReadOnlyLedger, address: Address): Hash32 = getStorageRoot(distinctBase db, address)
proc getBalance*(db: ReadOnlyLedger, address: Address): UInt256 = getBalance(distinctBase db, address)
proc getStorage*(db: ReadOnlyLedger, address: Address, slot: UInt256): UInt256 = getStorage(distinctBase db, address, slot)
proc getNonce*(db: ReadOnlyLedger, address: Address): AccountNonce = getNonce(distinctBase db, address)
proc getCode*(db: ReadOnlyLedger, address: Address): CodeBytesRef = getCode(distinctBase db, address)
proc getCodeSize*(db: ReadOnlyLedger, address: Address): int = getCodeSize(distinctBase db, address)
proc contractCollision*(db: ReadOnlyLedger, address: Address): bool = contractCollision(distinctBase db, address)
proc accountExists*(db: ReadOnlyLedger, address: Address): bool = accountExists(distinctBase db, address)
proc isDeadAccount*(db: ReadOnlyLedger, address: Address): bool = isDeadAccount(distinctBase db, address)
proc isEmptyAccount*(db: ReadOnlyLedger, address: Address): bool = isEmptyAccount(distinctBase db, address)
proc getCommittedStorage*(db: ReadOnlyLedger, address: Address, slot: UInt256): UInt256 = getCommittedStorage(distinctBase db, address, slot)
proc inAccessList*(db: ReadOnlyLedger, address: Address): bool = inAccessList(distinctBase db, address)
proc inAccessList*(db: ReadOnlyLedger, address: Address, slot: UInt256): bool = inAccessList(distinctBase db, address)
proc getTransientStorage*(db: ReadOnlyLedger,
address: Address, slot: UInt256): UInt256 = getTransientStorage(distinctBase db, address, slot)
proc getAccountProof*(db: ReadOnlyLedger, address: Address): seq[seq[byte]] = getAccountProof(distinctBase db, address)
proc getStorageProof*(db: ReadOnlyLedger, address: Address, slots: openArray[UInt256]): seq[seq[seq[byte]]] = getStorageProof(distinctBase db, address, slots)
proc resolveCode*(db: ReadOnlyLedger, address: Address): CodeBytesRef = resolveCode(distinctBase db, address)
# ------------------------------------------------------------------------------
# End
# ------------------------------------------------------------------------------