nimbus-eth1/nimbus/db/aristo/aristo_compute.nim
Jacek Sieka a056a722eb
Sort subkey lookups by VertexID when computing keys (#2918)
Since data is ordered by VertexID on disk, with this simple trick we can
make much better use of the various rocksdb caches.

Computing the state root of the full mainnet state is down to 4 hours
(from 9) on my laptop.
2024-12-09 08:16:02 +01:00

324 lines
11 KiB
Nim

# nimbus-eth1
# Copyright (c) 2023-2024 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/strformat,
chronicles,
eth/common,
results,
"."/[aristo_desc, aristo_get, aristo_walk/persistent],
./aristo_desc/desc_backend
type WriteBatch = tuple[writer: PutHdlRef, count: int, depth: int, prefix: uint64]
# Keep write batch size _around_ 1mb, give or take some overhead - this is a
# tradeoff between efficiency and memory usage with diminishing returns the
# larger it is..
const batchSize = 1024 * 1024 div (sizeof(RootedVertexID) + sizeof(HashKey))
proc flush(batch: var WriteBatch, db: AristoDbRef): Result[void, AristoError] =
if batch.writer != nil:
?db.backend.putEndFn batch.writer
batch.writer = nil
ok()
proc putVtx(
batch: var WriteBatch,
db: AristoDbRef,
rvid: RootedVertexID,
vtx: VertexRef,
key: HashKey,
): Result[void, AristoError] =
if batch.writer == nil:
doAssert db.backend != nil, "source data is from the backend"
batch.writer = ?db.backend.putBegFn()
db.backend.putVtxFn(batch.writer, rvid, vtx, key)
batch.count += 1
ok()
func progress(batch: WriteBatch): string =
# Return an approximation on how much of the keyspace has been covered by
# looking at the path prefix that we're currently processing
&"{(float(batch.prefix) / float(uint64.high)) * 100:02.2f}%"
func enter(batch: var WriteBatch, nibble: uint8) =
batch.depth += 1
if batch.depth <= 16:
batch.prefix += uint64(nibble) shl ((16 - batch.depth) * 4)
func leave(batch: var WriteBatch, nibble: uint8) =
if batch.depth <= 16:
batch.prefix -= uint64(nibble) shl ((16 - batch.depth) * 4)
batch.depth -= 1
proc putKeyAtLevel(
db: AristoDbRef,
rvid: RootedVertexID,
vtx: VertexRef,
key: HashKey,
level: int,
batch: var WriteBatch,
): Result[void, AristoError] =
## Store a hash key in the given layer or directly to the underlying database
## which helps ensure that memory usage is proportional to the pending change
## set (vertex data may have been committed to disk without computing the
## corresponding hash!)
if level == -2:
?batch.putVtx(db, rvid, vtx, key)
if batch.count mod batchSize == 0:
?batch.flush(db)
if batch.count mod (batchSize * 100) == 0:
info "Writing computeKey cache", keys = batch.count, accounts = batch.progress
else:
debug "Writing computeKey cache", keys = batch.count, accounts = batch.progress
else:
db.deltaAtLevel(level).sTab[rvid] = vtx
db.deltaAtLevel(level).kMap[rvid] = key
ok()
func maxLevel(cur, other: int): int =
# Compare two levels and return the topmost in the stack, taking into account
# the odd reversal of order around the zero point
if cur < 0:
max(cur, other) # >= 0 is always more topmost than <0
elif other < 0:
cur
else:
min(cur, other) # Here the order is reversed and 0 is the top layer
template encodeLeaf(w: var RlpWriter, pfx: NibblesBuf, leafData: untyped): HashKey =
w.startList(2)
w.append(pfx.toHexPrefix(isLeaf = true).data())
w.append(leafData)
w.finish().digestTo(HashKey)
template encodeBranch(w: var RlpWriter, vtx: VertexRef, subKeyForN: untyped): HashKey =
w.startList(17)
for (n {.inject.}, subvid {.inject.}) in vtx.allPairs():
w.append(subKeyForN)
w.append EmptyBlob
w.finish().digestTo(HashKey)
template encodeExt(w: var RlpWriter, pfx: NibblesBuf, branchKey: HashKey): HashKey =
w.startList(2)
w.append(pfx.toHexPrefix(isLeaf = false).data())
w.append(branchKey)
w.finish().digestTo(HashKey)
proc getKey(
db: AristoDbRef, rvid: RootedVertexID, skipLayers: static bool
): Result[((HashKey, VertexRef), int), AristoError] =
ok when skipLayers:
(?db.getKeyUbe(rvid, {GetVtxFlag.PeekCache}), -2)
else:
?db.getKeyRc(rvid, {})
template childVid(v: VertexRef): VertexID =
# If we have to recurse into a child, where would that recusion start?
case v.vType
of Leaf:
if v.lData.pType == AccountData and v.lData.stoID.isValid:
v.lData.stoID.vid
else:
default(VertexID)
of Branch:
v.startVid
proc computeKeyImpl(
db: AristoDbRef,
rvid: RootedVertexID,
batch: var WriteBatch,
vtx: VertexRef,
level: int,
skipLayers: static bool,
): Result[(HashKey, int), AristoError] =
# The bloom filter available used only when creating the key cache from an
# empty state
# Top-most level of all the verticies this hash computation depends on
var level = level
# TODO this is the same code as when serializing NodeRef, without the NodeRef
var writer = initRlpWriter()
let key =
case vtx.vType
of Leaf:
writer.encodeLeaf(vtx.pfx):
case vtx.lData.pType
of AccountData:
let
stoID = vtx.lData.stoID
skey =
if stoID.isValid:
let
keyvtxl = ?db.getKey((stoID.vid, stoID.vid), skipLayers)
(skey, sl) =
if keyvtxl[0][0].isValid:
(keyvtxl[0][0], keyvtxl[1])
else:
?db.computeKeyImpl(
(stoID.vid, stoID.vid),
batch,
keyvtxl[0][1],
keyvtxl[1],
skipLayers = skipLayers,
)
level = maxLevel(level, sl)
skey
else:
VOID_HASH_KEY
rlp.encode Account(
nonce: vtx.lData.account.nonce,
balance: vtx.lData.account.balance,
storageRoot: skey.to(Hash32),
codeHash: vtx.lData.account.codeHash,
)
of StoData:
# TODO avoid memory allocation when encoding storage data
rlp.encode(vtx.lData.stoData)
of Branch:
# For branches, we need to load the vertices before recursing into them
# to exploit their on-disk order
var keyvtxs: array[16, ((HashKey, VertexRef), int)]
for n, subvid in vtx.pairs:
keyvtxs[n] = ?db.getKey((rvid.root, subvid), skipLayers)
# Make sure we have keys computed for each hash
block keysComputed:
while true:
# Compute missing keys in the order of the child vid that we have to
# recurse into, again exploiting on-disk order - this more than
# doubles computeKey speed on a fresh database!
var
minVid = default(VertexID)
minIdx = keyvtxs.len + 1 # index where the minvid can be found
n = 0'u8 # number of already-processed keys, for the progress bar
# The O(n^2) sort/search here is fine given the small size of the list
for nibble, keyvtx in keyvtxs.mpairs:
let subvid = vtx.bVid(uint8 nibble)
if (not subvid.isValid) or keyvtx[0][0].isValid:
n += 1 # no need to compute key
continue
let childVid = keyvtx[0][1].childVid
if not childVid.isValid:
# leaf vertex without storage ID - we can compute the key trivially
(keyvtx[0][0], keyvtx[1]) =
?db.computeKeyImpl(
(rvid.root, subvid),
batch,
keyvtx[0][1],
keyvtx[1],
skipLayers = skipLayers,
)
n += 1
continue
if minIdx == keyvtxs.len + 1 or childVid < minVid:
minIdx = nibble
minVid = childVid
if minIdx == keyvtxs.len + 1: # no uncomputed key found!
break keysComputed
batch.enter(n)
(keyvtxs[minIdx][0][0], keyvtxs[minIdx][1]) =
?db.computeKeyImpl(
(rvid.root, vtx.bVid(uint8 minIdx)),
batch,
keyvtxs[minIdx][0][1],
keyvtxs[minIdx][1],
skipLayers = skipLayers,
)
batch.leave(n)
template writeBranch(w: var RlpWriter): HashKey =
w.encodeBranch(vtx):
if subvid.isValid:
level = maxLevel(level, keyvtxs[n][1])
keyvtxs[n][0][0]
else:
VOID_HASH_KEY
if vtx.pfx.len > 0: # Extension node
writer.encodeExt(vtx.pfx):
var bwriter = initRlpWriter()
bwriter.writeBranch()
else:
writer.writeBranch()
# Cache the hash into the same storage layer as the the top-most value that it
# depends on (recursively) - this could be an ephemeral in-memory layer or the
# underlying database backend - typically, values closer to the root are more
# likely to live in an in-memory layer since any leaf change will lead to the
# root key also changing while leaves that have never been hashed will see
# their hash being saved directly to the backend.
if vtx.vType != Leaf:
?db.putKeyAtLevel(rvid, vtx, key, level, batch)
ok (key, level)
proc computeKeyImpl(
db: AristoDbRef, rvid: RootedVertexID, skipLayers: static bool
): Result[HashKey, AristoError] =
let (keyvtx, level) =
when skipLayers:
(?db.getKeyUbe(rvid, {GetVtxFlag.PeekCache}), -2)
else:
?db.getKeyRc(rvid, {})
if keyvtx[0].isValid:
return ok(keyvtx[0])
var batch: WriteBatch
let res = computeKeyImpl(db, rvid, batch, keyvtx[1], level, skipLayers = skipLayers)
if res.isOk:
?batch.flush(db)
if batch.count > 0:
if batch.count >= batchSize * 100:
info "Wrote computeKey cache", keys = batch.count, accounts = "100.00%"
else:
debug "Wrote computeKey cache", keys = batch.count, accounts = "100.00%"
ok (?res)[0]
proc computeKey*(
db: AristoDbRef, # Database, top layer
rvid: RootedVertexID, # Vertex to convert
): Result[HashKey, AristoError] =
## Compute the key for an arbitrary vertex ID. If successful, the length of
## the resulting key might be smaller than 32. If it is used as a root vertex
## state/hash, it must be converted to a `Hash32` (using (`.to(Hash32)`) as
## in `db.computeKey(rvid).value.to(Hash32)` which always results in a
## 32 byte value.
computeKeyImpl(db, rvid, skipLayers = false)
proc computeKeys*(db: AristoDbRef, root: VertexID): Result[void, AristoError] =
## Ensure that key cache is topped up with the latest state root
discard db.computeKeyImpl((root, root), skipLayers = true)
ok()
# ------------------------------------------------------------------------------
# End
# ------------------------------------------------------------------------------