nimbus-eth1/nimbus/sync/snap/worker/db/hexary_desc.nim

429 lines
15 KiB
Nim
Raw Normal View History

# nimbus-eth1
# Copyright (c) 2021 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.
import
std/[algorithm, hashes, sequtils, sets, strutils, tables],
eth/[common/eth_types, p2p, trie/nibbles],
stint,
../../range_desc,
./hexary_error
{.push raises: [Defect].}
type
HexaryPpFn* = proc(key: RepairKey): string {.gcsafe.}
## For testing/debugging: key pretty printer function
ByteArray33* = array[33,byte]
## Used for 31 byte database keys, i.e. <marker> + <32-byte-key>
RepairKey* = distinct ByteArray33
## Byte prefixed `NodeKey` for internal DB records
# Example trie from https://eth.wiki/en/fundamentals/patricia-tree
#
# lookup data:
# "do": "verb"
# "dog": "puppy"
# "dodge": "coin"
# "horse": "stallion"
#
# trie DB:
# root: [16 A]
# A: [* * * * B * * * [20+"orse" "stallion"] * * * * * * * *]
# B: [00+"o" D]
# D: [* * * * * * E * * * * * * * * * "verb"]
# E: [17 [* * * * * * [35 "coin"] * * * * * * * * * "puppy"]]
#
# with first nibble of two-column rows:
# hex bits | node type length
# ---------+------------------
# 0 0000 | extension even
# 1 0001 | extension odd
# 2 0010 | leaf even
# 3 0011 | leaf odd
#
# and key path:
# "do": 6 4 6 f
# "dog": 6 4 6 f 6 7
# "dodge": 6 4 6 f 6 7 6 5
# "horse": 6 8 6 f 7 2 7 3 6 5
NodeKind* = enum
Branch
Extension
Leaf
RNodeState* = enum
Static = 0 ## Inserted as proof record
Locked ## Like `Static`, only added on-the-fly
Mutable ## Open for modification
TmpRoot ## Mutable root node
RNodeRef* = ref object
## Node for building a temporary hexary trie coined `repair tree`.
state*: RNodeState ## `Static` if added from proof data set
case kind*: NodeKind
of Leaf:
lPfx*: NibblesSeq ## Portion of path segment
lData*: Blob
of Extension:
ePfx*: NibblesSeq ## Portion of path segment
eLink*: RepairKey ## Single down link
of Branch:
bLink*: array[16,RepairKey] ## Down links
#
# Paraphrased comment from Andri's `stateless/readme.md` file in chapter
# `Deviation from yellow paper`, (also found here
# github.com/status-im/nimbus-eth1
# /tree/master/stateless#deviation-from-yellow-paper)
# [..] In the Yellow Paper, the 17th elem of the branch node can contain
# a value. But it is always empty in a real Ethereum state trie. The
# block witness spec also ignores this 17th elem when encoding or
# decoding a branch node. This can happen because in a Ethereum secure
# hexary trie, every keys have uniform length of 32 bytes or 64 nibbles.
# With the absence of the 17th element, a branch node will never contain
# a leaf value.
bData*: Blob
XNodeObj* = object
## Simplified version of `RNodeRef` to be used as a node for `XPathStep`
case kind*: NodeKind
of Leaf:
lPfx*: NibblesSeq ## Portion of path segment
lData*: Blob
of Extension:
ePfx*: NibblesSeq ## Portion of path segment
eLink*: Blob ## Single down link
of Branch:
bLink*: array[17,Blob] ## Down links followed by data
RPathStep* = object
## For constructing a repair tree traversal path `RPath`
key*: RepairKey ## Tree label, node hash
node*: RNodeRef ## Referes to data record
nibble*: int8 ## Branch node selector (if any)
RPath* = object
path*: seq[RPathStep]
tail*: NibblesSeq ## Stands for non completed leaf path
XPathStep* = object
## Similar to `RPathStep` for an arbitrary (sort of transparent) trie
key*: Blob ## Node hash implied by `node` data
node*: XNodeObj
nibble*: int8 ## Branch node selector (if any)
XPath* = object
path*: seq[XPathStep]
tail*: NibblesSeq ## Stands for non completed leaf path
depth*: int ## May indicate path length (typically 64)
RLeafSpecs* = object
## Temporarily stashed leaf data (as for an account.) Proper records
## have non-empty payload. Records with empty payload are administrative
## items, e.g. lower boundary records.
pathTag*: NodeTag ## Equivalent to account hash
nodeKey*: RepairKey ## Leaf hash into hexary repair table
payload*: Blob ## Data payload
TrieNodeStat* = object
## Trie inspection report
dangling*: seq[Blob] ## Paths from nodes with incomplete refs
level*: int ## Maximim nesting depth of dangling nodes
stopped*: bool ## Potential loop detected if `true`
HexaryTreeDbRef* = ref object
## Hexary trie plus helper structures
tab*: Table[RepairKey,RNodeRef] ## key-value trie table, in-memory db
repairKeyGen*: uint64 ## Unique tmp key generator
keyPp*: HexaryPpFn ## For debugging, might go away
HexaryGetFn* = proc(key: openArray[byte]): Blob {.gcsafe.}
## Persistent database `get()` function. For read-only cases, this function
## can be seen as the persistent alternative to ``tab[]` on a
## `HexaryTreeDbRef` descriptor.
HexaryNodeReport* = object
## Return code for single node operations
slot*: Option[int] ## May refer to indexed argument slots
kind*: Option[NodeKind] ## Node type (if any)
error*: HexaryDbError ## Error code, or `NothingSerious`
const
EmptyNodeBlob* = seq[byte].default
EmptyNibbleRange* = EmptyNodeBlob.initNibbleRange
static:
# Not that there is no doubt about this ...
doAssert NodeKey.default.ByteArray32.initNibbleRange.len == 64
var
disablePrettyKeys* = false ## Degugging, print raw keys if `true`
# ------------------------------------------------------------------------------
# Private helpers
# ------------------------------------------------------------------------------
proc initImpl(key: var RepairKey; data: openArray[byte]): bool =
key.reset
if 0 < data.len and data.len <= 33:
let trg = addr key.ByteArray33[33 - data.len]
trg.copyMem(unsafeAddr data[0], data.len)
return true
# ------------------------------------------------------------------------------
# Private debugging helpers
# ------------------------------------------------------------------------------
proc to*(key: NodeKey; T: type RepairKey): T {.gcsafe.}
proc toPfx(indent: int): string =
"\n" & " ".repeat(indent)
proc ppImpl(s: string; hex = false): string =
## For long strings print `begin..end` only
if hex:
let n = (s.len + 1) div 2
(if s.len < 20: s else: s[0 .. 5] & ".." & s[s.len-8 .. s.len-1]) &
"[" & (if 0 < n: "#" & $n else: "") & "]"
elif s.len <= 30:
s
else:
(if (s.len and 1) == 0: s[0 ..< 8] else: "0" & s[0 ..< 7]) &
"..(" & $s.len & ").." & s[s.len-16 ..< s.len]
proc ppImpl(key: RepairKey; db: HexaryTreeDbRef): string =
try:
if not disablePrettyKeys and not db.keyPp.isNil:
return db.keyPp(key)
except:
discard
key.ByteArray33.toSeq.mapIt(it.toHex(2)).join.toLowerAscii
proc ppImpl(key: NodeKey; db: HexaryTreeDbRef): string =
key.to(RepairKey).ppImpl(db)
proc ppImpl(w: openArray[RepairKey]; db: HexaryTreeDbRef): string =
w.mapIt(it.ppImpl(db)).join(",")
proc ppImpl(w: openArray[Blob]; db: HexaryTreeDbRef): string =
var q: seq[RepairKey]
for a in w:
var key: RepairKey
discard key.initImpl(a)
q.add key
q.ppImpl(db)
proc ppStr(blob: Blob): string =
if blob.len == 0: ""
else: blob.mapIt(it.toHex(2)).join.toLowerAscii.ppImpl(hex = true)
proc ppImpl(n: RNodeRef; db: HexaryTreeDbRef): string =
let so = n.state.ord
case n.kind:
of Leaf:
["l","ł","L","R"][so] & "(" & $n.lPfx & "," & n.lData.ppStr & ")"
of Extension:
["e","","E","R"][so] & "(" & $n.ePfx & "," & n.eLink.ppImpl(db) & ")"
of Branch:
["b","þ","B","R"][so] & "(" & n.bLink.ppImpl(db) & "," & n.bData.ppStr & ")"
proc ppImpl(n: XNodeObj; db: HexaryTreeDbRef): string =
case n.kind:
of Leaf:
"l(" & $n.lPfx & "," & n.lData.ppStr & ")"
of Extension:
var key: RepairKey
discard key.initImpl(n.eLink)
"e(" & $n.ePfx & "," & key.ppImpl(db) & ")"
of Branch:
"b(" & n.bLink[0..15].ppImpl(db) & "," & n.bLink[16].ppStr & ")"
proc ppImpl(w: RPathStep; db: HexaryTreeDbRef): string =
let
nibble = if 0 <= w.nibble: w.nibble.toHex(1).toLowerAscii else: "ø"
key = w.key.ppImpl(db)
"(" & key & "," & nibble & "," & w.node.ppImpl(db) & ")"
proc ppImpl(w: XPathStep; db: HexaryTreeDbRef): string =
let nibble = if 0 <= w.nibble: w.nibble.toHex(1).toLowerAscii else: "ø"
var key: RepairKey
discard key.initImpl(w.key)
"(" & key.ppImpl(db) & "," & $nibble & "," & w.node.ppImpl(db) & ")"
proc ppImpl(db: HexaryTreeDbRef; root: NodeKey): seq[string] =
## Dump the entries from the a generic repair tree. This function assumes
## that mapped keys are printed `$###` if a node is locked or static, and
## some substitute for the first letter `$` otherwise (if they are mutable.)
proc toKey(s: string): uint64 =
try:
result = s[1 ..< s.len].parseUint
except ValueError as e:
raiseAssert "Ooops ppImpl(s=" & s & "): name=" & $e.name & " msg=" & e.msg
if s[0] != '$':
result = result or (1u64 shl 63)
proc cmpIt(x, y: (uint64,string)): int =
cmp(x[0],y[0])
try:
var accu: seq[(uint64,string)]
if root.ByteArray32 != ByteArray32.default:
accu.add @[(0u64, "($0" & "," & root.ppImpl(db) & ")")]
for key,node in db.tab.pairs:
accu.add (
key.ppImpl(db).tokey,
"(" & key.ppImpl(db) & "," & node.ppImpl(db) & ")")
result = accu.sorted(cmpIt).mapIt(it[1])
except Exception as e:
result &= " ! Ooops ppImpl(): name=" & $e.name & " msg=" & e.msg
proc ppDangling(a: seq[Blob]; maxItems = 30): string =
proc ppBlob(w: Blob): string =
w.mapIt(it.toHex(2)).join.toLowerAscii
let
Prep for full sync after snap (#1253) * Split fetch accounts into sub-modules details: There will be separated modules for accounts snapshot, storage snapshot, and healing for either. * Allow to rebase pivot before negotiated header why: Peers seem to have not too many snapshots available. By setting back the pivot block header slightly, the chances might be higher to find more peers to serve this pivot. Experiment on mainnet showed that setting back too much (tested with 1024), the chances to find matching snapshot peers seem to decrease. * Add accounts healing * Update variable/field naming in `worker_desc` for readability * Handle leaf nodes in accounts healing why: There is no need to fetch accounts when they had been added by the healing process. On the flip side, these accounts must be checked for storage data and the batch queue updated, accordingly. * Reorganising accounts hash ranges batch queue why: The aim is to formally cover as many accounts as possible for different pivot state root environments. Formerly, this was tried by starting the accounts batch queue at a random value for each pivot (and wrapping around.) Now, each pivot environment starts with an interval set mutually disjunct from any interval set retrieved with other pivot state roots. also: Stop fishing for more pivots in `worker` if 100% download is reached * Reorganise/update accounts healing why: Error handling was wrong and the (math. complexity of) whole process could be better managed. details: Much of the algorithm is now documented at the top of the file `heal_accounts.nim`
2022-10-08 17:20:50 +00:00
q = a.mapIt(it.ppBlob)[0 ..< min(maxItems,a.len)]
andMore = if maxItems < a.len: ", ..[#" & $a.len & "].." else: ""
"{" & q.join(",") & andMore & "}"
# ------------------------------------------------------------------------------
# Public debugging helpers
# ------------------------------------------------------------------------------
proc pp*(s: string; hex = false): string =
## For long strings print `begin..end` only
s.ppImpl(hex)
proc pp*(w: NibblesSeq): string =
$w
proc pp*(key: RepairKey): string =
## Raw key, for referenced key dump use `key.pp(db)` below
key.ByteArray33.toSeq.mapIt(it.toHex(2)).join.tolowerAscii
proc pp*(key: NodeKey): string =
## Raw key, for referenced key dump use `key.pp(db)` below
key.ByteArray32.toSeq.mapIt(it.toHex(2)).join.tolowerAscii
proc pp*(key: NodeKey|RepairKey; db: HexaryTreeDbRef): string =
key.ppImpl(db)
proc pp*(w: RNodeRef|XNodeObj|RPathStep; db: HexaryTreeDbRef): string =
w.ppImpl(db)
proc pp*(w:openArray[RPathStep|XPathStep];db:HexaryTreeDbRef;indent=4): string =
w.toSeq.mapIt(it.ppImpl(db)).join(indent.toPfx)
proc pp*(w: RPath; db: HexaryTreeDbRef; indent=4): string =
w.path.pp(db,indent) & indent.toPfx & "(" & $w.tail & ")"
proc pp*(w: XPath; db: HexaryTreeDbRef; indent=4): string =
w.path.pp(db,indent) & indent.toPfx & "(" & $w.tail & "," & $w.depth & ")"
proc pp*(db: HexaryTreeDbRef; root: NodeKey; indent=4): string =
## Dump the entries from the a generic repair tree.
db.ppImpl(root).join(indent.toPfx)
proc pp*(db: HexaryTreeDbRef; indent=4): string =
## varinat of `pp()` above
db.ppImpl(NodeKey.default).join(indent.toPfx)
proc pp*(a: TrieNodeStat; db: HexaryTreeDbRef; maxItems = 30): string =
Prep for full sync after snap (#1253) * Split fetch accounts into sub-modules details: There will be separated modules for accounts snapshot, storage snapshot, and healing for either. * Allow to rebase pivot before negotiated header why: Peers seem to have not too many snapshots available. By setting back the pivot block header slightly, the chances might be higher to find more peers to serve this pivot. Experiment on mainnet showed that setting back too much (tested with 1024), the chances to find matching snapshot peers seem to decrease. * Add accounts healing * Update variable/field naming in `worker_desc` for readability * Handle leaf nodes in accounts healing why: There is no need to fetch accounts when they had been added by the healing process. On the flip side, these accounts must be checked for storage data and the batch queue updated, accordingly. * Reorganising accounts hash ranges batch queue why: The aim is to formally cover as many accounts as possible for different pivot state root environments. Formerly, this was tried by starting the accounts batch queue at a random value for each pivot (and wrapping around.) Now, each pivot environment starts with an interval set mutually disjunct from any interval set retrieved with other pivot state roots. also: Stop fishing for more pivots in `worker` if 100% download is reached * Reorganise/update accounts healing why: Error handling was wrong and the (math. complexity of) whole process could be better managed. details: Much of the algorithm is now documented at the top of the file `heal_accounts.nim`
2022-10-08 17:20:50 +00:00
result = "(" & $a.level
if a.stopped:
result &= "stopped,"
Prep for full sync after snap (#1253) * Split fetch accounts into sub-modules details: There will be separated modules for accounts snapshot, storage snapshot, and healing for either. * Allow to rebase pivot before negotiated header why: Peers seem to have not too many snapshots available. By setting back the pivot block header slightly, the chances might be higher to find more peers to serve this pivot. Experiment on mainnet showed that setting back too much (tested with 1024), the chances to find matching snapshot peers seem to decrease. * Add accounts healing * Update variable/field naming in `worker_desc` for readability * Handle leaf nodes in accounts healing why: There is no need to fetch accounts when they had been added by the healing process. On the flip side, these accounts must be checked for storage data and the batch queue updated, accordingly. * Reorganising accounts hash ranges batch queue why: The aim is to formally cover as many accounts as possible for different pivot state root environments. Formerly, this was tried by starting the accounts batch queue at a random value for each pivot (and wrapping around.) Now, each pivot environment starts with an interval set mutually disjunct from any interval set retrieved with other pivot state roots. also: Stop fishing for more pivots in `worker` if 100% download is reached * Reorganise/update accounts healing why: Error handling was wrong and the (math. complexity of) whole process could be better managed. details: Much of the algorithm is now documented at the top of the file `heal_accounts.nim`
2022-10-08 17:20:50 +00:00
result &= $a.dangling.len & "," &
a.dangling.ppDangling(maxItems) & ")"
# ------------------------------------------------------------------------------
# Public constructor (or similar)
# ------------------------------------------------------------------------------
proc init*(key: var RepairKey; data: openArray[byte]): bool =
key.initImpl(data)
proc newRepairKey*(db: HexaryTreeDbRef): RepairKey =
db.repairKeyGen.inc
var src = db.repairKeyGen.toBytesBE
(addr result.ByteArray33[25]).copyMem(addr src[0], 8)
result.ByteArray33[0] = 1
# ------------------------------------------------------------------------------
# Public functions
# ------------------------------------------------------------------------------
proc hash*(a: RepairKey): Hash =
## Tables mixin
a.ByteArray33.hash
proc `==`*(a, b: RepairKey): bool =
## Tables mixin
a.ByteArray33 == b.ByteArray33
proc to*(key: NodeKey; T: type NibblesSeq): T =
key.ByteArray32.initNibbleRange
proc to*(key: NodeKey; T: type RepairKey): T =
(addr result.ByteArray33[1]).copyMem(unsafeAddr key.ByteArray32[0], 32)
proc isZero*[T: NodeTag|NodeKey|RepairKey](a: T): bool =
a == T.default
proc isNodeKey*(a: RepairKey): bool =
a.ByteArray33[0] == 0
proc digestTo*(data: Blob; T: type NodeKey): T =
keccakHash(data).data.T
proc convertTo*(data: Blob; T: type NodeKey): T =
## Probably lossy conversion, use `init()` for safe conversion
discard result.init(data)
proc convertTo*(data: Blob; T: type RepairKey): T =
## Probably lossy conversion, use `init()` for safe conversion
discard result.initImpl(data)
proc convertTo*(node: RNodeRef; T: type Blob): T =
## Write the node as an RLP-encoded blob
var writer = initRlpWriter()
proc appendOk(writer: var RlpWriter; key: RepairKey): bool =
if key.isZero:
writer.append(EmptyNodeBlob)
elif key.isNodeKey:
var hash: Hash256
(addr hash.data[0]).copyMem(unsafeAddr key.ByteArray33[1], 32)
writer.append(hash)
else:
return false
true
case node.kind:
of Branch:
writer.startList(17)
for n in 0 ..< 16:
if not writer.appendOk(node.bLink[n]):
return # empty `Blob`
writer.append(node.bData)
of Extension:
writer.startList(2)
writer.append(node.ePfx.hexPrefixEncode(isleaf = false))
if not writer.appendOk(node.eLink):
return # empty `Blob`
of Leaf:
writer.startList(2)
writer.append(node.lPfx.hexPrefixEncode(isleaf = true))
writer.append(node.lData)
writer.finish()
# ------------------------------------------------------------------------------
# End
# ------------------------------------------------------------------------------