manual geth v1.11.1 merge

core/state/database.go

@@ -20,13 +20,12 @@ import (
 	"errors"
 	"fmt"
 
-	"github.com/VictoriaMetrics/fastcache"
 	"github.com/ethereum/go-ethereum/common"
+	"github.com/ethereum/go-ethereum/common/lru"
 	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/trie"
-	lru "github.com/hashicorp/golang-lru"
 )
 
 const (
@@ -43,7 +42,7 @@ type Database interface {
 	OpenTrie(root common.Hash) (Trie, error)
 
 	// OpenStorageTrie opens the storage trie of an account.
-	OpenStorageTrie(addrHash, root common.Hash) (Trie, error)
+	OpenStorageTrie(stateRoot common.Hash, addrHash, root common.Hash) (Trie, error)
 
 	// CopyTrie returns an independent copy of the given trie.
 	CopyTrie(Trie) Trie
@@ -54,6 +53,9 @@ type Database interface {
 	// ContractCodeSize retrieves a particular contracts code's size.
 	ContractCodeSize(addrHash, codeHash common.Hash) (int, error)
 
+	// DiskDB returns the underlying key-value disk database.
+	DiskDB() ethdb.KeyValueStore
+
 	// TrieDB retrieves the low level trie database used for data storage.
 	TrieDB() *trie.Database
 }
@@ -71,8 +73,13 @@ type Trie interface {
 	// trie.MissingNodeError is returned.
 	TryGet(key []byte) ([]byte, error)
 
-	// TryGetAccount abstract an account read from the trie.
-	TryGetAccount(key []byte) (*types.StateAccount, error)
+	// TryGetAccount abstracts an account read from the trie. It retrieves the
+	// account blob from the trie with provided account address and decodes it
+	// with associated decoding algorithm. If the specified account is not in
+	// the trie, nil will be returned. If the trie is corrupted(e.g. some nodes
+	// are missing or the account blob is incorrect for decoding), an error will
+	// be returned.
+	TryGetAccount(address common.Address) (*types.StateAccount, error)
 
 	// TryUpdate associates key with value in the trie. If value has length zero, any
 	// existing value is deleted from the trie. The value bytes must not be modified
@@ -80,15 +87,17 @@ type Trie interface {
 	// database, a trie.MissingNodeError is returned.
 	TryUpdate(key, value []byte) error
 
-	// TryUpdateAccount abstract an account write to the trie.
-	TryUpdateAccount(key []byte, account *types.StateAccount) error
+	// TryUpdateAccount abstracts an account write to the trie. It encodes the
+	// provided account object with associated algorithm and then updates it
+	// in the trie with provided address.
+	TryUpdateAccount(address common.Address, account *types.StateAccount) error
 
 	// TryDelete removes any existing value for key from the trie. If a node was not
 	// found in the database, a trie.MissingNodeError is returned.
 	TryDelete(key []byte) error
 
 	// TryDeleteAccount abstracts an account deletion from the trie.
-	TryDeleteAccount(key []byte) error
+	TryDeleteAccount(address common.Address) error
 
 	// Hash returns the root hash of the trie. It does not write to the database and
 	// can be used even if the trie doesn't have one.
@@ -100,7 +109,7 @@ type Trie interface {
 	// The returned nodeset can be nil if the trie is clean(nothing to commit).
 	// Once the trie is committed, it's not usable anymore. A new trie must
 	// be created with new root and updated trie database for following usage
-	Commit(collectLeaf bool) (common.Hash, *trie.NodeSet, error)
+	Commit(collectLeaf bool) (common.Hash, *trie.NodeSet)
 
 	// NodeIterator returns an iterator that returns nodes of the trie. Iteration
 	// starts at the key after the given start key.
@@ -127,23 +136,34 @@ func NewDatabase(db ethdb.Database) Database {
 // is safe for concurrent use and retains a lot of collapsed RLP trie nodes in a
 // large memory cache.
 func NewDatabaseWithConfig(db ethdb.Database, config *trie.Config) Database {
-	csc, _ := lru.New(codeSizeCacheSize)
 	return &cachingDB{
-		db:            trie.NewDatabaseWithConfig(db, config),
-		codeSizeCache: csc,
-		codeCache:     fastcache.New(codeCacheSize),
+		disk:          db,
+		codeSizeCache: lru.NewCache[common.Hash, int](codeSizeCacheSize),
+		codeCache:     lru.NewSizeConstrainedCache[common.Hash, []byte](codeCacheSize),
+		triedb:        trie.NewDatabaseWithConfig(db, config),
+	}
+}
+
+// NewDatabaseWithNodeDB creates a state database with an already initialized node database.
+func NewDatabaseWithNodeDB(db ethdb.Database, triedb *trie.Database) Database {
+	return &cachingDB{
+		disk:          db,
+		codeSizeCache: lru.NewCache[common.Hash, int](codeSizeCacheSize),
+		codeCache:     lru.NewSizeConstrainedCache[common.Hash, []byte](codeCacheSize),
+		triedb:        triedb,
 	}
 }
 
 type cachingDB struct {
-	db            *trie.Database
-	codeSizeCache *lru.Cache
-	codeCache     *fastcache.Cache
+	disk          ethdb.KeyValueStore
+	codeSizeCache *lru.Cache[common.Hash, int]
+	codeCache     *lru.SizeConstrainedCache[common.Hash, []byte]
+	triedb        *trie.Database
 }
 
 // OpenTrie opens the main account trie at a specific root hash.
 func (db *cachingDB) OpenTrie(root common.Hash) (Trie, error) {
-	tr, err := trie.NewStateTrie(common.Hash{}, root, db.db)
+	tr, err := trie.NewStateTrie(trie.StateTrieID(root), db.triedb)
 	if err != nil {
 		return nil, err
 	}
@@ -151,8 +171,8 @@ func (db *cachingDB) OpenTrie(root common.Hash) (Trie, error) {
 }
 
 // OpenStorageTrie opens the storage trie of an account.
-func (db *cachingDB) OpenStorageTrie(addrHash, root common.Hash) (Trie, error) {
-	tr, err := trie.NewStateTrie(addrHash, root, db.db)
+func (db *cachingDB) OpenStorageTrie(stateRoot common.Hash, addrHash, root common.Hash) (Trie, error) {
+	tr, err := trie.NewStateTrie(trie.StorageTrieID(stateRoot, addrHash, root), db.triedb)
 	if err != nil {
 		return nil, err
 	}
@@ -171,12 +191,13 @@ func (db *cachingDB) CopyTrie(t Trie) Trie {
 
 // ContractCode retrieves a particular contract's code.
 func (db *cachingDB) ContractCode(addrHash, codeHash common.Hash) ([]byte, error) {
-	if code := db.codeCache.Get(nil, codeHash.Bytes()); len(code) > 0 {
+	code, _ := db.codeCache.Get(codeHash)
+	if len(code) > 0 {
 		return code, nil
 	}
-	code := rawdb.ReadCode(db.db.DiskDB(), codeHash)
+	code = rawdb.ReadCode(db.disk, codeHash)
 	if len(code) > 0 {
-		db.codeCache.Set(codeHash.Bytes(), code)
+		db.codeCache.Add(codeHash, code)
 		db.codeSizeCache.Add(codeHash, len(code))
 		return code, nil
 	}
@@ -187,12 +208,13 @@ func (db *cachingDB) ContractCode(addrHash, codeHash common.Hash) ([]byte, error
 // code can't be found in the cache, then check the existence with **new**
 // db scheme.
 func (db *cachingDB) ContractCodeWithPrefix(addrHash, codeHash common.Hash) ([]byte, error) {
-	if code := db.codeCache.Get(nil, codeHash.Bytes()); len(code) > 0 {
+	code, _ := db.codeCache.Get(codeHash)
+	if len(code) > 0 {
 		return code, nil
 	}
-	code := rawdb.ReadCodeWithPrefix(db.db.DiskDB(), codeHash)
+	code = rawdb.ReadCodeWithPrefix(db.disk, codeHash)
 	if len(code) > 0 {
-		db.codeCache.Set(codeHash.Bytes(), code)
+		db.codeCache.Add(codeHash, code)
 		db.codeSizeCache.Add(codeHash, len(code))
 		return code, nil
 	}
@@ -202,13 +224,18 @@ func (db *cachingDB) ContractCodeWithPrefix(addrHash, codeHash common.Hash) ([]b
 // ContractCodeSize retrieves a particular contracts code's size.
 func (db *cachingDB) ContractCodeSize(addrHash, codeHash common.Hash) (int, error) {
 	if cached, ok := db.codeSizeCache.Get(codeHash); ok {
-		return cached.(int), nil
+		return cached, nil
 	}
 	code, err := db.ContractCode(addrHash, codeHash)
 	return len(code), err
 }
 
+// DiskDB returns the underlying key-value disk database.
+func (db *cachingDB) DiskDB() ethdb.KeyValueStore {
+	return db.disk
+}
+
 // TrieDB retrieves any intermediate trie-node caching layer.
 func (db *cachingDB) TrieDB() *trie.Database {
-	return db.db
+	return db.triedb
 }

core/state/dump.go

@@ -29,7 +29,7 @@ import (
 	"github.com/ethereum/go-ethereum/trie"
 )
 
-// DumpConfig is a set of options to control what portions of the statewill be
+// DumpConfig is a set of options to control what portions of the state will be
 // iterated and collected.
 type DumpConfig struct {
 	SkipCode          bool
@@ -168,7 +168,12 @@ func (s *StateDB) DumpToCollector(c DumpCollector, conf *DumpConfig) (nextKey []
 		}
 		if !conf.SkipStorage {
 			account.Storage = make(map[common.Hash]string)
-			storageIt := trie.NewIterator(obj.getTrie(s.db).NodeIterator(nil))
+			tr, err := obj.getTrie(s.db)
+			if err != nil {
+				log.Error("Failed to load storage trie", "err", err)
+				continue
+			}
+			storageIt := trie.NewIterator(tr.NodeIterator(nil))
 			for storageIt.Next() {
 				_, content, _, err := rlp.Split(storageIt.Value)
 				if err != nil {

core/state/iterator.go

@@ -109,7 +109,7 @@ func (it *NodeIterator) step() error {
 	if err := rlp.Decode(bytes.NewReader(it.stateIt.LeafBlob()), &account); err != nil {
 		return err
 	}
-	dataTrie, err := it.state.db.OpenStorageTrie(common.BytesToHash(it.stateIt.LeafKey()), account.Root)
+	dataTrie, err := it.state.db.OpenStorageTrie(it.state.originalRoot, common.BytesToHash(it.stateIt.LeafKey()), account.Root)
 	if err != nil {
 		return err
 	}

core/state/iterator_test.go

@@ -17,20 +17,19 @@
 package state
 
 import (
-	"bytes"
 	"testing"
 
 	"github.com/ethereum/go-ethereum/common"
-	"github.com/ethereum/go-ethereum/ethdb"
+	"github.com/ethereum/go-ethereum/core/rawdb"
 )
 
 // Tests that the node iterator indeed walks over the entire database contents.
 func TestNodeIteratorCoverage(t *testing.T) {
 	// Create some arbitrary test state to iterate
-	db, root, _ := makeTestState()
-	db.TrieDB().Commit(root, false, nil)
+	db, sdb, root, _ := makeTestState()
+	sdb.TrieDB().Commit(root, false)
 
-	state, err := New(root, db, nil)
+	state, err := New(root, sdb, nil)
 	if err != nil {
 		t.Fatalf("failed to create state trie at %x: %v", root, err)
 	}
@@ -41,29 +40,54 @@ func TestNodeIteratorCoverage(t *testing.T) {
 			hashes[it.Hash] = struct{}{}
 		}
 	}
+	// Check in-disk nodes
+	var (
+		seenNodes = make(map[common.Hash]struct{})
+		seenCodes = make(map[common.Hash]struct{})
+	)
+	it := db.NewIterator(nil, nil)
+	for it.Next() {
+		ok, hash := isTrieNode(sdb.TrieDB().Scheme(), it.Key(), it.Value())
+		if !ok {
+			continue
+		}
+		seenNodes[hash] = struct{}{}
+	}
+	it.Release()
+
+	// Check in-disk codes
+	it = db.NewIterator(nil, nil)
+	for it.Next() {
+		ok, hash := rawdb.IsCodeKey(it.Key())
+		if !ok {
+			continue
+		}
+		if _, ok := hashes[common.BytesToHash(hash)]; !ok {
+			t.Errorf("state entry not reported %x", it.Key())
+		}
+		seenCodes[common.BytesToHash(hash)] = struct{}{}
+	}
+	it.Release()
+
 	// Cross check the iterated hashes and the database/nodepool content
 	for hash := range hashes {
-		if _, err = db.TrieDB().Node(hash); err != nil {
-			_, err = db.ContractCode(common.Hash{}, hash)
+		_, ok := seenNodes[hash]
+		if !ok {
+			_, ok = seenCodes[hash]
 		}
-		if err != nil {
+		if !ok {
 			t.Errorf("failed to retrieve reported node %x", hash)
 		}
 	}
-	for _, hash := range db.TrieDB().Nodes() {
-		if _, ok := hashes[hash]; !ok {
-			t.Errorf("state entry not reported %x", hash)
+}
+
+// isTrieNode is a helper function which reports if the provided
+// database entry belongs to a trie node or not.
+func isTrieNode(scheme string, key, val []byte) (bool, common.Hash) {
+	if scheme == rawdb.HashScheme {
+		if len(key) == common.HashLength {
+			return true, common.BytesToHash(key)
 		}
 	}
-	it := db.TrieDB().DiskDB().(ethdb.Database).NewIterator(nil, nil)
-	for it.Next() {
-		key := it.Key()
-		if bytes.HasPrefix(key, []byte("secure-key-")) {
-			continue
-		}
-		if _, ok := hashes[common.BytesToHash(key)]; !ok {
-			t.Errorf("state entry not reported %x", key)
-		}
-	}
-	it.Release()
+	return false, common.Hash{}
 }

core/state/journal.go

@@ -138,6 +138,11 @@ type (
 		address *common.Address
 		slot    *common.Hash
 	}
+
+	transientStorageChange struct {
+		account       *common.Address
+		key, prevalue common.Hash
+	}
 )
 
 func (ch createObjectChange) revert(s *StateDB) {
@@ -151,8 +156,8 @@ func (ch createObjectChange) dirtied() *common.Address {
 
 func (ch resetObjectChange) revert(s *StateDB) {
 	s.setStateObject(ch.prev)
-	if !ch.prevdestruct && s.snap != nil {
-		delete(s.snapDestructs, ch.prev.addrHash)
+	if !ch.prevdestruct {
+		delete(s.stateObjectsDestruct, ch.prev.address)
 	}
 }
 
@@ -213,6 +218,14 @@ func (ch storageChange) dirtied() *common.Address {
 	return ch.account
 }
 
+func (ch transientStorageChange) revert(s *StateDB) {
+	s.setTransientState(*ch.account, ch.key, ch.prevalue)
+}
+
+func (ch transientStorageChange) dirtied() *common.Address {
+	return nil
+}
+
 func (ch refundChange) revert(s *StateDB) {
 	s.refund = ch.prev
 }

core/state/metrics.go

@@ -19,10 +19,12 @@ package state
 import "github.com/ethereum/go-ethereum/metrics"
 
 var (
-	accountUpdatedMeter        = metrics.NewRegisteredMeter("state/update/account", nil)
-	storageUpdatedMeter        = metrics.NewRegisteredMeter("state/update/storage", nil)
-	accountDeletedMeter        = metrics.NewRegisteredMeter("state/delete/account", nil)
-	storageDeletedMeter        = metrics.NewRegisteredMeter("state/delete/storage", nil)
-	accountTrieCommittedMeter  = metrics.NewRegisteredMeter("state/commit/accountnodes", nil)
-	storageTriesCommittedMeter = metrics.NewRegisteredMeter("state/commit/storagenodes", nil)
+	accountUpdatedMeter      = metrics.NewRegisteredMeter("state/update/account", nil)
+	storageUpdatedMeter      = metrics.NewRegisteredMeter("state/update/storage", nil)
+	accountDeletedMeter      = metrics.NewRegisteredMeter("state/delete/account", nil)
+	storageDeletedMeter      = metrics.NewRegisteredMeter("state/delete/storage", nil)
+	accountTrieUpdatedMeter  = metrics.NewRegisteredMeter("state/update/accountnodes", nil)
+	storageTriesUpdatedMeter = metrics.NewRegisteredMeter("state/update/storagenodes", nil)
+	accountTrieDeletedMeter  = metrics.NewRegisteredMeter("state/delete/accountnodes", nil)
+	storageTriesDeletedMeter = metrics.NewRegisteredMeter("state/delete/storagenodes", nil)
 )

core/state/pruner/pruner.go

@@ -63,52 +63,63 @@ var (
 	emptyCode = crypto.Keccak256(nil)
 )
 
+// Config includes all the configurations for pruning.
+type Config struct {
+	Datadir   string // The directory of the state database
+	Cachedir  string // The directory of state clean cache
+	BloomSize uint64 // The Megabytes of memory allocated to bloom-filter
+}
+
 // Pruner is an offline tool to prune the stale state with the
 // help of the snapshot. The workflow of pruner is very simple:
 //
-// - iterate the snapshot, reconstruct the relevant state
-// - iterate the database, delete all other state entries which
-//   don't belong to the target state and the genesis state
+//   - iterate the snapshot, reconstruct the relevant state
+//   - iterate the database, delete all other state entries which
+//     don't belong to the target state and the genesis state
 //
 // It can take several hours(around 2 hours for mainnet) to finish
 // the whole pruning work. It's recommended to run this offline tool
 // periodically in order to release the disk usage and improve the
 // disk read performance to some extent.
 type Pruner struct {
-	db            ethdb.Database
-	stateBloom    *stateBloom
-	datadir       string
-	trieCachePath string
-	headHeader    *types.Header
-	snaptree      *snapshot.Tree
+	config      Config
+	chainHeader *types.Header
+	db          ethdb.Database
+	stateBloom  *stateBloom
+	snaptree    *snapshot.Tree
 }
 
 // NewPruner creates the pruner instance.
-func NewPruner(db ethdb.Database, datadir, trieCachePath string, bloomSize uint64) (*Pruner, error) {
+func NewPruner(db ethdb.Database, config Config) (*Pruner, error) {
 	headBlock := rawdb.ReadHeadBlock(db)
 	if headBlock == nil {
-		return nil, errors.New("Failed to load head block")
+		return nil, errors.New("failed to load head block")
 	}
-	snaptree, err := snapshot.New(db, trie.NewDatabase(db), 256, headBlock.Root(), false, false, false)
+	snapconfig := snapshot.Config{
+		CacheSize:  256,
+		Recovery:   false,
+		NoBuild:    true,
+		AsyncBuild: false,
+	}
+	snaptree, err := snapshot.New(snapconfig, db, trie.NewDatabase(db), headBlock.Root())
 	if err != nil {
 		return nil, err // The relevant snapshot(s) might not exist
 	}
 	// Sanitize the bloom filter size if it's too small.
-	if bloomSize < 256 {
-		log.Warn("Sanitizing bloomfilter size", "provided(MB)", bloomSize, "updated(MB)", 256)
-		bloomSize = 256
+	if config.BloomSize < 256 {
+		log.Warn("Sanitizing bloomfilter size", "provided(MB)", config.BloomSize, "updated(MB)", 256)
+		config.BloomSize = 256
 	}
-	stateBloom, err := newStateBloomWithSize(bloomSize)
+	stateBloom, err := newStateBloomWithSize(config.BloomSize)
 	if err != nil {
 		return nil, err
 	}
 	return &Pruner{
-		db:            db,
-		stateBloom:    stateBloom,
-		datadir:       datadir,
-		trieCachePath: trieCachePath,
-		headHeader:    headBlock.Header(),
-		snaptree:      snaptree,
+		config:      config,
+		chainHeader: headBlock.Header(),
+		db:          db,
+		stateBloom:  stateBloom,
+		snaptree:    snaptree,
 	}, nil
 }
 
@@ -236,12 +247,12 @@ func (p *Pruner) Prune(root common.Hash) error {
 	// reuse it for pruning instead of generating a new one. It's
 	// mandatory because a part of state may already be deleted,
 	// the recovery procedure is necessary.
-	_, stateBloomRoot, err := findBloomFilter(p.datadir)
+	_, stateBloomRoot, err := findBloomFilter(p.config.Datadir)
 	if err != nil {
 		return err
 	}
 	if stateBloomRoot != (common.Hash{}) {
-		return RecoverPruning(p.datadir, p.db, p.trieCachePath)
+		return RecoverPruning(p.config.Datadir, p.db, p.config.Cachedir)
 	}
 	// If the target state root is not specified, use the HEAD-127 as the
 	// target. The reason for picking it is:
@@ -252,7 +263,7 @@ func (p *Pruner) Prune(root common.Hash) error {
 		// Retrieve all snapshot layers from the current HEAD.
 		// In theory there are 128 difflayers + 1 disk layer present,
 		// so 128 diff layers are expected to be returned.
-		layers = p.snaptree.Snapshots(p.headHeader.Root, 128, true)
+		layers = p.snaptree.Snapshots(p.chainHeader.Root, 128, true)
 		if len(layers) != 128 {
 			// Reject if the accumulated diff layers are less than 128. It
 			// means in most of normal cases, there is no associated state
@@ -265,7 +276,7 @@ func (p *Pruner) Prune(root common.Hash) error {
 	// Ensure the root is really present. The weak assumption
 	// is the presence of root can indicate the presence of the
 	// entire trie.
-	if !rawdb.HasTrieNode(p.db, root) {
+	if !rawdb.HasLegacyTrieNode(p.db, root) {
 		// The special case is for clique based networks(rinkeby, goerli
 		// and some other private networks), it's possible that two
 		// consecutive blocks will have same root. In this case snapshot
@@ -279,7 +290,7 @@ func (p *Pruner) Prune(root common.Hash) error {
 		// as the pruning target.
 		var found bool
 		for i := len(layers) - 2; i >= 2; i-- {
-			if rawdb.HasTrieNode(p.db, layers[i].Root()) {
+			if rawdb.HasLegacyTrieNode(p.db, layers[i].Root()) {
 				root = layers[i].Root()
 				found = true
 				log.Info("Selecting middle-layer as the pruning target", "root", root, "depth", i)
@@ -294,7 +305,7 @@ func (p *Pruner) Prune(root common.Hash) error {
 		}
 	} else {
 		if len(layers) > 0 {
-			log.Info("Selecting bottom-most difflayer as the pruning target", "root", root, "height", p.headHeader.Number.Uint64()-127)
+			log.Info("Selecting bottom-most difflayer as the pruning target", "root", root, "height", p.chainHeader.Number.Uint64()-127)
 		} else {
 			log.Info("Selecting user-specified state as the pruning target", "root", root)
 		}
@@ -303,7 +314,7 @@ func (p *Pruner) Prune(root common.Hash) error {
 	// It's necessary otherwise in the next restart we will hit the
 	// deleted state root in the "clean cache" so that the incomplete
 	// state is picked for usage.
-	deleteCleanTrieCache(p.trieCachePath)
+	deleteCleanTrieCache(p.config.Cachedir)
 
 	// All the state roots of the middle layer should be forcibly pruned,
 	// otherwise the dangling state will be left.
@@ -325,7 +336,7 @@ func (p *Pruner) Prune(root common.Hash) error {
 	if err := extractGenesis(p.db, p.stateBloom); err != nil {
 		return err
 	}
-	filterName := bloomFilterName(p.datadir, root)
+	filterName := bloomFilterName(p.config.Datadir, root)
 
 	log.Info("Writing state bloom to disk", "name", filterName)
 	if err := p.stateBloom.Commit(filterName, filterName+stateBloomFileTempSuffix); err != nil {
@@ -352,7 +363,7 @@ func RecoverPruning(datadir string, db ethdb.Database, trieCachePath string) err
 	}
 	headBlock := rawdb.ReadHeadBlock(db)
 	if headBlock == nil {
-		return errors.New("Failed to load head block")
+		return errors.New("failed to load head block")
 	}
 	// Initialize the snapshot tree in recovery mode to handle this special case:
 	// - Users run the `prune-state` command multiple times
@@ -362,7 +373,13 @@ func RecoverPruning(datadir string, db ethdb.Database, trieCachePath string) err
 	// - The state HEAD is rewound already because of multiple incomplete `prune-state`
 	// In this case, even the state HEAD is not exactly matched with snapshot, it
 	// still feasible to recover the pruning correctly.
-	snaptree, err := snapshot.New(db, trie.NewDatabase(db), 256, headBlock.Root(), false, false, true)
+	snapconfig := snapshot.Config{
+		CacheSize:  256,
+		Recovery:   true,
+		NoBuild:    true,
+		AsyncBuild: false,
+	}
+	snaptree, err := snapshot.New(snapconfig, db, trie.NewDatabase(db), headBlock.Root())
 	if err != nil {
 		return err // The relevant snapshot(s) might not exist
 	}
@@ -410,7 +427,7 @@ func extractGenesis(db ethdb.Database, stateBloom *stateBloom) error {
 	if genesis == nil {
 		return errors.New("missing genesis block")
 	}
-	t, err := trie.NewStateTrie(common.Hash{}, genesis.Root(), trie.NewDatabase(db))
+	t, err := trie.NewStateTrie(trie.StateTrieID(genesis.Root()), trie.NewDatabase(db))
 	if err != nil {
 		return err
 	}
@@ -430,7 +447,8 @@ func extractGenesis(db ethdb.Database, stateBloom *stateBloom) error {
 				return err
 			}
 			if acc.Root != emptyRoot {
-				storageTrie, err := trie.NewStateTrie(common.BytesToHash(accIter.LeafKey()), acc.Root, trie.NewDatabase(db))
+				id := trie.StorageTrieID(genesis.Root(), common.BytesToHash(accIter.LeafKey()), acc.Root)
+				storageTrie, err := trie.NewStateTrie(id, trie.NewDatabase(db))
 				if err != nil {
 					return err
 				}

core/state/snapshot/conversion.go

@@ -43,7 +43,7 @@ type trieKV struct {
 type (
 	// trieGeneratorFn is the interface of trie generation which can
 	// be implemented by different trie algorithm.
-	trieGeneratorFn func(db ethdb.KeyValueWriter, owner common.Hash, in chan (trieKV), out chan (common.Hash))
+	trieGeneratorFn func(db ethdb.KeyValueWriter, scheme string, owner common.Hash, in chan (trieKV), out chan (common.Hash))
 
 	// leafCallbackFn is the callback invoked at the leaves of the trie,
 	// returns the subtrie root with the specified subtrie identifier.
@@ -52,12 +52,12 @@ type (
 
 // GenerateAccountTrieRoot takes an account iterator and reproduces the root hash.
 func GenerateAccountTrieRoot(it AccountIterator) (common.Hash, error) {
-	return generateTrieRoot(nil, it, common.Hash{}, stackTrieGenerate, nil, newGenerateStats(), true)
+	return generateTrieRoot(nil, "", it, common.Hash{}, stackTrieGenerate, nil, newGenerateStats(), true)
 }
 
 // GenerateStorageTrieRoot takes a storage iterator and reproduces the root hash.
 func GenerateStorageTrieRoot(account common.Hash, it StorageIterator) (common.Hash, error) {
-	return generateTrieRoot(nil, it, account, stackTrieGenerate, nil, newGenerateStats(), true)
+	return generateTrieRoot(nil, "", it, account, stackTrieGenerate, nil, newGenerateStats(), true)
 }
 
 // GenerateTrie takes the whole snapshot tree as the input, traverses all the
@@ -71,7 +71,8 @@ func GenerateTrie(snaptree *Tree, root common.Hash, src ethdb.Database, dst ethd
 	}
 	defer acctIt.Release()
 
-	got, err := generateTrieRoot(dst, acctIt, common.Hash{}, stackTrieGenerate, func(dst ethdb.KeyValueWriter, accountHash, codeHash common.Hash, stat *generateStats) (common.Hash, error) {
+	scheme := snaptree.triedb.Scheme()
+	got, err := generateTrieRoot(dst, scheme, acctIt, common.Hash{}, stackTrieGenerate, func(dst ethdb.KeyValueWriter, accountHash, codeHash common.Hash, stat *generateStats) (common.Hash, error) {
 		// Migrate the code first, commit the contract code into the tmp db.
 		if codeHash != emptyCode {
 			code := rawdb.ReadCode(src, codeHash)
@@ -87,7 +88,7 @@ func GenerateTrie(snaptree *Tree, root common.Hash, src ethdb.Database, dst ethd
 		}
 		defer storageIt.Release()
 
-		hash, err := generateTrieRoot(dst, storageIt, accountHash, stackTrieGenerate, nil, stat, false)
+		hash, err := generateTrieRoot(dst, scheme, storageIt, accountHash, stackTrieGenerate, nil, stat, false)
 		if err != nil {
 			return common.Hash{}, err
 		}
@@ -136,7 +137,7 @@ func (stat *generateStats) progressAccounts(account common.Hash, done uint64) {
 	stat.head = account
 }
 
-// finishAccounts updates the gemerator stats for the finished account range.
+// finishAccounts updates the generator stats for the finished account range.
 func (stat *generateStats) finishAccounts(done uint64) {
 	stat.lock.Lock()
 	defer stat.lock.Unlock()
@@ -242,7 +243,7 @@ func runReport(stats *generateStats, stop chan bool) {
 // generateTrieRoot generates the trie hash based on the snapshot iterator.
 // It can be used for generating account trie, storage trie or even the
 // whole state which connects the accounts and the corresponding storages.
-func generateTrieRoot(db ethdb.KeyValueWriter, it Iterator, account common.Hash, generatorFn trieGeneratorFn, leafCallback leafCallbackFn, stats *generateStats, report bool) (common.Hash, error) {
+func generateTrieRoot(db ethdb.KeyValueWriter, scheme string, it Iterator, account common.Hash, generatorFn trieGeneratorFn, leafCallback leafCallbackFn, stats *generateStats, report bool) (common.Hash, error) {
 	var (
 		in      = make(chan trieKV)         // chan to pass leaves
 		out     = make(chan common.Hash, 1) // chan to collect result
@@ -253,7 +254,7 @@ func generateTrieRoot(db ethdb.KeyValueWriter, it Iterator, account common.Hash,
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
-		generatorFn(db, account, in, out)
+		generatorFn(db, scheme, account, in, out)
 	}()
 	// Spin up a go-routine for progress logging
 	if report && stats != nil {
@@ -360,8 +361,14 @@ func generateTrieRoot(db ethdb.KeyValueWriter, it Iterator, account common.Hash,
 	return stop(nil)
 }
 
-func stackTrieGenerate(db ethdb.KeyValueWriter, owner common.Hash, in chan trieKV, out chan common.Hash) {
-	t := trie.NewStackTrieWithOwner(db, owner)
+func stackTrieGenerate(db ethdb.KeyValueWriter, scheme string, owner common.Hash, in chan trieKV, out chan common.Hash) {
+	var nodeWriter trie.NodeWriteFunc
+	if db != nil {
+		nodeWriter = func(owner common.Hash, path []byte, hash common.Hash, blob []byte) {
+			rawdb.WriteTrieNode(db, owner, path, hash, blob, scheme)
+		}
+	}
+	t := trie.NewStackTrieWithOwner(nodeWriter, owner)
 	for leaf := range in {
 		t.TryUpdate(leaf.key[:], leaf.value)
 	}

core/state/snapshot/difflayer.go

@@ -68,7 +68,7 @@ var (
 	bloomFuncs = math.Round((bloomSize / float64(aggregatorItemLimit)) * math.Log(2))
 
 	// the bloom offsets are runtime constants which determines which part of the
-	// the account/storage hash the hasher functions looks at, to determine the
+	// account/storage hash the hasher functions looks at, to determine the
 	// bloom key for an account/slot. This is randomized at init(), so that the
 	// global population of nodes do not all display the exact same behaviour with
 	// regards to bloom content

core/state/snapshot/difflayer_test.go

@@ -18,6 +18,7 @@ package snapshot
 
 import (
 	"bytes"
+	crand "crypto/rand"
 	"math/rand"
 	"testing"
 
@@ -73,7 +74,7 @@ func TestMergeBasics(t *testing.T) {
 		if rand.Intn(2) == 0 {
 			accStorage := make(map[common.Hash][]byte)
 			value := make([]byte, 32)
-			rand.Read(value)
+			crand.Read(value)
 			accStorage[randomHash()] = value
 			storage[h] = accStorage
 		}
@@ -294,7 +295,7 @@ func BenchmarkSearchSlot(b *testing.B) {
 		accStorage := make(map[common.Hash][]byte)
 		for i := 0; i < 5; i++ {
 			value := make([]byte, 32)
-			rand.Read(value)
+			crand.Read(value)
 			accStorage[randomHash()] = value
 			storage[accountKey] = accStorage
 		}
@@ -330,7 +331,7 @@ func BenchmarkFlatten(b *testing.B) {
 			accStorage := make(map[common.Hash][]byte)
 			for i := 0; i < 20; i++ {
 				value := make([]byte, 32)
-				rand.Read(value)
+				crand.Read(value)
 				accStorage[randomHash()] = value
 			}
 			storage[accountKey] = accStorage
@@ -379,7 +380,7 @@ func BenchmarkJournal(b *testing.B) {
 			accStorage := make(map[common.Hash][]byte)
 			for i := 0; i < 200; i++ {
 				value := make([]byte, 32)
-				rand.Read(value)
+				crand.Read(value)
 				accStorage[randomHash()] = value
 			}
 			storage[accountKey] = accStorage

core/state/snapshot/generate.go

@@ -29,7 +29,6 @@ import (
 	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/ethdb"
-	"github.com/ethereum/go-ethereum/ethdb/memorydb"
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
@@ -166,7 +165,7 @@ func (result *proofResult) forEach(callback func(key []byte, val []byte) error)
 //
 // The proof result will be returned if the range proving is finished, otherwise
 // the error will be returned to abort the entire procedure.
-func (dl *diskLayer) proveRange(ctx *generatorContext, owner common.Hash, root common.Hash, prefix []byte, kind string, origin []byte, max int, valueConvertFn func([]byte) ([]byte, error)) (*proofResult, error) {
+func (dl *diskLayer) proveRange(ctx *generatorContext, trieId *trie.ID, prefix []byte, kind string, origin []byte, max int, valueConvertFn func([]byte) ([]byte, error)) (*proofResult, error) {
 	var (
 		keys     [][]byte
 		vals     [][]byte
@@ -233,8 +232,9 @@ func (dl *diskLayer) proveRange(ctx *generatorContext, owner common.Hash, root c
 	}(time.Now())
 
 	// The snap state is exhausted, pass the entire key/val set for verification
+	root := trieId.Root
 	if origin == nil && !diskMore {
-		stackTr := trie.NewStackTrieWithOwner(nil, owner)
+		stackTr := trie.NewStackTrie(nil)
 		for i, key := range keys {
 			stackTr.TryUpdate(key, vals[i])
 		}
@@ -248,7 +248,7 @@ func (dl *diskLayer) proveRange(ctx *generatorContext, owner common.Hash, root c
 		return &proofResult{keys: keys, vals: vals}, nil
 	}
 	// Snap state is chunked, generate edge proofs for verification.
-	tr, err := trie.New(owner, root, dl.triedb)
+	tr, err := trie.New(trieId, dl.triedb)
 	if err != nil {
 		ctx.stats.Log("Trie missing, state snapshotting paused", dl.root, dl.genMarker)
 		return nil, errMissingTrie
@@ -313,9 +313,9 @@ type onStateCallback func(key []byte, val []byte, write bool, delete bool) error
 // generateRange generates the state segment with particular prefix. Generation can
 // either verify the correctness of existing state through range-proof and skip
 // generation, or iterate trie to regenerate state on demand.
-func (dl *diskLayer) generateRange(ctx *generatorContext, owner common.Hash, root common.Hash, prefix []byte, kind string, origin []byte, max int, onState onStateCallback, valueConvertFn func([]byte) ([]byte, error)) (bool, []byte, error) {
+func (dl *diskLayer) generateRange(ctx *generatorContext, trieId *trie.ID, prefix []byte, kind string, origin []byte, max int, onState onStateCallback, valueConvertFn func([]byte) ([]byte, error)) (bool, []byte, error) {
 	// Use range prover to check the validity of the flat state in the range
-	result, err := dl.proveRange(ctx, owner, root, prefix, kind, origin, max, valueConvertFn)
+	result, err := dl.proveRange(ctx, trieId, prefix, kind, origin, max, valueConvertFn)
 	if err != nil {
 		return false, nil, err
 	}
@@ -359,25 +359,28 @@ func (dl *diskLayer) generateRange(ctx *generatorContext, owner common.Hash, roo
 	}
 	// We use the snap data to build up a cache which can be used by the
 	// main account trie as a primary lookup when resolving hashes
-	var snapNodeCache ethdb.KeyValueStore
+	var resolver trie.NodeResolver
 	if len(result.keys) > 0 {
-		snapNodeCache = memorydb.New()
-		snapTrieDb := trie.NewDatabase(snapNodeCache)
-		snapTrie, _ := trie.New(owner, common.Hash{}, snapTrieDb)
+		mdb := rawdb.NewMemoryDatabase()
+		tdb := trie.NewDatabase(mdb)
+		snapTrie := trie.NewEmpty(tdb)
 		for i, key := range result.keys {
 			snapTrie.Update(key, result.vals[i])
 		}
-		root, nodes, _ := snapTrie.Commit(false)
+		root, nodes := snapTrie.Commit(false)
 		if nodes != nil {
-			snapTrieDb.Update(trie.NewWithNodeSet(nodes))
+			tdb.Update(trie.NewWithNodeSet(nodes))
+			tdb.Commit(root, false)
+		}
+		resolver = func(owner common.Hash, path []byte, hash common.Hash) []byte {
+			return rawdb.ReadTrieNode(mdb, owner, path, hash, tdb.Scheme())
 		}
-		snapTrieDb.Commit(root, false, nil)
 	}
 	// Construct the trie for state iteration, reuse the trie
 	// if it's already opened with some nodes resolved.
 	tr := result.tr
 	if tr == nil {
-		tr, err = trie.New(owner, root, dl.triedb)
+		tr, err = trie.New(trieId, dl.triedb)
 		if err != nil {
 			ctx.stats.Log("Trie missing, state snapshotting paused", dl.root, dl.genMarker)
 			return false, nil, errMissingTrie
@@ -400,7 +403,7 @@ func (dl *diskLayer) generateRange(ctx *generatorContext, owner common.Hash, roo
 		start    = time.Now()
 		internal time.Duration
 	)
-	nodeIt.AddResolver(snapNodeCache)
+	nodeIt.AddResolver(resolver)
 
 	for iter.Next() {
 		if last != nil && bytes.Compare(iter.Key, last) > 0 {
@@ -460,7 +463,7 @@ func (dl *diskLayer) generateRange(ctx *generatorContext, owner common.Hash, roo
 	} else {
 		snapAccountTrieReadCounter.Inc((time.Since(start) - internal).Nanoseconds())
 	}
-	logger.Debug("Regenerated state range", "root", root, "last", hexutil.Encode(last),
+	logger.Debug("Regenerated state range", "root", trieId.Root, "last", hexutil.Encode(last),
 		"count", count, "created", created, "updated", updated, "untouched", untouched, "deleted", deleted)
 
 	// If there are either more trie items, or there are more snap items
@@ -511,7 +514,7 @@ func (dl *diskLayer) checkAndFlush(ctx *generatorContext, current []byte) error
 
 // generateStorages generates the missing storage slots of the specific contract.
 // It's supposed to restart the generation from the given origin position.
-func generateStorages(ctx *generatorContext, dl *diskLayer, account common.Hash, storageRoot common.Hash, storeMarker []byte) error {
+func generateStorages(ctx *generatorContext, dl *diskLayer, stateRoot common.Hash, account common.Hash, storageRoot common.Hash, storeMarker []byte) error {
 	onStorage := func(key []byte, val []byte, write bool, delete bool) error {
 		defer func(start time.Time) {
 			snapStorageWriteCounter.Inc(time.Since(start).Nanoseconds())
@@ -540,7 +543,8 @@ func generateStorages(ctx *generatorContext, dl *diskLayer, account common.Hash,
 	// Loop for re-generating the missing storage slots.
 	var origin = common.CopyBytes(storeMarker)
 	for {
-		exhausted, last, err := dl.generateRange(ctx, account, storageRoot, append(rawdb.SnapshotStoragePrefix, account.Bytes()...), snapStorage, origin, storageCheckRange, onStorage, nil)
+		id := trie.StorageTrieID(stateRoot, account, storageRoot)
+		exhausted, last, err := dl.generateRange(ctx, id, append(rawdb.SnapshotStoragePrefix, account.Bytes()...), snapStorage, origin, storageCheckRange, onStorage, nil)
 		if err != nil {
 			return err // The procedure it aborted, either by external signal or internal error.
 		}
@@ -624,7 +628,7 @@ func generateAccounts(ctx *generatorContext, dl *diskLayer, accMarker []byte) er
 			if accMarker != nil && bytes.Equal(account[:], accMarker) && len(dl.genMarker) > common.HashLength {
 				storeMarker = dl.genMarker[common.HashLength:]
 			}
-			if err := generateStorages(ctx, dl, account, acc.Root, storeMarker); err != nil {
+			if err := generateStorages(ctx, dl, dl.root, account, acc.Root, storeMarker); err != nil {
 				return err
 			}
 		}
@@ -640,7 +644,8 @@ func generateAccounts(ctx *generatorContext, dl *diskLayer, accMarker []byte) er
 	}
 	origin := common.CopyBytes(accMarker)
 	for {
-		exhausted, last, err := dl.generateRange(ctx, common.Hash{}, dl.root, rawdb.SnapshotAccountPrefix, snapAccount, origin, accountRange, onAccount, FullAccountRLP)
+		id := trie.StateTrieID(dl.root)
+		exhausted, last, err := dl.generateRange(ctx, id, rawdb.SnapshotAccountPrefix, snapAccount, origin, accountRange, onAccount, FullAccountRLP)
 		if err != nil {
 			return err // The procedure it aborted, either by external signal or internal error.
 		}

core/state/snapshot/generate_test.go

@@ -117,12 +117,12 @@ func checkSnapRoot(t *testing.T, snap *diskLayer, trieRoot common.Hash) {
 	accIt := snap.AccountIterator(common.Hash{})
 	defer accIt.Release()
 
-	snapRoot, err := generateTrieRoot(nil, accIt, common.Hash{}, stackTrieGenerate,
+	snapRoot, err := generateTrieRoot(nil, "", accIt, common.Hash{}, stackTrieGenerate,
 		func(db ethdb.KeyValueWriter, accountHash, codeHash common.Hash, stat *generateStats) (common.Hash, error) {
 			storageIt, _ := snap.StorageIterator(accountHash, common.Hash{})
 			defer storageIt.Release()
 
-			hash, err := generateTrieRoot(nil, storageIt, accountHash, stackTrieGenerate, nil, stat, false)
+			hash, err := generateTrieRoot(nil, "", storageIt, accountHash, stackTrieGenerate, nil, stat, false)
 			if err != nil {
 				return common.Hash{}, err
 			}
@@ -149,7 +149,7 @@ type testHelper struct {
 func newHelper() *testHelper {
 	diskdb := rawdb.NewMemoryDatabase()
 	triedb := trie.NewDatabase(diskdb)
-	accTrie, _ := trie.NewStateTrie(common.Hash{}, common.Hash{}, triedb)
+	accTrie, _ := trie.NewStateTrie(trie.StateTrieID(common.Hash{}), triedb)
 	return &testHelper{
 		diskdb:  diskdb,
 		triedb:  triedb,
@@ -182,14 +182,15 @@ func (t *testHelper) addSnapStorage(accKey string, keys []string, vals []string)
 }
 
 func (t *testHelper) makeStorageTrie(stateRoot, owner common.Hash, keys []string, vals []string, commit bool) []byte {
-	stTrie, _ := trie.NewStateTrie(owner, common.Hash{}, t.triedb)
+	id := trie.StorageTrieID(stateRoot, owner, common.Hash{})
+	stTrie, _ := trie.NewStateTrie(id, t.triedb)
 	for i, k := range keys {
 		stTrie.Update([]byte(k), []byte(vals[i]))
 	}
 	if !commit {
 		return stTrie.Hash().Bytes()
 	}
-	root, nodes, _ := stTrie.Commit(false)
+	root, nodes := stTrie.Commit(false)
 	if nodes != nil {
 		t.nodes.Merge(nodes)
 	}
@@ -197,12 +198,12 @@ func (t *testHelper) makeStorageTrie(stateRoot, owner common.Hash, keys []string
 }
 
 func (t *testHelper) Commit() common.Hash {
-	root, nodes, _ := t.accTrie.Commit(true)
+	root, nodes := t.accTrie.Commit(true)
 	if nodes != nil {
 		t.nodes.Merge(nodes)
 	}
 	t.triedb.Update(t.nodes)
-	t.triedb.Commit(root, false, nil)
+	t.triedb.Commit(root, false)
 	return root
 }
 
@@ -220,10 +221,12 @@ func (t *testHelper) CommitAndGenerate() (common.Hash, *diskLayer) {
 //   - miss in the beginning
 //   - miss in the middle
 //   - miss in the end
+//
 // - the contract(non-empty storage) has wrong storage slots
 //   - wrong slots in the beginning
 //   - wrong slots in the middle
 //   - wrong slots in the end
+//
 // - the contract(non-empty storage) has extra storage slots
 //   - extra slots in the beginning
 //   - extra slots in the middle
@@ -388,7 +391,7 @@ func TestGenerateCorruptAccountTrie(t *testing.T) {
 	root := helper.Commit() // Root: 0xa04693ea110a31037fb5ee814308a6f1d76bdab0b11676bdf4541d2de55ba978
 
 	// Delete an account trie leaf and ensure the generator chokes
-	helper.triedb.Commit(root, false, nil)
+	helper.triedb.Commit(root, false)
 	helper.diskdb.Delete(common.HexToHash("0x65145f923027566669a1ae5ccac66f945b55ff6eaeb17d2ea8e048b7d381f2d7").Bytes())
 
 	snap := generateSnapshot(helper.diskdb, helper.triedb, 16, root)
@@ -491,12 +494,12 @@ func TestGenerateWithExtraAccounts(t *testing.T) {
 
 		// Identical in the snap
 		key := hashData([]byte("acc-1"))
-		rawdb.WriteAccountSnapshot(helper.triedb.DiskDB(), key, val)
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("key-1")), []byte("val-1"))
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("key-2")), []byte("val-2"))
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("key-3")), []byte("val-3"))
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("key-4")), []byte("val-4"))
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("key-5")), []byte("val-5"))
+		rawdb.WriteAccountSnapshot(helper.diskdb, key, val)
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("key-1")), []byte("val-1"))
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("key-2")), []byte("val-2"))
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("key-3")), []byte("val-3"))
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("key-4")), []byte("val-4"))
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("key-5")), []byte("val-5"))
 	}
 	{
 		// Account two exists only in the snapshot
@@ -508,15 +511,15 @@ func TestGenerateWithExtraAccounts(t *testing.T) {
 		acc := &Account{Balance: big.NewInt(1), Root: stRoot, CodeHash: emptyCode.Bytes()}
 		val, _ := rlp.EncodeToBytes(acc)
 		key := hashData([]byte("acc-2"))
-		rawdb.WriteAccountSnapshot(helper.triedb.DiskDB(), key, val)
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("b-key-1")), []byte("b-val-1"))
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("b-key-2")), []byte("b-val-2"))
-		rawdb.WriteStorageSnapshot(helper.triedb.DiskDB(), key, hashData([]byte("b-key-3")), []byte("b-val-3"))
+		rawdb.WriteAccountSnapshot(helper.diskdb, key, val)
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("b-key-1")), []byte("b-val-1"))
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("b-key-2")), []byte("b-val-2"))
+		rawdb.WriteStorageSnapshot(helper.diskdb, key, hashData([]byte("b-key-3")), []byte("b-val-3"))
 	}
 	root := helper.Commit()
 
 	// To verify the test: If we now inspect the snap db, there should exist extraneous storage items
-	if data := rawdb.ReadStorageSnapshot(helper.triedb.DiskDB(), hashData([]byte("acc-2")), hashData([]byte("b-key-1"))); data == nil {
+	if data := rawdb.ReadStorageSnapshot(helper.diskdb, hashData([]byte("acc-2")), hashData([]byte("b-key-1"))); data == nil {
 		t.Fatalf("expected snap storage to exist")
 	}
 	snap := generateSnapshot(helper.diskdb, helper.triedb, 16, root)
@@ -534,7 +537,7 @@ func TestGenerateWithExtraAccounts(t *testing.T) {
 	snap.genAbort <- stop
 	<-stop
 	// If we now inspect the snap db, there should exist no extraneous storage items
-	if data := rawdb.ReadStorageSnapshot(helper.triedb.DiskDB(), hashData([]byte("acc-2")), hashData([]byte("b-key-1"))); data != nil {
+	if data := rawdb.ReadStorageSnapshot(helper.diskdb, hashData([]byte("acc-2")), hashData([]byte("b-key-1"))); data != nil {
 		t.Fatalf("expected slot to be removed, got %v", string(data))
 	}
 }

core/state/snapshot/iterator_fast.go

@@ -276,7 +276,7 @@ func (fi *fastIterator) next(idx int) bool {
 			return false
 		}
 		// The elem we're placing it next to has the same value,
-		// so whichever winds up on n+1 will need further iteraton
+		// so whichever winds up on n+1 will need further iteration
 		clash = n + 1
 
 		return cur.priority < fi.iterators[n+1].priority

core/state/snapshot/iterator_test.go

@@ -18,6 +18,7 @@ package snapshot
 
 import (
 	"bytes"
+	crand "crypto/rand"
 	"encoding/binary"
 	"fmt"
 	"math/rand"
@@ -47,7 +48,7 @@ func TestAccountIteratorBasics(t *testing.T) {
 		if rand.Intn(2) == 0 {
 			accStorage := make(map[common.Hash][]byte)
 			value := make([]byte, 32)
-			rand.Read(value)
+			crand.Read(value)
 			accStorage[randomHash()] = value
 			storage[h] = accStorage
 		}
@@ -79,7 +80,7 @@ func TestStorageIteratorBasics(t *testing.T) {
 
 		var nilstorage int
 		for i := 0; i < 100; i++ {
-			rand.Read(value)
+			crand.Read(value)
 			if rand.Intn(2) == 0 {
 				accStorage[randomHash()] = common.CopyBytes(value)
 			} else {
@@ -819,7 +820,7 @@ func TestStorageIteratorDeletions(t *testing.T) {
 // only spit out 200 values eventually.
 //
 // The value-fetching benchmark is easy on the binary iterator, since it never has to reach
-// down at any depth for retrieving the values -- all are on the toppmost layer
+// down at any depth for retrieving the values -- all are on the topmost layer
 //
 // BenchmarkAccountIteratorTraversal/binary_iterator_keys-6         	    2239	    483674 ns/op
 // BenchmarkAccountIteratorTraversal/binary_iterator_values-6       	    2403	    501810 ns/op

core/state/snapshot/journal.go

@@ -120,7 +120,7 @@ func loadAndParseJournal(db ethdb.KeyValueStore, base *diskLayer) (snapshot, jou
 }
 
 // loadSnapshot loads a pre-existing state snapshot backed by a key-value store.
-func loadSnapshot(diskdb ethdb.KeyValueStore, triedb *trie.Database, cache int, root common.Hash, recovery bool) (snapshot, bool, error) {
+func loadSnapshot(diskdb ethdb.KeyValueStore, triedb *trie.Database, root common.Hash, cache int, recovery bool, noBuild bool) (snapshot, bool, error) {
 	// If snapshotting is disabled (initial sync in progress), don't do anything,
 	// wait for the chain to permit us to do something meaningful
 	if rawdb.ReadSnapshotDisabled(diskdb) {
@@ -140,7 +140,7 @@ func loadSnapshot(diskdb ethdb.KeyValueStore, triedb *trie.Database, cache int,
 	}
 	snapshot, generator, err := loadAndParseJournal(diskdb, base)
 	if err != nil {
-		log.Warn("Failed to load new-format journal", "error", err)
+		log.Warn("Failed to load journal", "error", err)
 		return nil, false, err
 	}
 	// Entire snapshot journal loaded, sanity check the head. If the loaded
@@ -164,13 +164,16 @@ func loadSnapshot(diskdb ethdb.KeyValueStore, triedb *trie.Database, cache int,
 		// disk layer.
 		log.Warn("Snapshot is not continuous with chain", "snaproot", head, "chainroot", root)
 	}
-	// Everything loaded correctly, resume any suspended operations
+	// Load the disk layer status from the generator if it's not complete
 	if !generator.Done {
-		// Whether or not wiping was in progress, load any generator progress too
 		base.genMarker = generator.Marker
 		if base.genMarker == nil {
 			base.genMarker = []byte{}
 		}
+	}
+	// Everything loaded correctly, resume any suspended operations
+	// if the background generation is allowed
+	if !generator.Done && !noBuild {
 		base.genPending = make(chan struct{})
 		base.genAbort = make(chan chan *generatorStats)
 

core/state/snapshot/metrics.go

@@ -36,7 +36,7 @@ var (
 	snapAccountProveCounter = metrics.NewRegisteredCounter("state/snapshot/generation/duration/account/prove", nil)
 	// snapAccountTrieReadCounter measures time spent on the account trie iteration
 	snapAccountTrieReadCounter = metrics.NewRegisteredCounter("state/snapshot/generation/duration/account/trieread", nil)
-	// snapAccountSnapReadCounter measues time spent on the snapshot account iteration
+	// snapAccountSnapReadCounter measures time spent on the snapshot account iteration
 	snapAccountSnapReadCounter = metrics.NewRegisteredCounter("state/snapshot/generation/duration/account/snapread", nil)
 	// snapAccountWriteCounter measures time spent on writing/updating/deleting accounts
 	snapAccountWriteCounter = metrics.NewRegisteredCounter("state/snapshot/generation/duration/account/write", nil)

core/state/snapshot/snapshot.go

@@ -148,6 +148,14 @@ type snapshot interface {
 	StorageIterator(account common.Hash, seek common.Hash) (StorageIterator, bool)
 }
 
+// Config includes the configurations for snapshots.
+type Config struct {
+	CacheSize  int  // Megabytes permitted to use for read caches
+	Recovery   bool // Indicator that the snapshots is in the recovery mode
+	NoBuild    bool // Indicator that the snapshots generation is disallowed
+	AsyncBuild bool // The snapshot generation is allowed to be constructed asynchronously
+}
+
 // Tree is an Ethereum state snapshot tree. It consists of one persistent base
 // layer backed by a key-value store, on top of which arbitrarily many in-memory
 // diff layers are topped. The memory diffs can form a tree with branching, but
@@ -158,9 +166,9 @@ type snapshot interface {
 // storage data to avoid expensive multi-level trie lookups; and to allow sorted,
 // cheap iteration of the account/storage tries for sync aid.
 type Tree struct {
+	config Config                   // Snapshots configurations
 	diskdb ethdb.KeyValueStore      // Persistent database to store the snapshot
 	triedb *trie.Database           // In-memory cache to access the trie through
-	cache  int                      // Megabytes permitted to use for read caches
 	layers map[common.Hash]snapshot // Collection of all known layers
 	lock   sync.RWMutex
 
@@ -179,30 +187,32 @@ type Tree struct {
 // If the memory layers in the journal do not match the disk layer (e.g. there is
 // a gap) or the journal is missing, there are two repair cases:
 //
-// - if the 'recovery' parameter is true, all memory diff-layers will be discarded.
-//   This case happens when the snapshot is 'ahead' of the state trie.
-// - otherwise, the entire snapshot is considered invalid and will be recreated on
-//   a background thread.
-func New(diskdb ethdb.KeyValueStore, triedb *trie.Database, cache int, root common.Hash, async bool, rebuild bool, recovery bool) (*Tree, error) {
+//   - if the 'recovery' parameter is true, memory diff-layers and the disk-layer
+//     will all be kept. This case happens when the snapshot is 'ahead' of the
+//     state trie.
+//   - otherwise, the entire snapshot is considered invalid and will be recreated on
+//     a background thread.
+func New(config Config, diskdb ethdb.KeyValueStore, triedb *trie.Database, root common.Hash) (*Tree, error) {
 	// Create a new, empty snapshot tree
 	snap := &Tree{
+		config: config,
 		diskdb: diskdb,
 		triedb: triedb,
-		cache:  cache,
 		layers: make(map[common.Hash]snapshot),
 	}
-	if !async {
-		defer snap.waitBuild()
-	}
 	// Attempt to load a previously persisted snapshot and rebuild one if failed
-	head, disabled, err := loadSnapshot(diskdb, triedb, cache, root, recovery)
+	head, disabled, err := loadSnapshot(diskdb, triedb, root, config.CacheSize, config.Recovery, config.NoBuild)
 	if disabled {
 		log.Warn("Snapshot maintenance disabled (syncing)")
 		return snap, nil
 	}
+	// Create the building waiter iff the background generation is allowed
+	if !config.NoBuild && !config.AsyncBuild {
+		defer snap.waitBuild()
+	}
 	if err != nil {
-		if rebuild {
-			log.Warn("Failed to load snapshot, regenerating", "err", err)
+		log.Warn("Failed to load snapshot", "err", err)
+		if !config.NoBuild {
 			snap.Rebuild(root)
 			return snap, nil
 		}
@@ -727,7 +737,7 @@ func (t *Tree) Rebuild(root common.Hash) {
 	// generator will run a wiper first if there's not one running right now.
 	log.Info("Rebuilding state snapshot")
 	t.layers = map[common.Hash]snapshot{
-		root: generateSnapshot(t.diskdb, t.triedb, t.cache, root),
+		root: generateSnapshot(t.diskdb, t.triedb, t.config.CacheSize, root),
 	}
 }
 
@@ -766,14 +776,14 @@ func (t *Tree) Verify(root common.Hash) error {
 	}
 	defer acctIt.Release()
 
-	got, err := generateTrieRoot(nil, acctIt, common.Hash{}, stackTrieGenerate, func(db ethdb.KeyValueWriter, accountHash, codeHash common.Hash, stat *generateStats) (common.Hash, error) {
+	got, err := generateTrieRoot(nil, "", acctIt, common.Hash{}, stackTrieGenerate, func(db ethdb.KeyValueWriter, accountHash, codeHash common.Hash, stat *generateStats) (common.Hash, error) {
 		storageIt, err := t.StorageIterator(root, accountHash, common.Hash{})
 		if err != nil {
 			return common.Hash{}, err
 		}
 		defer storageIt.Release()
 
-		hash, err := generateTrieRoot(nil, storageIt, accountHash, stackTrieGenerate, nil, stat, false)
+		hash, err := generateTrieRoot(nil, "", storageIt, accountHash, stackTrieGenerate, nil, stat, false)
 		if err != nil {
 			return common.Hash{}, err
 		}
@@ -835,7 +845,7 @@ func (t *Tree) generating() (bool, error) {
 	return layer.genMarker != nil, nil
 }
 
-// diskRoot is a external helper function to return the disk layer root.
+// DiskRoot is a external helper function to return the disk layer root.
 func (t *Tree) DiskRoot() common.Hash {
 	t.lock.Lock()
 	defer t.lock.Unlock()

core/state/snapshot/snapshot_test.go

@@ -17,6 +17,7 @@
 package snapshot
 
 import (
+	crand "crypto/rand"
 	"encoding/binary"
 	"fmt"
 	"math/big"
@@ -33,7 +34,7 @@ import (
 // randomHash generates a random blob of data and returns it as a hash.
 func randomHash() common.Hash {
 	var hash common.Hash
-	if n, err := rand.Read(hash[:]); n != common.HashLength || err != nil {
+	if n, err := crand.Read(hash[:]); n != common.HashLength || err != nil {
 		panic(err)
 	}
 	return hash
@@ -166,7 +167,7 @@ func TestDiskLayerExternalInvalidationPartialFlatten(t *testing.T) {
 	if err := snaps.Cap(common.HexToHash("0x03"), 1); err != nil {
 		t.Fatalf("failed to merge accumulator onto disk: %v", err)
 	}
-	// Since the base layer was modified, ensure that data retrievald on the external reference fail
+	// Since the base layer was modified, ensure that data retrievals on the external reference fail
 	if acc, err := ref.Account(common.HexToHash("0x01")); err != ErrSnapshotStale {
 		t.Errorf("stale reference returned account: %#x (err: %v)", acc, err)
 	}
@@ -224,7 +225,7 @@ func TestDiffLayerExternalInvalidationPartialFlatten(t *testing.T) {
 	if err := snaps.Cap(common.HexToHash("0x04"), 1); err != nil {
 		t.Fatalf("failed to flatten diff layer into accumulator: %v", err)
 	}
-	// Since the accumulator diff layer was modified, ensure that data retrievald on the external reference fail
+	// Since the accumulator diff layer was modified, ensure that data retrievals on the external reference fail
 	if acc, err := ref.Account(common.HexToHash("0x01")); err != ErrSnapshotStale {
 		t.Errorf("stale reference returned account: %#x (err: %v)", acc, err)
 	}

core/state/state_object.go

@@ -63,7 +63,7 @@ func (s Storage) Copy() Storage {
 // The usage pattern is as follows:
 // First you need to obtain a state object.
 // Account values can be accessed and modified through the object.
-// Finally, call CommitTrie to write the modified storage trie into a database.
+// Finally, call commitTrie to write the modified storage trie into a database.
 type stateObject struct {
 	address  common.Address
 	addrHash common.Hash // hash of ethereum address of the account
@@ -84,7 +84,6 @@ type stateObject struct {
 	originStorage  Storage // Storage cache of original entries to dedup rewrites, reset for every transaction
 	pendingStorage Storage // Storage entries that need to be flushed to disk, at the end of an entire block
 	dirtyStorage   Storage // Storage entries that have been modified in the current transaction execution
-	fakeStorage    Storage // Fake storage which constructed by caller for debugging purpose.
 
 	// Cache flags.
 	// When an object is marked suicided it will be delete from the trie
@@ -148,7 +147,10 @@ func (s *stateObject) touch() {
 	}
 }
 
-func (s *stateObject) getTrie(db Database) Trie {
+// getTrie returns the associated storage trie. The trie will be opened
+// if it's not loaded previously. An error will be returned if trie can't
+// be loaded.
+func (s *stateObject) getTrie(db Database) (Trie, error) {
 	if s.trie == nil {
 		// Try fetching from prefetcher first
 		// We don't prefetch empty tries
@@ -158,23 +160,18 @@ func (s *stateObject) getTrie(db Database) Trie {
 			s.trie = s.db.prefetcher.trie(s.addrHash, s.data.Root)
 		}
 		if s.trie == nil {
-			var err error
-			s.trie, err = db.OpenStorageTrie(s.addrHash, s.data.Root)
+			tr, err := db.OpenStorageTrie(s.db.originalRoot, s.addrHash, s.data.Root)
 			if err != nil {
-				s.trie, _ = db.OpenStorageTrie(s.addrHash, common.Hash{})
-				s.setError(fmt.Errorf("can't create storage trie: %v", err))
+				return nil, err
 			}
+			s.trie = tr
 		}
 	}
-	return s.trie
+	return s.trie, nil
 }
 
 // GetState retrieves a value from the account storage trie.
 func (s *stateObject) GetState(db Database, key common.Hash) common.Hash {
-	// If the fake storage is set, only lookup the state here(in the debugging mode)
-	if s.fakeStorage != nil {
-		return s.fakeStorage[key]
-	}
 	// If we have a dirty value for this state entry, return it
 	value, dirty := s.dirtyStorage[key]
 	if dirty {
@@ -186,10 +183,6 @@ func (s *stateObject) GetState(db Database, key common.Hash) common.Hash {
 
 // GetCommittedState retrieves a value from the committed account storage trie.
 func (s *stateObject) GetCommittedState(db Database, key common.Hash) common.Hash {
-	// If the fake storage is set, only lookup the state here(in the debugging mode)
-	if s.fakeStorage != nil {
-		return s.fakeStorage[key]
-	}
 	// If we have a pending write or clean cached, return that
 	if value, pending := s.pendingStorage[key]; pending {
 		return value
@@ -197,21 +190,21 @@ func (s *stateObject) GetCommittedState(db Database, key common.Hash) common.Has
 	if value, cached := s.originStorage[key]; cached {
 		return value
 	}
+	// If the object was destructed in *this* block (and potentially resurrected),
+	// the storage has been cleared out, and we should *not* consult the previous
+	// database about any storage values. The only possible alternatives are:
+	//   1) resurrect happened, and new slot values were set -- those should
+	//      have been handles via pendingStorage above.
+	//   2) we don't have new values, and can deliver empty response back
+	if _, destructed := s.db.stateObjectsDestruct[s.address]; destructed {
+		return common.Hash{}
+	}
 	// If no live objects are available, attempt to use snapshots
 	var (
 		enc []byte
 		err error
 	)
 	if s.db.snap != nil {
-		// If the object was destructed in *this* block (and potentially resurrected),
-		// the storage has been cleared out, and we should *not* consult the previous
-		// snapshot about any storage values. The only possible alternatives are:
-		//   1) resurrect happened, and new slot values were set -- those should
-		//      have been handles via pendingStorage above.
-		//   2) we don't have new values, and can deliver empty response back
-		if _, destructed := s.db.snapDestructs[s.addrHash]; destructed {
-			return common.Hash{}
-		}
 		start := time.Now()
 		enc, err = s.db.snap.Storage(s.addrHash, crypto.Keccak256Hash(key.Bytes()))
 		if metrics.EnabledExpensive {
@@ -221,7 +214,12 @@ func (s *stateObject) GetCommittedState(db Database, key common.Hash) common.Has
 	// If the snapshot is unavailable or reading from it fails, load from the database.
 	if s.db.snap == nil || err != nil {
 		start := time.Now()
-		enc, err = s.getTrie(db).TryGet(key.Bytes())
+		tr, err := s.getTrie(db)
+		if err != nil {
+			s.setError(err)
+			return common.Hash{}
+		}
+		enc, err = tr.TryGet(key.Bytes())
 		if metrics.EnabledExpensive {
 			s.db.StorageReads += time.Since(start)
 		}
@@ -244,11 +242,6 @@ func (s *stateObject) GetCommittedState(db Database, key common.Hash) common.Has
 
 // SetState updates a value in account storage.
 func (s *stateObject) SetState(db Database, key, value common.Hash) {
-	// If the fake storage is set, put the temporary state update here.
-	if s.fakeStorage != nil {
-		s.fakeStorage[key] = value
-		return
-	}
 	// If the new value is the same as old, don't set
 	prev := s.GetState(db, key)
 	if prev == value {
@@ -263,24 +256,6 @@ func (s *stateObject) SetState(db Database, key, value common.Hash) {
 	s.setState(key, value)
 }
 
-// SetStorage replaces the entire state storage with the given one.
-//
-// After this function is called, all original state will be ignored and state
-// lookup only happens in the fake state storage.
-//
-// Note this function should only be used for debugging purpose.
-func (s *stateObject) SetStorage(storage map[common.Hash]common.Hash) {
-	// Allocate fake storage if it's nil.
-	if s.fakeStorage == nil {
-		s.fakeStorage = make(Storage)
-	}
-	for key, value := range storage {
-		s.fakeStorage[key] = value
-	}
-	// Don't bother journal since this function should only be used for
-	// debugging and the `fake` storage won't be committed to database.
-}
-
 func (s *stateObject) setState(key, value common.Hash) {
 	s.dirtyStorage[key] = value
 }
@@ -304,23 +279,29 @@ func (s *stateObject) finalise(prefetch bool) {
 }
 
 // updateTrie writes cached storage modifications into the object's storage trie.
-// It will return nil if the trie has not been loaded and no changes have been made
-func (s *stateObject) updateTrie(db Database) Trie {
+// It will return nil if the trie has not been loaded and no changes have been
+// made. An error will be returned if the trie can't be loaded/updated correctly.
+func (s *stateObject) updateTrie(db Database) (Trie, error) {
 	// Make sure all dirty slots are finalized into the pending storage area
 	s.finalise(false) // Don't prefetch anymore, pull directly if need be
 	if len(s.pendingStorage) == 0 {
-		return s.trie
+		return s.trie, nil
 	}
 	// Track the amount of time wasted on updating the storage trie
 	if metrics.EnabledExpensive {
 		defer func(start time.Time) { s.db.StorageUpdates += time.Since(start) }(time.Now())
 	}
 	// The snapshot storage map for the object
-	var storage map[common.Hash][]byte
+	var (
+		storage map[common.Hash][]byte
+		hasher  = s.db.hasher
+	)
+	tr, err := s.getTrie(db)
+	if err != nil {
+		s.setError(err)
+		return nil, err
+	}
 	// Insert all the pending updates into the trie
-	tr := s.getTrie(db)
-	hasher := s.db.hasher
-
 	usedStorage := make([][]byte, 0, len(s.pendingStorage))
 	for key, value := range s.pendingStorage {
 		// Skip noop changes, persist actual changes
@@ -331,12 +312,18 @@ func (s *stateObject) updateTrie(db Database) Trie {
 
 		var v []byte
 		if (value == common.Hash{}) {
-			s.setError(tr.TryDelete(key[:]))
+			if err := tr.TryDelete(key[:]); err != nil {
+				s.setError(err)
+				return nil, err
+			}
 			s.db.StorageDeleted += 1
 		} else {
 			// Encoding []byte cannot fail, ok to ignore the error.
 			v, _ = rlp.EncodeToBytes(common.TrimLeftZeroes(value[:]))
-			s.setError(tr.TryUpdate(key[:], v))
+			if err := tr.TryUpdate(key[:], v); err != nil {
+				s.setError(err)
+				return nil, err
+			}
 			s.db.StorageUpdated += 1
 		}
 		// If state snapshotting is active, cache the data til commit
@@ -358,40 +345,48 @@ func (s *stateObject) updateTrie(db Database) Trie {
 	if len(s.pendingStorage) > 0 {
 		s.pendingStorage = make(Storage)
 	}
-	return tr
+	return tr, nil
 }
 
-// UpdateRoot sets the trie root to the current root hash of
+// UpdateRoot sets the trie root to the current root hash of. An error
+// will be returned if trie root hash is not computed correctly.
 func (s *stateObject) updateRoot(db Database) {
+	tr, err := s.updateTrie(db)
+	if err != nil {
+		s.setError(fmt.Errorf("updateRoot (%x) error: %w", s.address, err))
+		return
+	}
 	// If nothing changed, don't bother with hashing anything
-	if s.updateTrie(db) == nil {
+	if tr == nil {
 		return
 	}
 	// Track the amount of time wasted on hashing the storage trie
 	if metrics.EnabledExpensive {
 		defer func(start time.Time) { s.db.StorageHashes += time.Since(start) }(time.Now())
 	}
-	s.data.Root = s.trie.Hash()
+	s.data.Root = tr.Hash()
 }
 
-// CommitTrie the storage trie of the object to db.
-// This updates the trie root.
-func (s *stateObject) CommitTrie(db Database) (*trie.NodeSet, error) {
-	// If nothing changed, don't bother with hashing anything
-	if s.updateTrie(db) == nil {
-		return nil, nil
+// commitTrie submits the storage changes into the storage trie and re-computes
+// the root. Besides, all trie changes will be collected in a nodeset and returned.
+func (s *stateObject) commitTrie(db Database) (*trie.NodeSet, error) {
+	tr, err := s.updateTrie(db)
+	if err != nil {
+		return nil, err
 	}
 	if s.dbErr != nil {
 		return nil, s.dbErr
 	}
+	// If nothing changed, don't bother with committing anything
+	if tr == nil {
+		return nil, nil
+	}
 	// Track the amount of time wasted on committing the storage trie
 	if metrics.EnabledExpensive {
 		defer func(start time.Time) { s.db.StorageCommits += time.Since(start) }(time.Now())
 	}
-	root, nodes, err := s.trie.Commit(false)
-	if err == nil {
-		s.data.Root = root
-	}
+	root, nodes := tr.Commit(false)
+	s.data.Root = root
 	return nodes, err
 }
 
@@ -449,7 +444,7 @@ func (s *stateObject) deepCopy(db *StateDB) *stateObject {
 // Attribute accessors
 //
 
-// Returns the address of the contract/account
+// Address returns the address of the contract/account
 func (s *stateObject) Address() common.Address {
 	return s.address
 }
@@ -527,7 +522,7 @@ func (s *stateObject) Nonce() uint64 {
 	return s.data.Nonce
 }
 
-// Never called, but must be present to allow stateObject to be used
+// Value is never called, but must be present to allow stateObject to be used
 // as a vm.Account interface that also satisfies the vm.ContractRef
 // interface. Interfaces are awesome.
 func (s *stateObject) Value() *big.Int {

core/state/statedb.go

@@ -31,6 +31,7 @@ import (
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/metrics"
+	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
 )
@@ -71,16 +72,16 @@ type StateDB struct {
 	// It will be updated when the Commit is called.
 	originalRoot common.Hash
 
-	snaps         *snapshot.Tree
-	snap          snapshot.Snapshot
-	snapDestructs map[common.Hash]struct{}
-	snapAccounts  map[common.Hash][]byte
-	snapStorage   map[common.Hash]map[common.Hash][]byte
+	snaps        *snapshot.Tree
+	snap         snapshot.Snapshot
+	snapAccounts map[common.Hash][]byte
+	snapStorage  map[common.Hash]map[common.Hash][]byte
 
 	// This map holds 'live' objects, which will get modified while processing a state transition.
-	stateObjects        map[common.Address]*stateObject
-	stateObjectsPending map[common.Address]struct{} // State objects finalized but not yet written to the trie
-	stateObjectsDirty   map[common.Address]struct{} // State objects modified in the current execution
+	stateObjects         map[common.Address]*stateObject
+	stateObjectsPending  map[common.Address]struct{} // State objects finalized but not yet written to the trie
+	stateObjectsDirty    map[common.Address]struct{} // State objects modified in the current execution
+	stateObjectsDestruct map[common.Address]struct{} // State objects destructed in the block
 
 	// DB error.
 	// State objects are used by the consensus core and VM which are
@@ -102,6 +103,9 @@ type StateDB struct {
 	// Per-transaction access list
 	accessList *accessList
 
+	// Transient storage
+	transientStorage transientStorage
+
 	// Journal of state modifications. This is the backbone of
 	// Snapshot and RevertToSnapshot.
 	journal        *journal
@@ -120,6 +124,7 @@ type StateDB struct {
 	SnapshotAccountReads time.Duration
 	SnapshotStorageReads time.Duration
 	SnapshotCommits      time.Duration
+	TrieDBCommits        time.Duration
 
 	AccountUpdated int
 	StorageUpdated int
@@ -134,22 +139,23 @@ func New(root common.Hash, db Database, snaps *snapshot.Tree) (*StateDB, error)
 		return nil, err
 	}
 	sdb := &StateDB{
-		db:                  db,
-		trie:                tr,
-		originalRoot:        root,
-		snaps:               snaps,
-		stateObjects:        make(map[common.Address]*stateObject),
-		stateObjectsPending: make(map[common.Address]struct{}),
-		stateObjectsDirty:   make(map[common.Address]struct{}),
-		logs:                make(map[common.Hash][]*types.Log),
-		preimages:           make(map[common.Hash][]byte),
-		journal:             newJournal(),
-		accessList:          newAccessList(),
-		hasher:              crypto.NewKeccakState(),
+		db:                   db,
+		trie:                 tr,
+		originalRoot:         root,
+		snaps:                snaps,
+		stateObjects:         make(map[common.Address]*stateObject),
+		stateObjectsPending:  make(map[common.Address]struct{}),
+		stateObjectsDirty:    make(map[common.Address]struct{}),
+		stateObjectsDestruct: make(map[common.Address]struct{}),
+		logs:                 make(map[common.Hash][]*types.Log),
+		preimages:            make(map[common.Hash][]byte),
+		journal:              newJournal(),
+		accessList:           newAccessList(),
+		transientStorage:     newTransientStorage(),
+		hasher:               crypto.NewKeccakState(),
 	}
 	if sdb.snaps != nil {
 		if sdb.snap = sdb.snaps.Snapshot(root); sdb.snap != nil {
-			sdb.snapDestructs = make(map[common.Hash]struct{})
 			sdb.snapAccounts = make(map[common.Hash][]byte)
 			sdb.snapStorage = make(map[common.Hash]map[common.Hash][]byte)
 		}
@@ -209,9 +215,12 @@ func (s *StateDB) AddLog(log *types.Log) {
 	s.logSize++
 }
 
-func (s *StateDB) GetLogs(hash common.Hash, blockHash common.Hash) []*types.Log {
+// GetLogs returns the logs matching the specified transaction hash, and annotates
+// them with the given blockNumber and blockHash.
+func (s *StateDB) GetLogs(hash common.Hash, blockNumber uint64, blockHash common.Hash) []*types.Log {
 	logs := s.logs[hash]
 	for _, l := range logs {
+		l.BlockNumber = blockNumber
 		l.BlockHash = blockHash
 	}
 	return logs
@@ -339,13 +348,19 @@ func (s *StateDB) GetProofByHash(addrHash common.Hash) ([][]byte, error) {
 
 // GetStorageProof returns the Merkle proof for given storage slot.
 func (s *StateDB) GetStorageProof(a common.Address, key common.Hash) ([][]byte, error) {
-	var proof proofList
-	trie := s.StorageTrie(a)
-	if trie == nil {
-		return proof, errors.New("storage trie for requested address does not exist")
+	trie, err := s.StorageTrie(a)
+	if err != nil {
+		return nil, err
 	}
-	err := trie.Prove(crypto.Keccak256(key.Bytes()), 0, &proof)
-	return proof, err
+	if trie == nil {
+		return nil, errors.New("storage trie for requested address does not exist")
+	}
+	var proof proofList
+	err = trie.Prove(crypto.Keccak256(key.Bytes()), 0, &proof)
+	if err != nil {
+		return nil, err
+	}
+	return proof, nil
 }
 
 // GetCommittedState retrieves a value from the given account's committed storage trie.
@@ -362,15 +377,18 @@ func (s *StateDB) Database() Database {
 	return s.db
 }
 
-// StorageTrie returns the storage trie of an account.
-// The return value is a copy and is nil for non-existent accounts.
-func (s *StateDB) StorageTrie(addr common.Address) Trie {
+// StorageTrie returns the storage trie of an account. The return value is a copy
+// and is nil for non-existent accounts. An error will be returned if storage trie
+// is existent but can't be loaded correctly.
+func (s *StateDB) StorageTrie(addr common.Address) (Trie, error) {
 	stateObject := s.getStateObject(addr)
 	if stateObject == nil {
-		return nil
+		return nil, nil
 	}
 	cpy := stateObject.deepCopy(s)
-	cpy.updateTrie(s.db)
+	if _, err := cpy.updateTrie(s.db); err != nil {
+		return nil, err
+	}
 	return cpy.getTrie(s.db)
 }
 
@@ -433,9 +451,15 @@ func (s *StateDB) SetState(addr common.Address, key, value common.Hash) {
 // SetStorage replaces the entire storage for the specified account with given
 // storage. This function should only be used for debugging.
 func (s *StateDB) SetStorage(addr common.Address, storage map[common.Hash]common.Hash) {
+	// SetStorage needs to wipe existing storage. We achieve this by pretending
+	// that the account self-destructed earlier in this block, by flagging
+	// it in stateObjectsDestruct. The effect of doing so is that storage lookups
+	// will not hit disk, since it is assumed that the disk-data is belonging
+	// to a previous incarnation of the object.
+	s.stateObjectsDestruct[addr] = struct{}{}
 	stateObject := s.GetOrNewStateObject(addr)
-	if stateObject != nil {
-		stateObject.SetStorage(storage)
+	for k, v := range storage {
+		stateObject.SetState(s.db, k, v)
 	}
 }
 
@@ -460,6 +484,35 @@ func (s *StateDB) Suicide(addr common.Address) bool {
 	return true
 }
 
+// SetTransientState sets transient storage for a given account. It
+// adds the change to the journal so that it can be rolled back
+// to its previous value if there is a revert.
+func (s *StateDB) SetTransientState(addr common.Address, key, value common.Hash) {
+	prev := s.GetTransientState(addr, key)
+	if prev == value {
+		return
+	}
+
+	s.journal.append(transientStorageChange{
+		account:  &addr,
+		key:      key,
+		prevalue: prev,
+	})
+
+	s.setTransientState(addr, key, value)
+}
+
+// setTransientState is a lower level setter for transient storage. It
+// is called during a revert to prevent modifications to the journal.
+func (s *StateDB) setTransientState(addr common.Address, key, value common.Hash) {
+	s.transientStorage.Set(addr, key, value)
+}
+
+// GetTransientState gets transient storage for a given account.
+func (s *StateDB) GetTransientState(addr common.Address, key common.Hash) common.Hash {
+	return s.transientStorage.Get(addr, key)
+}
+
 //
 // Setting, updating & deleting state object methods.
 //
@@ -472,7 +525,7 @@ func (s *StateDB) updateStateObject(obj *stateObject) {
 	}
 	// Encode the account and update the account trie
 	addr := obj.Address()
-	if err := s.trie.TryUpdateAccount(addr[:], &obj.data); err != nil {
+	if err := s.trie.TryUpdateAccount(addr, &obj.data); err != nil {
 		s.setError(fmt.Errorf("updateStateObject (%x) error: %v", addr[:], err))
 	}
 
@@ -493,7 +546,7 @@ func (s *StateDB) deleteStateObject(obj *stateObject) {
 	}
 	// Delete the account from the trie
 	addr := obj.Address()
-	if err := s.trie.TryDeleteAccount(addr[:]); err != nil {
+	if err := s.trie.TryDeleteAccount(addr); err != nil {
 		s.setError(fmt.Errorf("deleteStateObject (%x) error: %v", addr[:], err))
 	}
 }
@@ -547,7 +600,7 @@ func (s *StateDB) getDeletedStateObject(addr common.Address) *stateObject {
 	if data == nil {
 		start := time.Now()
 		var err error
-		data, err = s.trie.TryGetAccount(addr.Bytes())
+		data, err = s.trie.TryGetAccount(addr)
 		if metrics.EnabledExpensive {
 			s.AccountReads += time.Since(start)
 		}
@@ -584,10 +637,10 @@ func (s *StateDB) createObject(addr common.Address) (newobj, prev *stateObject)
 	prev = s.getDeletedStateObject(addr) // Note, prev might have been deleted, we need that!
 
 	var prevdestruct bool
-	if s.snap != nil && prev != nil {
-		_, prevdestruct = s.snapDestructs[prev.addrHash]
+	if prev != nil {
+		_, prevdestruct = s.stateObjectsDestruct[prev.address]
 		if !prevdestruct {
-			s.snapDestructs[prev.addrHash] = struct{}{}
+			s.stateObjectsDestruct[prev.address] = struct{}{}
 		}
 	}
 	newobj = newObject(s, addr, types.StateAccount{})
@@ -609,8 +662,8 @@ func (s *StateDB) createObject(addr common.Address) (newobj, prev *stateObject)
 // CreateAccount is called during the EVM CREATE operation. The situation might arise that
 // a contract does the following:
 //
-//   1. sends funds to sha(account ++ (nonce + 1))
-//   2. tx_create(sha(account ++ nonce)) (note that this gets the address of 1)
+//  1. sends funds to sha(account ++ (nonce + 1))
+//  2. tx_create(sha(account ++ nonce)) (note that this gets the address of 1)
 //
 // Carrying over the balance ensures that Ether doesn't disappear.
 func (s *StateDB) CreateAccount(addr common.Address) {
@@ -625,7 +678,11 @@ func (db *StateDB) ForEachStorage(addr common.Address, cb func(key, value common
 	if so == nil {
 		return nil
 	}
-	it := trie.NewIterator(so.getTrie(db.db).NodeIterator(nil))
+	tr, err := so.getTrie(db.db)
+	if err != nil {
+		return err
+	}
+	it := trie.NewIterator(tr.NodeIterator(nil))
 
 	for it.Next() {
 		key := common.BytesToHash(db.trie.GetKey(it.Key))
@@ -654,18 +711,19 @@ func (db *StateDB) ForEachStorage(addr common.Address, cb func(key, value common
 func (s *StateDB) Copy() *StateDB {
 	// Copy all the basic fields, initialize the memory ones
 	state := &StateDB{
-		db:                  s.db,
-		trie:                s.db.CopyTrie(s.trie),
-		originalRoot:        s.originalRoot,
-		stateObjects:        make(map[common.Address]*stateObject, len(s.journal.dirties)),
-		stateObjectsPending: make(map[common.Address]struct{}, len(s.stateObjectsPending)),
-		stateObjectsDirty:   make(map[common.Address]struct{}, len(s.journal.dirties)),
-		refund:              s.refund,
-		logs:                make(map[common.Hash][]*types.Log, len(s.logs)),
-		logSize:             s.logSize,
-		preimages:           make(map[common.Hash][]byte, len(s.preimages)),
-		journal:             newJournal(),
-		hasher:              crypto.NewKeccakState(),
+		db:                   s.db,
+		trie:                 s.db.CopyTrie(s.trie),
+		originalRoot:         s.originalRoot,
+		stateObjects:         make(map[common.Address]*stateObject, len(s.journal.dirties)),
+		stateObjectsPending:  make(map[common.Address]struct{}, len(s.stateObjectsPending)),
+		stateObjectsDirty:    make(map[common.Address]struct{}, len(s.journal.dirties)),
+		stateObjectsDestruct: make(map[common.Address]struct{}, len(s.stateObjectsDestruct)),
+		refund:               s.refund,
+		logs:                 make(map[common.Hash][]*types.Log, len(s.logs)),
+		logSize:              s.logSize,
+		preimages:            make(map[common.Hash][]byte, len(s.preimages)),
+		journal:              newJournal(),
+		hasher:               crypto.NewKeccakState(),
 	}
 	// Copy the dirty states, logs, and preimages
 	for addr := range s.journal.dirties {
@@ -675,7 +733,7 @@ func (s *StateDB) Copy() *StateDB {
 		// nil
 		if object, exist := s.stateObjects[addr]; exist {
 			// Even though the original object is dirty, we are not copying the journal,
-			// so we need to make sure that anyside effect the journal would have caused
+			// so we need to make sure that any side-effect the journal would have caused
 			// during a commit (or similar op) is already applied to the copy.
 			state.stateObjects[addr] = object.deepCopy(state)
 
@@ -683,9 +741,10 @@ func (s *StateDB) Copy() *StateDB {
 			state.stateObjectsPending[addr] = struct{}{} // Mark the copy pending to force external (account) commits
 		}
 	}
-	// Above, we don't copy the actual journal. This means that if the copy is copied, the
-	// loop above will be a no-op, since the copy's journal is empty.
-	// Thus, here we iterate over stateObjects, to enable copies of copies
+	// Above, we don't copy the actual journal. This means that if the copy
+	// is copied, the loop above will be a no-op, since the copy's journal
+	// is empty. Thus, here we iterate over stateObjects, to enable copies
+	// of copies.
 	for addr := range s.stateObjectsPending {
 		if _, exist := state.stateObjects[addr]; !exist {
 			state.stateObjects[addr] = s.stateObjects[addr].deepCopy(state)
@@ -698,6 +757,10 @@ func (s *StateDB) Copy() *StateDB {
 		}
 		state.stateObjectsDirty[addr] = struct{}{}
 	}
+	// Deep copy the destruction flag.
+	for addr := range s.stateObjectsDestruct {
+		state.stateObjectsDestruct[addr] = struct{}{}
+	}
 	for hash, logs := range s.logs {
 		cpy := make([]*types.Log, len(logs))
 		for i, l := range logs {
@@ -709,12 +772,14 @@ func (s *StateDB) Copy() *StateDB {
 	for hash, preimage := range s.preimages {
 		state.preimages[hash] = preimage
 	}
-	// Do we need to copy the access list? In practice: No. At the start of a
-	// transaction, the access list is empty. In practice, we only ever copy state
-	// _between_ transactions/blocks, never in the middle of a transaction.
-	// However, it doesn't cost us much to copy an empty list, so we do it anyway
-	// to not blow up if we ever decide copy it in the middle of a transaction
+	// Do we need to copy the access list and transient storage?
+	// In practice: No. At the start of a transaction, these two lists are empty.
+	// In practice, we only ever copy state _between_ transactions/blocks, never
+	// in the middle of a transaction. However, it doesn't cost us much to copy
+	// empty lists, so we do it anyway to not blow up if we ever decide copy them
+	// in the middle of a transaction.
 	state.accessList = s.accessList.Copy()
+	state.transientStorage = s.transientStorage.Copy()
 
 	// If there's a prefetcher running, make an inactive copy of it that can
 	// only access data but does not actively preload (since the user will not
@@ -724,16 +789,13 @@ func (s *StateDB) Copy() *StateDB {
 	}
 	if s.snaps != nil {
 		// In order for the miner to be able to use and make additions
-		// to the snapshot tree, we need to copy that aswell.
+		// to the snapshot tree, we need to copy that as well.
 		// Otherwise, any block mined by ourselves will cause gaps in the tree,
 		// and force the miner to operate trie-backed only
 		state.snaps = s.snaps
 		state.snap = s.snap
+
 		// deep copy needed
-		state.snapDestructs = make(map[common.Hash]struct{})
-		for k, v := range s.snapDestructs {
-			state.snapDestructs[k] = v
-		}
 		state.snapAccounts = make(map[common.Hash][]byte)
 		for k, v := range s.snapAccounts {
 			state.snapAccounts[k] = v
@@ -798,14 +860,17 @@ func (s *StateDB) Finalise(deleteEmptyObjects bool) {
 		if obj.suicided || (deleteEmptyObjects && obj.empty()) {
 			obj.deleted = true
 
+			// We need to maintain account deletions explicitly (will remain
+			// set indefinitely).
+			s.stateObjectsDestruct[obj.address] = struct{}{}
+
 			// If state snapshotting is active, also mark the destruction there.
 			// Note, we can't do this only at the end of a block because multiple
 			// transactions within the same block might self destruct and then
 			// resurrect an account; but the snapshotter needs both events.
 			if s.snap != nil {
-				s.snapDestructs[obj.addrHash] = struct{}{} // We need to maintain account deletions explicitly (will remain set indefinitely)
-				delete(s.snapAccounts, obj.addrHash)       // Clear out any previously updated account data (may be recreated via a ressurrect)
-				delete(s.snapStorage, obj.addrHash)        // Clear out any previously updated storage data (may be recreated via a ressurrect)
+				delete(s.snapAccounts, obj.addrHash) // Clear out any previously updated account data (may be recreated via a resurrect)
+				delete(s.snapStorage, obj.addrHash)  // Clear out any previously updated storage data (may be recreated via a resurrect)
 			}
 		} else {
 			obj.finalise(true) // Prefetch slots in the background
@@ -888,9 +953,10 @@ func (s *StateDB) IntermediateRoot(deleteEmptyObjects bool) common.Hash {
 	return s.trie.Hash()
 }
 
-// Prepare sets the current transaction hash and index which are
-// used when the EVM emits new state logs.
-func (s *StateDB) Prepare(thash common.Hash, ti int) {
+// SetTxContext sets the current transaction hash and index which are
+// used when the EVM emits new state logs. It should be invoked before
+// transaction execution.
+func (s *StateDB) SetTxContext(thash common.Hash, ti int) {
 	s.thash = thash
 	s.txIndex = ti
 }
@@ -913,9 +979,11 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 
 	// Commit objects to the trie, measuring the elapsed time
 	var (
-		accountTrieNodes int
-		storageTrieNodes int
-		nodes            = trie.NewMergedNodeSet()
+		accountTrieNodesUpdated int
+		accountTrieNodesDeleted int
+		storageTrieNodesUpdated int
+		storageTrieNodesDeleted int
+		nodes                   = trie.NewMergedNodeSet()
 	)
 	// begin PluGeth injection
 	codeUpdates := make(map[common.Hash][]byte)
@@ -932,7 +1000,7 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 				obj.dirtyCode = false
 			}
 			// Write any storage changes in the state object to its storage trie
-			set, err := obj.CommitTrie(s.db)
+			set, err := obj.commitTrie(s.db)
 			if err != nil {
 				return common.Hash{}, err
 			}
@@ -941,9 +1009,17 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 				if err := nodes.Merge(set); err != nil {
 					return common.Hash{}, err
 				}
-				storageTrieNodes += set.Len()
+				updates, deleted := set.Size()
+				storageTrieNodesUpdated += updates
+				storageTrieNodesDeleted += deleted
 			}
 		}
+		// If the contract is destructed, the storage is still left in the
+		// database as dangling data. Theoretically it's should be wiped from
+		// database as well, but in hash-based-scheme it's extremely hard to
+		// determine that if the trie nodes are also referenced by other storage,
+		// and in path-based-scheme some technical challenges are still unsolved.
+		// Although it won't affect the correctness but please fix it TODO(rjl493456442).
 	}
 	if len(s.stateObjectsDirty) > 0 {
 		s.stateObjectsDirty = make(map[common.Address]struct{})
@@ -958,16 +1034,13 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 	if metrics.EnabledExpensive {
 		start = time.Now()
 	}
-	root, set, err := s.trie.Commit(true)
-	if err != nil {
-		return common.Hash{}, err
-	}
+	root, set := s.trie.Commit(true)
 	// Merge the dirty nodes of account trie into global set
 	if set != nil {
 		if err := nodes.Merge(set); err != nil {
 			return common.Hash{}, err
 		}
-		accountTrieNodes = set.Len()
+		accountTrieNodesUpdated, accountTrieNodesDeleted = set.Size()
 	}
 	if metrics.EnabledExpensive {
 		s.AccountCommits += time.Since(start)
@@ -976,16 +1049,16 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 		storageUpdatedMeter.Mark(int64(s.StorageUpdated))
 		accountDeletedMeter.Mark(int64(s.AccountDeleted))
 		storageDeletedMeter.Mark(int64(s.StorageDeleted))
-		accountTrieCommittedMeter.Mark(int64(accountTrieNodes))
-		storageTriesCommittedMeter.Mark(int64(storageTrieNodes))
+		accountTrieUpdatedMeter.Mark(int64(accountTrieNodesUpdated))
+		accountTrieDeletedMeter.Mark(int64(accountTrieNodesDeleted))
+		storageTriesUpdatedMeter.Mark(int64(storageTrieNodesUpdated))
+		storageTriesDeletedMeter.Mark(int64(storageTrieNodesDeleted))
 		s.AccountUpdated, s.AccountDeleted = 0, 0
 		s.StorageUpdated, s.StorageDeleted = 0, 0
 	}
 	// If snapshotting is enabled, update the snapshot tree with this new version
 	if s.snap != nil {
-		if metrics.EnabledExpensive {
-			defer func(start time.Time) { s.SnapshotCommits += time.Since(start) }(time.Now())
-		}
+		start := time.Now()
 		// Only update if there's a state transition (skip empty Clique blocks)
 		if parent := s.snap.Root(); parent != root {
 			//begin PluGeth code injection
@@ -1004,42 +1077,73 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 				}
 			}
 		}
-		s.snap, s.snapDestructs, s.snapAccounts, s.snapStorage = nil, nil, nil, nil
+		if metrics.EnabledExpensive {
+			s.SnapshotCommits += time.Since(start)
+		}
+		s.snap, s.snapAccounts, s.snapStorage = nil, nil, nil
 	}
-	if err := s.db.TrieDB().Update(nodes); err != nil {
-		return common.Hash{}, err
+	if len(s.stateObjectsDestruct) > 0 {
+		s.stateObjectsDestruct = make(map[common.Address]struct{})
 	}
-	s.originalRoot = root
-	return root, err
+	if root == (common.Hash{}) {
+		root = emptyRoot
+	}
+	origin := s.originalRoot
+	if origin == (common.Hash{}) {
+		origin = emptyRoot
+	}
+	if root != origin {
+		start := time.Now()
+		if err := s.db.TrieDB().Update(nodes); err != nil {
+			return common.Hash{}, err
+		}
+		s.originalRoot = root
+		if metrics.EnabledExpensive {
+			s.TrieDBCommits += time.Since(start)
+		}
+	}
+	return root, nil
 }
 
-// PrepareAccessList handles the preparatory steps for executing a state transition with
-// regards to both EIP-2929 and EIP-2930:
+// Prepare handles the preparatory steps for executing a state transition with.
+// This method must be invoked before state transition.
 //
+// Berlin fork:
 // - Add sender to access list (2929)
 // - Add destination to access list (2929)
 // - Add precompiles to access list (2929)
 // - Add the contents of the optional tx access list (2930)
 //
-// This method should only be called if Berlin/2929+2930 is applicable at the current number.
-func (s *StateDB) PrepareAccessList(sender common.Address, dst *common.Address, precompiles []common.Address, list types.AccessList) {
-	// Clear out any leftover from previous executions
-	s.accessList = newAccessList()
+// Potential EIPs:
+// - Reset access list (Berlin)
+// - Add coinbase to access list (EIP-3651)
+// - Reset transient storage (EIP-1153)
+func (s *StateDB) Prepare(rules params.Rules, sender, coinbase common.Address, dst *common.Address, precompiles []common.Address, list types.AccessList) {
+	if rules.IsBerlin {
+		// Clear out any leftover from previous executions
+		al := newAccessList()
+		s.accessList = al
 
-	s.AddAddressToAccessList(sender)
-	if dst != nil {
-		s.AddAddressToAccessList(*dst)
-		// If it's a create-tx, the destination will be added inside evm.create
-	}
-	for _, addr := range precompiles {
-		s.AddAddressToAccessList(addr)
-	}
-	for _, el := range list {
-		s.AddAddressToAccessList(el.Address)
-		for _, key := range el.StorageKeys {
-			s.AddSlotToAccessList(el.Address, key)
+		al.AddAddress(sender)
+		if dst != nil {
+			al.AddAddress(*dst)
+			// If it's a create-tx, the destination will be added inside evm.create
+		}
+		for _, addr := range precompiles {
+			al.AddAddress(addr)
+		}
+		for _, el := range list {
+			al.AddAddress(el.Address)
+			for _, key := range el.StorageKeys {
+				al.AddSlot(el.Address, key)
+			}
+		}
+		if rules.IsShanghai { // EIP-3651: warm coinbase
+			al.AddAddress(coinbase)
 		}
 	}
+	// Reset transient storage at the beginning of transaction execution
+	s.transientStorage = newTransientStorage()
 }
 
 // AddAddressToAccessList adds the given address to the access list
@@ -1076,3 +1180,17 @@ func (s *StateDB) AddressInAccessList(addr common.Address) bool {
 func (s *StateDB) SlotInAccessList(addr common.Address, slot common.Hash) (addressPresent bool, slotPresent bool) {
 	return s.accessList.Contains(addr, slot)
 }
+
+// convertAccountSet converts a provided account set from address keyed to hash keyed.
+func (s *StateDB) convertAccountSet(set map[common.Address]struct{}) map[common.Hash]struct{} {
+	ret := make(map[common.Hash]struct{})
+	for addr := range set {
+		obj, exist := s.stateObjects[addr]
+		if !exist {
+			ret[crypto.Keccak256Hash(addr[:])] = struct{}{}
+		} else {
+			ret[obj.addrHash] = struct{}{}
+		}
+	}
+	return ret
+}

core/state/statedb_test.go

@@ -55,7 +55,7 @@ func TestUpdateLeaks(t *testing.T) {
 	}
 
 	root := state.IntermediateRoot(false)
-	if err := state.Database().TrieDB().Commit(root, false, nil); err != nil {
+	if err := state.Database().TrieDB().Commit(root, false); err != nil {
 		t.Errorf("can not commit trie %v to persistent database", root.Hex())
 	}
 
@@ -106,7 +106,7 @@ func TestIntermediateLeaks(t *testing.T) {
 	if err != nil {
 		t.Fatalf("failed to commit transition state: %v", err)
 	}
-	if err = transState.Database().TrieDB().Commit(transRoot, false, nil); err != nil {
+	if err = transState.Database().TrieDB().Commit(transRoot, false); err != nil {
 		t.Errorf("can not commit trie %v to persistent database", transRoot.Hex())
 	}
 
@@ -114,7 +114,7 @@ func TestIntermediateLeaks(t *testing.T) {
 	if err != nil {
 		t.Fatalf("failed to commit final state: %v", err)
 	}
-	if err = finalState.Database().TrieDB().Commit(finalRoot, false, nil); err != nil {
+	if err = finalState.Database().TrieDB().Commit(finalRoot, false); err != nil {
 		t.Errorf("can not commit trie %v to persistent database", finalRoot.Hex())
 	}
 
@@ -342,6 +342,16 @@ func newTestAction(addr common.Address, r *rand.Rand) testAction {
 			},
 			args: make([]int64, 1),
 		},
+		{
+			name: "SetTransientState",
+			fn: func(a testAction, s *StateDB) {
+				var key, val common.Hash
+				binary.BigEndian.PutUint16(key[:], uint16(a.args[0]))
+				binary.BigEndian.PutUint16(val[:], uint16(a.args[1]))
+				s.SetTransientState(addr, key, val)
+			},
+			args: make([]int64, 2),
+		},
 	}
 	action := actions[r.Intn(len(actions))]
 	var nameargs []string
@@ -463,9 +473,9 @@ func (test *snapshotTest) checkEqual(state, checkstate *StateDB) error {
 		return fmt.Errorf("got GetRefund() == %d, want GetRefund() == %d",
 			state.GetRefund(), checkstate.GetRefund())
 	}
-	if !reflect.DeepEqual(state.GetLogs(common.Hash{}, common.Hash{}), checkstate.GetLogs(common.Hash{}, common.Hash{})) {
+	if !reflect.DeepEqual(state.GetLogs(common.Hash{}, 0, common.Hash{}), checkstate.GetLogs(common.Hash{}, 0, common.Hash{})) {
 		return fmt.Errorf("got GetLogs(common.Hash{}) == %v, want GetLogs(common.Hash{}) == %v",
-			state.GetLogs(common.Hash{}, common.Hash{}), checkstate.GetLogs(common.Hash{}, common.Hash{}))
+			state.GetLogs(common.Hash{}, 0, common.Hash{}), checkstate.GetLogs(common.Hash{}, 0, common.Hash{}))
 	}
 	return nil
 }
@@ -938,7 +948,7 @@ func TestFlushOrderDataLoss(t *testing.T) {
 	if err := statedb.TrieDB().Cap(1024); err != nil {
 		t.Fatalf("failed to cap trie dirty cache: %v", err)
 	}
-	if err := statedb.TrieDB().Commit(root, false, nil); err != nil {
+	if err := statedb.TrieDB().Commit(root, false); err != nil {
 		t.Fatalf("failed to commit state trie: %v", err)
 	}
 	// Reopen the state trie from flushed disk and verify it
@@ -954,3 +964,37 @@ func TestFlushOrderDataLoss(t *testing.T) {
 		}
 	}
 }
+
+func TestStateDBTransientStorage(t *testing.T) {
+	memDb := rawdb.NewMemoryDatabase()
+	db := NewDatabase(memDb)
+	state, _ := New(common.Hash{}, db, nil)
+
+	key := common.Hash{0x01}
+	value := common.Hash{0x02}
+	addr := common.Address{}
+
+	state.SetTransientState(addr, key, value)
+	if exp, got := 1, state.journal.length(); exp != got {
+		t.Fatalf("journal length mismatch: have %d, want %d", got, exp)
+	}
+	// the retrieved value should equal what was set
+	if got := state.GetTransientState(addr, key); got != value {
+		t.Fatalf("transient storage mismatch: have %x, want %x", got, value)
+	}
+
+	// revert the transient state being set and then check that the
+	// value is now the empty hash
+	state.journal.revert(state, 0)
+	if got, exp := state.GetTransientState(addr, key), (common.Hash{}); exp != got {
+		t.Fatalf("transient storage mismatch: have %x, want %x", got, exp)
+	}
+
+	// set transient state and then copy the statedb and ensure that
+	// the transient state is copied
+	state.SetTransientState(addr, key, value)
+	cpy := state.Copy()
+	if got := cpy.GetTransientState(addr, key); got != value {
+		t.Fatalf("transient storage mismatch: have %x, want %x", got, value)
+	}
+}

core/state/sync.go

@@ -27,7 +27,7 @@ import (
 )
 
 // NewStateSync create a new state trie download scheduler.
-func NewStateSync(root common.Hash, database ethdb.KeyValueReader, onLeaf func(keys [][]byte, leaf []byte) error) *trie.Sync {
+func NewStateSync(root common.Hash, database ethdb.KeyValueReader, onLeaf func(keys [][]byte, leaf []byte) error, scheme string) *trie.Sync {
 	// Register the storage slot callback if the external callback is specified.
 	var onSlot func(keys [][]byte, path []byte, leaf []byte, parent common.Hash, parentPath []byte) error
 	if onLeaf != nil {
@@ -52,6 +52,6 @@ func NewStateSync(root common.Hash, database ethdb.KeyValueReader, onLeaf func(k
 		syncer.AddCodeEntry(common.BytesToHash(obj.CodeHash), path, parent, parentPath)
 		return nil
 	}
-	syncer = trie.NewSync(root, database, onAccount)
+	syncer = trie.NewSync(root, database, onAccount, scheme)
 	return syncer
 }

core/state/sync_test.go

@@ -39,10 +39,11 @@ type testAccount struct {
 }
 
 // makeTestState create a sample test state to test node-wise reconstruction.
-func makeTestState() (Database, common.Hash, []*testAccount) {
+func makeTestState() (ethdb.Database, Database, common.Hash, []*testAccount) {
 	// Create an empty state
-	db := NewDatabase(rawdb.NewMemoryDatabase())
-	state, _ := New(common.Hash{}, db, nil)
+	db := rawdb.NewMemoryDatabase()
+	sdb := NewDatabase(db)
+	state, _ := New(common.Hash{}, sdb, nil)
 
 	// Fill it with some arbitrary data
 	var accounts []*testAccount
@@ -63,7 +64,7 @@ func makeTestState() (Database, common.Hash, []*testAccount) {
 		if i%5 == 0 {
 			for j := byte(0); j < 5; j++ {
 				hash := crypto.Keccak256Hash([]byte{i, i, i, i, i, j, j})
-				obj.SetState(db, hash, hash)
+				obj.SetState(sdb, hash, hash)
 			}
 		}
 		state.updateStateObject(obj)
@@ -72,7 +73,7 @@ func makeTestState() (Database, common.Hash, []*testAccount) {
 	root, _ := state.Commit(false)
 
 	// Return the generated state
-	return db, root, accounts
+	return db, sdb, root, accounts
 }
 
 // checkStateAccounts cross references a reconstructed state with an expected
@@ -104,7 +105,7 @@ func checkTrieConsistency(db ethdb.Database, root common.Hash) error {
 	if v, _ := db.Get(root[:]); v == nil {
 		return nil // Consider a non existent state consistent.
 	}
-	trie, err := trie.New(common.Hash{}, root, trie.NewDatabase(db))
+	trie, err := trie.New(trie.StateTrieID(root), trie.NewDatabase(db))
 	if err != nil {
 		return err
 	}
@@ -132,8 +133,9 @@ func checkStateConsistency(db ethdb.Database, root common.Hash) error {
 
 // Tests that an empty state is not scheduled for syncing.
 func TestEmptyStateSync(t *testing.T) {
+	db := trie.NewDatabase(rawdb.NewMemoryDatabase())
 	empty := common.HexToHash("56e81f171bcc55a6ff8345e692c0f86e5b48e01b996cadc001622fb5e363b421")
-	sync := NewStateSync(empty, rawdb.NewMemoryDatabase(), nil)
+	sync := NewStateSync(empty, rawdb.NewMemoryDatabase(), nil, db.Scheme())
 	if paths, nodes, codes := sync.Missing(1); len(paths) != 0 || len(nodes) != 0 || len(codes) != 0 {
 		t.Errorf("content requested for empty state: %v, %v, %v", nodes, paths, codes)
 	}
@@ -170,15 +172,15 @@ type stateElement struct {
 
 func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 	// Create a random state to copy
-	srcDb, srcRoot, srcAccounts := makeTestState()
+	_, srcDb, srcRoot, srcAccounts := makeTestState()
 	if commit {
-		srcDb.TrieDB().Commit(srcRoot, false, nil)
+		srcDb.TrieDB().Commit(srcRoot, false)
 	}
-	srcTrie, _ := trie.New(common.Hash{}, srcRoot, srcDb.TrieDB())
+	srcTrie, _ := trie.New(trie.StateTrieID(srcRoot), srcDb.TrieDB())
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil)
+	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
 
 	var (
 		nodeElements []stateElement
@@ -222,7 +224,8 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 					if err := rlp.DecodeBytes(srcTrie.Get(node.syncPath[0]), &acc); err != nil {
 						t.Fatalf("failed to decode account on path %x: %v", node.syncPath[0], err)
 					}
-					stTrie, err := trie.New(common.BytesToHash(node.syncPath[0]), acc.Root, srcDb.TrieDB())
+					id := trie.StorageTrieID(srcRoot, common.BytesToHash(node.syncPath[0]), acc.Root)
+					stTrie, err := trie.New(id, srcDb.TrieDB())
 					if err != nil {
 						t.Fatalf("failed to retriev storage trie for path %x: %v", node.syncPath[1], err)
 					}
@@ -280,11 +283,11 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 // partial results are returned, and the others sent only later.
 func TestIterativeDelayedStateSync(t *testing.T) {
 	// Create a random state to copy
-	srcDb, srcRoot, srcAccounts := makeTestState()
+	_, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil)
+	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
 
 	var (
 		nodeElements []stateElement
@@ -373,11 +376,11 @@ func TestIterativeRandomStateSyncBatched(t *testing.T)    { testIterativeRandomS
 
 func testIterativeRandomStateSync(t *testing.T, count int) {
 	// Create a random state to copy
-	srcDb, srcRoot, srcAccounts := makeTestState()
+	_, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil)
+	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
 
 	nodeQueue := make(map[string]stateElement)
 	codeQueue := make(map[common.Hash]struct{})
@@ -453,11 +456,11 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 // partial results are returned (Even those randomly), others sent only later.
 func TestIterativeRandomDelayedStateSync(t *testing.T) {
 	// Create a random state to copy
-	srcDb, srcRoot, srcAccounts := makeTestState()
+	_, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil)
+	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
 
 	nodeQueue := make(map[string]stateElement)
 	codeQueue := make(map[common.Hash]struct{})
@@ -543,7 +546,7 @@ func TestIterativeRandomDelayedStateSync(t *testing.T) {
 // the database.
 func TestIncompleteStateSync(t *testing.T) {
 	// Create a random state to copy
-	srcDb, srcRoot, srcAccounts := makeTestState()
+	db, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// isCodeLookup to save some hashing
 	var isCode = make(map[common.Hash]struct{})
@@ -553,15 +556,16 @@ func TestIncompleteStateSync(t *testing.T) {
 		}
 	}
 	isCode[common.BytesToHash(emptyCodeHash)] = struct{}{}
-	checkTrieConsistency(srcDb.TrieDB().DiskDB().(ethdb.Database), srcRoot)
+	checkTrieConsistency(db, srcRoot)
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil)
+	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
 
 	var (
-		addedCodes []common.Hash
-		addedNodes []common.Hash
+		addedCodes  []common.Hash
+		addedPaths  []string
+		addedHashes []common.Hash
 	)
 	nodeQueue := make(map[string]stateElement)
 	codeQueue := make(map[common.Hash]struct{})
@@ -598,15 +602,16 @@ func TestIncompleteStateSync(t *testing.T) {
 		var nodehashes []common.Hash
 		if len(nodeQueue) > 0 {
 			results := make([]trie.NodeSyncResult, 0, len(nodeQueue))
-			for key, element := range nodeQueue {
+			for path, element := range nodeQueue {
 				data, err := srcDb.TrieDB().Node(element.hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for %x", element.hash)
 				}
-				results = append(results, trie.NodeSyncResult{Path: key, Data: data})
+				results = append(results, trie.NodeSyncResult{Path: path, Data: data})
 
 				if element.hash != srcRoot {
-					addedNodes = append(addedNodes, element.hash)
+					addedPaths = append(addedPaths, element.path)
+					addedHashes = append(addedHashes, element.hash)
 				}
 				nodehashes = append(nodehashes, element.hash)
 			}
@@ -654,12 +659,18 @@ func TestIncompleteStateSync(t *testing.T) {
 		}
 		rawdb.WriteCode(dstDb, node, val)
 	}
-	for _, node := range addedNodes {
-		val := rawdb.ReadTrieNode(dstDb, node)
-		rawdb.DeleteTrieNode(dstDb, node)
-		if err := checkStateConsistency(dstDb, srcRoot); err == nil {
-			t.Errorf("trie inconsistency not caught, missing: %v", node.Hex())
+	scheme := srcDb.TrieDB().Scheme()
+	for i, path := range addedPaths {
+		owner, inner := trie.ResolvePath([]byte(path))
+		hash := addedHashes[i]
+		val := rawdb.ReadTrieNode(dstDb, owner, inner, hash, scheme)
+		if val == nil {
+			t.Error("missing trie node")
 		}
-		rawdb.WriteTrieNode(dstDb, node, val)
+		rawdb.DeleteTrieNode(dstDb, owner, inner, hash, scheme)
+		if err := checkStateConsistency(dstDb, srcRoot); err == nil {
+			t.Errorf("trie inconsistency not caught, missing: %v", path)
+		}
+		rawdb.WriteTrieNode(dstDb, owner, inner, hash, val, scheme)
 	}
 }

core/state/transient_storage.go

@@ -0,0 +1,55 @@
+// Copyright 2022 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 (
+	"github.com/ethereum/go-ethereum/common"
+)
+
+// transientStorage is a representation of EIP-1153 "Transient Storage".
+type transientStorage map[common.Address]Storage
+
+// newTransientStorage creates a new instance of a transientStorage.
+func newTransientStorage() transientStorage {
+	return make(transientStorage)
+}
+
+// Set sets the transient-storage `value` for `key` at the given `addr`.
+func (t transientStorage) Set(addr common.Address, key, value common.Hash) {
+	if _, ok := t[addr]; !ok {
+		t[addr] = make(Storage)
+	}
+	t[addr][key] = value
+}
+
+// Get gets the transient storage for `key` at the given `addr`.
+func (t transientStorage) Get(addr common.Address, key common.Hash) common.Hash {
+	val, ok := t[addr]
+	if !ok {
+		return common.Hash{}
+	}
+	return val[key]
+}
+
+// Copy does a deep copy of the transientStorage
+func (t transientStorage) Copy() transientStorage {
+	storage := make(transientStorage)
+	for key, value := range t {
+		storage[key] = value.Copy()
+	}
+	return storage
+}

core/state/trie_prefetcher.go

@@ -126,6 +126,9 @@ func (p *triePrefetcher) copy() *triePrefetcher {
 	// If the prefetcher is already a copy, duplicate the data
 	if p.fetches != nil {
 		for root, fetch := range p.fetches {
+			if fetch == nil {
+				continue
+			}
 			copy.fetches[root] = p.db.CopyTrie(fetch)
 		}
 		return copy
@@ -147,7 +150,7 @@ func (p *triePrefetcher) prefetch(owner common.Hash, root common.Hash, keys [][]
 	id := p.trieID(owner, root)
 	fetcher := p.fetchers[id]
 	if fetcher == nil {
-		fetcher = newSubfetcher(p.db, owner, root)
+		fetcher = newSubfetcher(p.db, p.root, owner, root)
 		p.fetchers[id] = fetcher
 	}
 	fetcher.schedule(keys)
@@ -203,6 +206,7 @@ func (p *triePrefetcher) trieID(owner common.Hash, root common.Hash) string {
 // the trie being worked on is retrieved from the prefetcher.
 type subfetcher struct {
 	db    Database    // Database to load trie nodes through
+	state common.Hash // Root hash of the state to prefetch
 	owner common.Hash // Owner of the trie, usually account hash
 	root  common.Hash // Root hash of the trie to prefetch
 	trie  Trie        // Trie being populated with nodes
@@ -222,9 +226,10 @@ type subfetcher struct {
 
 // newSubfetcher creates a goroutine to prefetch state items belonging to a
 // particular root hash.
-func newSubfetcher(db Database, owner common.Hash, root common.Hash) *subfetcher {
+func newSubfetcher(db Database, state common.Hash, owner common.Hash, root common.Hash) *subfetcher {
 	sf := &subfetcher{
 		db:    db,
+		state: state,
 		owner: owner,
 		root:  root,
 		wake:  make(chan struct{}, 1),
@@ -295,7 +300,7 @@ func (sf *subfetcher) loop() {
 		}
 		sf.trie = trie
 	} else {
-		trie, err := sf.db.OpenStorageTrie(sf.owner, sf.root)
+		trie, err := sf.db.OpenStorageTrie(sf.state, sf.owner, sf.root)
 		if err != nil {
 			log.Warn("Trie prefetcher failed opening trie", "root", sf.root, "err", err)
 			return
@@ -331,11 +336,7 @@ func (sf *subfetcher) loop() {
 					if _, ok := sf.seen[string(task)]; ok {
 						sf.dups++
 					} else {
-						if len(task) == len(common.Address{}) {
-							sf.trie.TryGetAccount(task)
-						} else {
-							sf.trie.TryGet(task)
-						}
+						sf.trie.TryGet(task)
 						sf.seen[string(task)] = struct{}{}
 					}
 				}