bsc/core/state/journal.go
zjubfd 2ce00adb55
[R4R] performance improvement in many aspects (#257)
* focus on performance improvement in many aspects.

1. Do BlockBody verification concurrently;
2. Do calculation of intermediate root concurrently;
3. Preload accounts before processing blocks;
4. Make the snapshot layers configurable.
5. Reuse some object to reduce GC.

add

* rlp: improve decoder stream implementation (#22858)

This commit makes various cleanup changes to rlp.Stream.

* rlp: shrink Stream struct

This removes a lot of unused padding space in Stream by reordering the
fields. The size of Stream changes from 120 bytes to 88 bytes. Stream
instances are internally cached and reused using sync.Pool, so this does
not improve performance.

* rlp: simplify list stack

The list stack kept track of the size of the current list context as
well as the current offset into it. The size had to be stored in the
stack in order to subtract it from the remaining bytes of any enclosing
list in ListEnd. It seems that this can be implemented in a simpler
way: just subtract the size from the enclosing list context in List instead.

* rlp: use atomic.Value for type cache (#22902)

All encoding/decoding operations read the type cache to find the
writer/decoder function responsible for a type. When analyzing CPU
profiles of geth during sync, I found that the use of sync.RWMutex in
cache lookups appears in the profiles. It seems we are running into
CPU cache contention problems when package rlp is heavily used
on all CPU cores during sync.

This change makes it use atomic.Value + a writer lock instead of
sync.RWMutex. In the common case where the typeinfo entry is present in
the cache, we simply fetch the map and lookup the type.

* rlp: optimize byte array handling (#22924)

This change improves the performance of encoding/decoding [N]byte.

    name                     old time/op    new time/op    delta
    DecodeByteArrayStruct-8     336ns ± 0%     246ns ± 0%  -26.98%  (p=0.000 n=9+10)
    EncodeByteArrayStruct-8     225ns ± 1%     148ns ± 1%  -34.12%  (p=0.000 n=10+10)

    name                     old alloc/op   new alloc/op   delta
    DecodeByteArrayStruct-8      120B ± 0%       48B ± 0%  -60.00%  (p=0.000 n=10+10)
    EncodeByteArrayStruct-8     0.00B          0.00B          ~     (all equal)

* rlp: optimize big.Int decoding for size <= 32 bytes (#22927)

This change grows the static integer buffer in Stream to 32 bytes,
making it possible to decode 256bit integers without allocating a
temporary buffer.

In the recent commit 088da24, Stream struct size decreased from 120
bytes down to 88 bytes. This commit grows the struct to 112 bytes again,
but the size change will not degrade performance because Stream
instances are internally cached in sync.Pool.

    name             old time/op    new time/op    delta
    DecodeBigInts-8    12.2µs ± 0%     8.6µs ± 4%  -29.58%  (p=0.000 n=9+10)

    name             old speed      new speed      delta
    DecodeBigInts-8   230MB/s ± 0%   326MB/s ± 4%  +42.04%  (p=0.000 n=9+10)

* eth/protocols/eth, les: avoid Raw() when decoding HashOrNumber (#22841)

Getting the raw value is not necessary to decode this type, and
decoding it directly from the stream is faster.

* fix testcase

* debug no lazy

* fix can not repair

* address comments

Co-authored-by: Felix Lange <fjl@twurst.com>
2021-07-29 17:16:53 +08:00

275 lines
6.8 KiB
Go

// Copyright 2016 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
package state
import (
"math/big"
"github.com/ethereum/go-ethereum/common"
)
// journalEntry is a modification entry in the state change journal that can be
// reverted on demand.
type journalEntry interface {
// revert undoes the changes introduced by this journal entry.
revert(*StateDB)
// dirtied returns the Ethereum address modified by this journal entry.
dirtied() *common.Address
}
// journal contains the list of state modifications applied since the last state
// commit. These are tracked to be able to be reverted in case of an execution
// exception or revertal request.
type journal struct {
entries []journalEntry // Current changes tracked by the journal
dirties map[common.Address]int // Dirty accounts and the number of changes
}
// newJournal create a new initialized journal.
func newJournal() *journal {
return &journal{
dirties: make(map[common.Address]int, defaultNumOfSlots),
entries: make([]journalEntry, 0, defaultNumOfSlots),
}
}
// append inserts a new modification entry to the end of the change journal.
func (j *journal) append(entry journalEntry) {
j.entries = append(j.entries, entry)
if addr := entry.dirtied(); addr != nil {
j.dirties[*addr]++
}
}
// revert undoes a batch of journalled modifications along with any reverted
// dirty handling too.
func (j *journal) revert(statedb *StateDB, snapshot int) {
for i := len(j.entries) - 1; i >= snapshot; i-- {
// Undo the changes made by the operation
j.entries[i].revert(statedb)
// Drop any dirty tracking induced by the change
if addr := j.entries[i].dirtied(); addr != nil {
if j.dirties[*addr]--; j.dirties[*addr] == 0 {
delete(j.dirties, *addr)
}
}
}
j.entries = j.entries[:snapshot]
}
// dirty explicitly sets an address to dirty, even if the change entries would
// otherwise suggest it as clean. This method is an ugly hack to handle the RIPEMD
// precompile consensus exception.
func (j *journal) dirty(addr common.Address) {
j.dirties[addr]++
}
// length returns the current number of entries in the journal.
func (j *journal) length() int {
return len(j.entries)
}
type (
// Changes to the account trie.
createObjectChange struct {
account *common.Address
}
resetObjectChange struct {
prev *StateObject
prevdestruct bool
}
suicideChange struct {
account *common.Address
prev bool // whether account had already suicided
prevbalance *big.Int
}
// Changes to individual accounts.
balanceChange struct {
account *common.Address
prev *big.Int
}
nonceChange struct {
account *common.Address
prev uint64
}
storageChange struct {
account *common.Address
key, prevalue common.Hash
}
codeChange struct {
account *common.Address
prevcode, prevhash []byte
}
// Changes to other state values.
refundChange struct {
prev uint64
}
addLogChange struct {
txhash common.Hash
}
addPreimageChange struct {
hash common.Hash
}
touchChange struct {
account *common.Address
}
// Changes to the access list
accessListAddAccountChange struct {
address *common.Address
}
accessListAddSlotChange struct {
address *common.Address
slot *common.Hash
}
)
func (ch createObjectChange) revert(s *StateDB) {
delete(s.stateObjects, *ch.account)
delete(s.stateObjectsDirty, *ch.account)
}
func (ch createObjectChange) dirtied() *common.Address {
return ch.account
}
func (ch resetObjectChange) revert(s *StateDB) {
s.SetStateObject(ch.prev)
if !ch.prevdestruct && s.snap != nil {
delete(s.snapDestructs, ch.prev.addrHash)
}
}
func (ch resetObjectChange) dirtied() *common.Address {
return nil
}
func (ch suicideChange) revert(s *StateDB) {
obj := s.getStateObject(*ch.account)
if obj != nil {
obj.suicided = ch.prev
obj.setBalance(ch.prevbalance)
}
}
func (ch suicideChange) dirtied() *common.Address {
return ch.account
}
var ripemd = common.HexToAddress("0000000000000000000000000000000000000003")
func (ch touchChange) revert(s *StateDB) {
}
func (ch touchChange) dirtied() *common.Address {
return ch.account
}
func (ch balanceChange) revert(s *StateDB) {
s.getStateObject(*ch.account).setBalance(ch.prev)
}
func (ch balanceChange) dirtied() *common.Address {
return ch.account
}
func (ch nonceChange) revert(s *StateDB) {
s.getStateObject(*ch.account).setNonce(ch.prev)
}
func (ch nonceChange) dirtied() *common.Address {
return ch.account
}
func (ch codeChange) revert(s *StateDB) {
s.getStateObject(*ch.account).setCode(common.BytesToHash(ch.prevhash), ch.prevcode)
}
func (ch codeChange) dirtied() *common.Address {
return ch.account
}
func (ch storageChange) revert(s *StateDB) {
s.getStateObject(*ch.account).setState(ch.key, ch.prevalue)
}
func (ch storageChange) dirtied() *common.Address {
return ch.account
}
func (ch refundChange) revert(s *StateDB) {
s.refund = ch.prev
}
func (ch refundChange) dirtied() *common.Address {
return nil
}
func (ch addLogChange) revert(s *StateDB) {
logs := s.logs[ch.txhash]
if len(logs) == 1 {
delete(s.logs, ch.txhash)
} else {
s.logs[ch.txhash] = logs[:len(logs)-1]
}
s.logSize--
}
func (ch addLogChange) dirtied() *common.Address {
return nil
}
func (ch addPreimageChange) revert(s *StateDB) {
delete(s.preimages, ch.hash)
}
func (ch addPreimageChange) dirtied() *common.Address {
return nil
}
func (ch accessListAddAccountChange) revert(s *StateDB) {
/*
One important invariant here, is that whenever a (addr, slot) is added, if the
addr is not already present, the add causes two journal entries:
- one for the address,
- one for the (address,slot)
Therefore, when unrolling the change, we can always blindly delete the
(addr) at this point, since no storage adds can remain when come upon
a single (addr) change.
*/
if s.accessList != nil {
s.accessList.DeleteAddress(*ch.address)
}
}
func (ch accessListAddAccountChange) dirtied() *common.Address {
return nil
}
func (ch accessListAddSlotChange) revert(s *StateDB) {
if s.accessList != nil {
s.accessList.DeleteSlot(*ch.address, *ch.slot)
}
}
func (ch accessListAddSlotChange) dirtied() *common.Address {
return nil
}