all: activate pbss as experimental feature (#26274)

* all: activate pbss

* core/rawdb: fix compilation error

* cma, core, eth, les, trie: address comments

* cmd, core, eth, trie: polish code

* core, cmd, eth: address comments

* cmd, core, eth, les, light, tests: address comment

* cmd/utils: shorten log message

* trie/triedb/pathdb: limit node buffer size to 1gb

* cmd/utils: fix opening non-existing db

* cmd/utils: rename flag name

* cmd, core: group chain history flags and fix tests

* core, eth, trie: fix memory leak in snapshot generation

* cmd, eth, internal: deprecate flags

* all: enable state tests for pathdb, fixes

* cmd, core: polish code

* trie/triedb/pathdb: limit the node buffer size to 256mb

---------

Co-authored-by: Martin Holst Swende <martin@swende.se>
Co-authored-by: Péter Szilágyi <peterke@gmail.com>

cmd/evm/blockrunner.go

@@ -22,6 +22,7 @@ import (
 	"fmt"
 	"os"
 
+	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/eth/tracers/logger"
 	"github.com/ethereum/go-ethereum/log"
@@ -64,7 +65,7 @@ func blockTestCmd(ctx *cli.Context) error {
 		return err
 	}
 	for i, test := range tests {
-		if err := test.Run(false, tracer); err != nil {
+		if err := test.Run(false, rawdb.HashScheme, tracer); err != nil {
 			return fmt.Errorf("test %v: %w", i, err)
 		}
 	}

cmd/evm/runner.go

@@ -42,6 +42,7 @@ import (
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
 	"github.com/urfave/cli/v2"
 )
 
@@ -142,12 +143,23 @@ func runCmd(ctx *cli.Context) error {
 		gen := readGenesis(ctx.String(GenesisFlag.Name))
 		genesisConfig = gen
 		db := rawdb.NewMemoryDatabase()
-		genesis := gen.MustCommit(db)
+		triedb := trie.NewDatabase(db, &trie.Config{
-		sdb := state.NewDatabaseWithConfig(db, &trie.Config{Preimages: preimages})
+			Preimages: preimages,
+			HashDB:    hashdb.Defaults,
+		})
+		defer triedb.Close()
+		genesis := gen.MustCommit(db, triedb)
+		sdb := state.NewDatabaseWithNodeDB(db, triedb)
 		statedb, _ = state.New(genesis.Root(), sdb, nil)
 		chainConfig = gen.Config
 	} else {
-		sdb := state.NewDatabaseWithConfig(rawdb.NewMemoryDatabase(), &trie.Config{Preimages: preimages})
+		db := rawdb.NewMemoryDatabase()
+		triedb := trie.NewDatabase(db, &trie.Config{
+			Preimages: preimages,
+			HashDB:    hashdb.Defaults,
+		})
+		defer triedb.Close()
+		sdb := state.NewDatabaseWithNodeDB(db, triedb)
 		statedb, _ = state.New(types.EmptyRootHash, sdb, nil)
 		genesisConfig = new(core.Genesis)
 	}

cmd/evm/staterunner.go

@@ -23,7 +23,9 @@ import (
 	"os"
 
 	"github.com/ethereum/go-ethereum/common"
+	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/state"
+	"github.com/ethereum/go-ethereum/core/state/snapshot"
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/eth/tracers/logger"
 	"github.com/ethereum/go-ethereum/log"
@@ -104,25 +106,22 @@ func runStateTest(fname string, cfg vm.Config, jsonOut, dump bool) error {
 		for _, st := range test.Subtests() {
 			// Run the test and aggregate the result
 			result := &StatetestResult{Name: key, Fork: st.Fork, Pass: true}
-			_, s, err := test.Run(st, cfg, false)
+			test.Run(st, cfg, false, rawdb.HashScheme, func(err error, snaps *snapshot.Tree, state *state.StateDB) {
-			// print state root for evmlab tracing
+				if err != nil {
-			if s != nil {
+					// Test failed, mark as so and dump any state to aid debugging
-				root := s.IntermediateRoot(false)
+					result.Pass, result.Error = false, err.Error()
+					if dump {
+						dump := state.RawDump(nil)
+						result.State = &dump
+					}
+				} else {
+					root := state.IntermediateRoot(false)
 					result.Root = &root
 					if jsonOut {
 						fmt.Fprintf(os.Stderr, "{\"stateRoot\": \"%#x\"}\n", root)
 					}
 				}
-			if err != nil {
+			})
-				// Test failed, mark as so and dump any state to aid debugging
-				result.Pass, result.Error = false, err.Error()
-				if dump && s != nil {
-					s, _ = state.New(*result.Root, s.Database(), nil)
-					dump := s.RawDump(nil)
-					result.State = &dump
-				}
-			}
-
 			results = append(results, *result)
 		}
 	}

cmd/geth/chaincmd.go

@@ -39,7 +39,6 @@ import (
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/metrics"
 	"github.com/ethereum/go-ethereum/node"
-	"github.com/ethereum/go-ethereum/trie"
 	"github.com/urfave/cli/v2"
 )
 
@@ -49,7 +48,10 @@ var (
 		Name:      "init",
 		Usage:     "Bootstrap and initialize a new genesis block",
 		ArgsUsage: "<genesisPath>",
-		Flags:     flags.Merge([]cli.Flag{utils.CachePreimagesFlag}, utils.DatabasePathFlags),
+		Flags: flags.Merge([]cli.Flag{
+			utils.CachePreimagesFlag,
+			utils.StateSchemeFlag,
+		}, utils.DatabasePathFlags),
 		Description: `
 The init command initializes a new genesis block and definition for the network.
 This is a destructive action and changes the network in which you will be
@@ -94,6 +96,9 @@ if one is set.  Otherwise it prints the genesis from the datadir.`,
 			utils.MetricsInfluxDBBucketFlag,
 			utils.MetricsInfluxDBOrganizationFlag,
 			utils.TxLookupLimitFlag,
+			utils.TransactionHistoryFlag,
+			utils.StateSchemeFlag,
+			utils.StateHistoryFlag,
 		}, utils.DatabasePathFlags),
 		Description: `
 The import command imports blocks from an RLP-encoded form. The form can be one file
@@ -110,6 +115,7 @@ processing will proceed even if an individual RLP-file import failure occurs.`,
 		Flags: flags.Merge([]cli.Flag{
 			utils.CacheFlag,
 			utils.SyncModeFlag,
+			utils.StateSchemeFlag,
 		}, utils.DatabasePathFlags),
 		Description: `
 Requires a first argument of the file to write to.
@@ -159,6 +165,7 @@ It's deprecated, please use "geth db export" instead.
 			utils.IncludeIncompletesFlag,
 			utils.StartKeyFlag,
 			utils.DumpLimitFlag,
+			utils.StateSchemeFlag,
 		}, utils.DatabasePathFlags),
 		Description: `
 This command dumps out the state for a given block (or latest, if none provided).
@@ -195,14 +202,15 @@ func initGenesis(ctx *cli.Context) error {
 		if err != nil {
 			utils.Fatalf("Failed to open database: %v", err)
 		}
-		triedb := trie.NewDatabaseWithConfig(chaindb, &trie.Config{
+		defer chaindb.Close()
-			Preimages: ctx.Bool(utils.CachePreimagesFlag.Name),
+
-		})
+		triedb := utils.MakeTrieDatabase(ctx, chaindb, ctx.Bool(utils.CachePreimagesFlag.Name), false)
+		defer triedb.Close()
+
 		_, hash, err := core.SetupGenesisBlock(chaindb, triedb, genesis)
 		if err != nil {
 			utils.Fatalf("Failed to write genesis block: %v", err)
 		}
-		chaindb.Close()
 		log.Info("Successfully wrote genesis state", "database", name, "hash", hash)
 	}
 	return nil
@@ -241,7 +249,7 @@ func dumpGenesis(ctx *cli.Context) error {
 	if ctx.IsSet(utils.DataDirFlag.Name) {
 		utils.Fatalf("no existing datadir at %s", stack.Config().DataDir)
 	}
-	utils.Fatalf("no network preset provided.  no exisiting genesis in the default datadir")
+	utils.Fatalf("no network preset provided, no existing genesis in the default datadir")
 	return nil
 }
 
@@ -465,10 +473,10 @@ func dump(ctx *cli.Context) error {
 	if err != nil {
 		return err
 	}
-	config := &trie.Config{
+	triedb := utils.MakeTrieDatabase(ctx, db, true, false) // always enable preimage lookup
-		Preimages: true, // always enable preimage lookup
+	defer triedb.Close()
-	}
+
-	state, err := state.New(root, state.NewDatabaseWithConfig(db, config), nil)
+	state, err := state.New(root, state.NewDatabaseWithNodeDB(db, triedb), nil)
 	if err != nil {
 		return err
 	}

cmd/geth/dbcmd.go

@@ -151,6 +151,7 @@ WARNING: This is a low-level operation which may cause database corruption!`,
 		ArgsUsage: "<hex-encoded state root> <hex-encoded account hash> <hex-encoded storage trie root> <hex-encoded start (optional)> <int max elements (optional)>",
 		Flags: flags.Merge([]cli.Flag{
 			utils.SyncModeFlag,
+			utils.StateSchemeFlag,
 		}, utils.NetworkFlags, utils.DatabasePathFlags),
 		Description: "This command looks up the specified database key from the database.",
 	}
@@ -482,6 +483,9 @@ func dbDumpTrie(ctx *cli.Context) error {
 	db := utils.MakeChainDatabase(ctx, stack, true)
 	defer db.Close()
 
+	triedb := utils.MakeTrieDatabase(ctx, db, false, true)
+	defer triedb.Close()
+
 	var (
 		state   []byte
 		storage []byte
@@ -515,7 +519,7 @@ func dbDumpTrie(ctx *cli.Context) error {
 		}
 	}
 	id := trie.StorageTrieID(common.BytesToHash(state), common.BytesToHash(account), common.BytesToHash(storage))
-	theTrie, err := trie.New(id, trie.NewDatabase(db))
+	theTrie, err := trie.New(id, triedb)
 	if err != nil {
 		return err
 	}

cmd/geth/genesis_test.go

@@ -176,12 +176,12 @@ func TestCustomBackend(t *testing.T) {
 		{ // Can't start pebble on top of leveldb
 			initArgs:   []string{"--db.engine", "leveldb"},
 			execArgs:   []string{"--db.engine", "pebble"},
-			execExpect: `Fatal: Failed to register the Ethereum service: db.engine choice was pebble but found pre-existing leveldb database in specified data directory`,
+			execExpect: `Fatal: Could not open database: db.engine choice was pebble but found pre-existing leveldb database in specified data directory`,
 		},
 		{ // Can't start leveldb on top of pebble
 			initArgs:   []string{"--db.engine", "pebble"},
 			execArgs:   []string{"--db.engine", "leveldb"},
-			execExpect: `Fatal: Failed to register the Ethereum service: db.engine choice was leveldb but found pre-existing pebble database in specified data directory`,
+			execExpect: `Fatal: Could not open database: db.engine choice was leveldb but found pre-existing pebble database in specified data directory`,
 		},
 		{ // Reject invalid backend choice
 			initArgs:   []string{"--db.engine", "mssql"},

cmd/geth/main.go

@@ -88,6 +88,9 @@ var (
 		utils.GCModeFlag,
 		utils.SnapshotFlag,
 		utils.TxLookupLimitFlag,
+		utils.TransactionHistoryFlag,
+		utils.StateSchemeFlag,
+		utils.StateHistoryFlag,
 		utils.LightServeFlag,
 		utils.LightIngressFlag,
 		utils.LightEgressFlag,

cmd/geth/snapshot.go

@@ -61,10 +61,7 @@ two version states are available: genesis and the specific one.
 
 The default pruning target is the HEAD-127 state.
 
-WARNING: It's necessary to delete the trie clean cache after the pruning.
+WARNING: it's only supported in hash mode(--state.scheme=hash)".
-If you specify another directory for the trie clean cache via "--cache.trie.journal"
-during the use of Geth, please also specify it here for correct deletion. Otherwise
-the trie clean cache with default directory will be deleted.
 `,
 			},
 			{
@@ -72,7 +69,9 @@ the trie clean cache with default directory will be deleted.
 				Usage:     "Recalculate state hash based on the snapshot for verification",
 				ArgsUsage: "<root>",
 				Action:    verifyState,
-				Flags:     flags.Merge(utils.NetworkFlags, utils.DatabasePathFlags),
+				Flags: flags.Merge([]cli.Flag{
+					utils.StateSchemeFlag,
+				}, utils.NetworkFlags, utils.DatabasePathFlags),
 				Description: `
 geth snapshot verify-state <state-root>
 will traverse the whole accounts and storages set based on the specified
@@ -107,7 +106,9 @@ information about the specified address.
 				Usage:     "Traverse the state with given root hash and perform quick verification",
 				ArgsUsage: "<root>",
 				Action:    traverseState,
-				Flags:     flags.Merge(utils.NetworkFlags, utils.DatabasePathFlags),
+				Flags: flags.Merge([]cli.Flag{
+					utils.StateSchemeFlag,
+				}, utils.NetworkFlags, utils.DatabasePathFlags),
 				Description: `
 geth snapshot traverse-state <state-root>
 will traverse the whole state from the given state root and will abort if any
@@ -122,7 +123,9 @@ It's also usable without snapshot enabled.
 				Usage:     "Traverse the state with given root hash and perform detailed verification",
 				ArgsUsage: "<root>",
 				Action:    traverseRawState,
-				Flags:     flags.Merge(utils.NetworkFlags, utils.DatabasePathFlags),
+				Flags: flags.Merge([]cli.Flag{
+					utils.StateSchemeFlag,
+				}, utils.NetworkFlags, utils.DatabasePathFlags),
 				Description: `
 geth snapshot traverse-rawstate <state-root>
 will traverse the whole state from the given root and will abort if any referenced
@@ -143,6 +146,7 @@ It's also usable without snapshot enabled.
 					utils.ExcludeStorageFlag,
 					utils.StartKeyFlag,
 					utils.DumpLimitFlag,
+					utils.StateSchemeFlag,
 				}, utils.NetworkFlags, utils.DatabasePathFlags),
 				Description: `
 This command is semantically equivalent to 'geth dump', but uses the snapshots
@@ -165,6 +169,9 @@ func pruneState(ctx *cli.Context) error {
 	chaindb := utils.MakeChainDatabase(ctx, stack, false)
 	defer chaindb.Close()
 
+	if rawdb.ReadStateScheme(chaindb) != rawdb.HashScheme {
+		log.Crit("Offline pruning is not required for path scheme")
+	}
 	prunerconfig := pruner.Config{
 		Datadir:   stack.ResolvePath(""),
 		BloomSize: ctx.Uint64(utils.BloomFilterSizeFlag.Name),
@@ -205,13 +212,16 @@ func verifyState(ctx *cli.Context) error {
 		log.Error("Failed to load head block")
 		return errors.New("no head block")
 	}
-	snapconfig := snapshot.Config{
+	triedb := utils.MakeTrieDatabase(ctx, chaindb, false, true)
+	defer triedb.Close()
+
+	snapConfig := snapshot.Config{
 		CacheSize:  256,
 		Recovery:   false,
 		NoBuild:    true,
 		AsyncBuild: false,
 	}
-	snaptree, err := snapshot.New(snapconfig, chaindb, trie.NewDatabase(chaindb), headBlock.Root())
+	snaptree, err := snapshot.New(snapConfig, chaindb, triedb, headBlock.Root())
 	if err != nil {
 		log.Error("Failed to open snapshot tree", "err", err)
 		return err
@@ -253,6 +263,11 @@ func traverseState(ctx *cli.Context) error {
 	defer stack.Close()
 
 	chaindb := utils.MakeChainDatabase(ctx, stack, true)
+	defer chaindb.Close()
+
+	triedb := utils.MakeTrieDatabase(ctx, chaindb, false, true)
+	defer triedb.Close()
+
 	headBlock := rawdb.ReadHeadBlock(chaindb)
 	if headBlock == nil {
 		log.Error("Failed to load head block")
@@ -277,7 +292,6 @@ func traverseState(ctx *cli.Context) error {
 		root = headBlock.Root()
 		log.Info("Start traversing the state", "root", root, "number", headBlock.NumberU64())
 	}
-	triedb := trie.NewDatabase(chaindb)
 	t, err := trie.NewStateTrie(trie.StateTrieID(root), triedb)
 	if err != nil {
 		log.Error("Failed to open trie", "root", root, "err", err)
@@ -353,6 +367,11 @@ func traverseRawState(ctx *cli.Context) error {
 	defer stack.Close()
 
 	chaindb := utils.MakeChainDatabase(ctx, stack, true)
+	defer chaindb.Close()
+
+	triedb := utils.MakeTrieDatabase(ctx, chaindb, false, true)
+	defer triedb.Close()
+
 	headBlock := rawdb.ReadHeadBlock(chaindb)
 	if headBlock == nil {
 		log.Error("Failed to load head block")
@@ -377,7 +396,6 @@ func traverseRawState(ctx *cli.Context) error {
 		root = headBlock.Root()
 		log.Info("Start traversing the state", "root", root, "number", headBlock.NumberU64())
 	}
-	triedb := trie.NewDatabase(chaindb)
 	t, err := trie.NewStateTrie(trie.StateTrieID(root), triedb)
 	if err != nil {
 		log.Error("Failed to open trie", "root", root, "err", err)
@@ -398,6 +416,11 @@ func traverseRawState(ctx *cli.Context) error {
 		log.Error("Failed to open iterator", "root", root, "err", err)
 		return err
 	}
+	reader, err := triedb.Reader(root)
+	if err != nil {
+		log.Error("State is non-existent", "root", root)
+		return nil
+	}
 	for accIter.Next(true) {
 		nodes += 1
 		node := accIter.Hash()
@@ -405,7 +428,7 @@ func traverseRawState(ctx *cli.Context) error {
 		// Check the present for non-empty hash node(embedded node doesn't
 		// have their own hash).
 		if node != (common.Hash{}) {
-			blob := rawdb.ReadLegacyTrieNode(chaindb, node)
+			blob, _ := reader.Node(common.Hash{}, accIter.Path(), node)
 			if len(blob) == 0 {
 				log.Error("Missing trie node(account)", "hash", node)
 				return errors.New("missing account")
@@ -446,7 +469,7 @@ func traverseRawState(ctx *cli.Context) error {
 					// Check the presence for non-empty hash node(embedded node doesn't
 					// have their own hash).
 					if node != (common.Hash{}) {
-						blob := rawdb.ReadLegacyTrieNode(chaindb, node)
+						blob, _ := reader.Node(common.BytesToHash(accIter.LeafKey()), storageIter.Path(), node)
 						if len(blob) == 0 {
 							log.Error("Missing trie node(storage)", "hash", node)
 							return errors.New("missing storage")
@@ -506,13 +529,16 @@ func dumpState(ctx *cli.Context) error {
 	if err != nil {
 		return err
 	}
+	triedb := utils.MakeTrieDatabase(ctx, db, false, true)
+	defer triedb.Close()
+
 	snapConfig := snapshot.Config{
 		CacheSize:  256,
 		Recovery:   false,
 		NoBuild:    true,
 		AsyncBuild: false,
 	}
-	snaptree, err := snapshot.New(snapConfig, db, trie.NewDatabase(db), root)
+	snaptree, err := snapshot.New(snapConfig, db, triedb, root)
 	if err != nil {
 		return err
 	}

cmd/utils/flags.go

@@ -48,7 +48,7 @@ import (
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/crypto/kzg4844"
 	"github.com/ethereum/go-ethereum/eth"
-	ethcatalyst "github.com/ethereum/go-ethereum/eth/catalyst"
+	"github.com/ethereum/go-ethereum/eth/catalyst"
 	"github.com/ethereum/go-ethereum/eth/downloader"
 	"github.com/ethereum/go-ethereum/eth/ethconfig"
 	"github.com/ethereum/go-ethereum/eth/filters"
@@ -74,6 +74,9 @@ import (
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/rpc"
+	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 	pcsclite "github.com/gballet/go-libpcsclite"
 	gopsutil "github.com/shirou/gopsutil/mem"
 	"github.com/urfave/cli/v2"
@@ -218,30 +221,12 @@ var (
 	}
 
 	defaultSyncMode = ethconfig.Defaults.SyncMode
-	SyncModeFlag    = &flags.TextMarshalerFlag{
-		Name:     "syncmode",
-		Usage:    `Blockchain sync mode ("snap", "full" or "light")`,
-		Value:    &defaultSyncMode,
-		Category: flags.EthCategory,
-	}
-	GCModeFlag = &cli.StringFlag{
-		Name:     "gcmode",
-		Usage:    `Blockchain garbage collection mode ("full", "archive")`,
-		Value:    "full",
-		Category: flags.EthCategory,
-	}
 	SnapshotFlag    = &cli.BoolFlag{
 		Name:     "snapshot",
 		Usage:    `Enables snapshot-database mode (default = enable)`,
 		Value:    true,
 		Category: flags.EthCategory,
 	}
-	TxLookupLimitFlag = &cli.Uint64Flag{
-		Name:     "txlookuplimit",
-		Usage:    "Number of recent blocks to maintain transactions index for (default = about one year, 0 = entire chain)",
-		Value:    ethconfig.Defaults.TxLookupLimit,
-		Category: flags.EthCategory,
-	}
 	LightKDFFlag = &cli.BoolFlag{
 		Name:     "lightkdf",
 		Usage:    "Reduce key-derivation RAM & CPU usage at some expense of KDF strength",
@@ -268,6 +253,36 @@ var (
 		Usage:    "Manually specify the Verkle fork timestamp, overriding the bundled setting",
 		Category: flags.EthCategory,
 	}
+	SyncModeFlag = &flags.TextMarshalerFlag{
+		Name:     "syncmode",
+		Usage:    `Blockchain sync mode ("snap", "full" or "light")`,
+		Value:    &defaultSyncMode,
+		Category: flags.StateCategory,
+	}
+	GCModeFlag = &cli.StringFlag{
+		Name:     "gcmode",
+		Usage:    `Blockchain garbage collection mode, only relevant in state.scheme=hash ("full", "archive")`,
+		Value:    "full",
+		Category: flags.StateCategory,
+	}
+	StateSchemeFlag = &cli.StringFlag{
+		Name:     "state.scheme",
+		Usage:    "Scheme to use for storing ethereum state ('hash' or 'path')",
+		Value:    rawdb.HashScheme,
+		Category: flags.StateCategory,
+	}
+	StateHistoryFlag = &cli.Uint64Flag{
+		Name:     "history.state",
+		Usage:    "Number of recent blocks to retain state history for (default = 90,000 blocks, 0 = entire chain)",
+		Value:    ethconfig.Defaults.StateHistory,
+		Category: flags.StateCategory,
+	}
+	TransactionHistoryFlag = &cli.Uint64Flag{
+		Name:     "history.transactions",
+		Usage:    "Number of recent blocks to maintain transactions index for (default = about one year, 0 = entire chain)",
+		Value:    ethconfig.Defaults.TransactionHistory,
+		Category: flags.StateCategory,
+	}
 	// Light server and client settings
 	LightServeFlag = &cli.IntFlag{
 		Name:     "light.serve",
@@ -1624,13 +1639,8 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *ethconfig.Config) {
 	CheckExclusive(ctx, MainnetFlag, DeveloperFlag, GoerliFlag, SepoliaFlag)
 	CheckExclusive(ctx, LightServeFlag, SyncModeFlag, "light")
 	CheckExclusive(ctx, DeveloperFlag, ExternalSignerFlag) // Can't use both ephemeral unlocked and external signer
-	if ctx.String(GCModeFlag.Name) == "archive" && ctx.Uint64(TxLookupLimitFlag.Name) != 0 {
+
-		ctx.Set(TxLookupLimitFlag.Name, "0")
+	// Set configurations from CLI flags
-		log.Warn("Disable transaction unindexing for archive node")
-	}
-	if ctx.IsSet(LightServeFlag.Name) && ctx.Uint64(TxLookupLimitFlag.Name) != 0 {
-		log.Warn("LES server cannot serve old transaction status and cannot connect below les/4 protocol version if transaction lookup index is limited")
-	}
 	setEtherbase(ctx, cfg)
 	setGPO(ctx, &cfg.GPO, ctx.String(SyncModeFlag.Name) == "light")
 	setTxPool(ctx, &cfg.TxPool)
@@ -1687,8 +1697,36 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *ethconfig.Config) {
 		cfg.Preimages = true
 		log.Info("Enabling recording of key preimages since archive mode is used")
 	}
-	if ctx.IsSet(TxLookupLimitFlag.Name) {
+	if ctx.IsSet(StateHistoryFlag.Name) {
-		cfg.TxLookupLimit = ctx.Uint64(TxLookupLimitFlag.Name)
+		cfg.StateHistory = ctx.Uint64(StateHistoryFlag.Name)
+	}
+	// Parse state scheme, abort the process if it's not compatible.
+	chaindb := tryMakeReadOnlyDatabase(ctx, stack)
+	scheme, err := ParseStateScheme(ctx, chaindb)
+	chaindb.Close()
+	if err != nil {
+		Fatalf("%v", err)
+	}
+	cfg.StateScheme = scheme
+
+	// Parse transaction history flag, if user is still using legacy config
+	// file with 'TxLookupLimit' configured, copy the value to 'TransactionHistory'.
+	if cfg.TransactionHistory == ethconfig.Defaults.TransactionHistory && cfg.TxLookupLimit != ethconfig.Defaults.TxLookupLimit {
+		log.Warn("The config option 'TxLookupLimit' is deprecated and will be removed, please use 'TransactionHistory'")
+		cfg.TransactionHistory = cfg.TxLookupLimit
+	}
+	if ctx.IsSet(TransactionHistoryFlag.Name) {
+		cfg.TransactionHistory = ctx.Uint64(TransactionHistoryFlag.Name)
+	} else if ctx.IsSet(TxLookupLimitFlag.Name) {
+		log.Warn("The flag --txlookuplimit is deprecated and will be removed, please use --history.transactions")
+		cfg.TransactionHistory = ctx.Uint64(TransactionHistoryFlag.Name)
+	}
+	if ctx.String(GCModeFlag.Name) == "archive" && cfg.TransactionHistory != 0 {
+		cfg.TransactionHistory = 0
+		log.Warn("Disabled transaction unindexing for archive node")
+	}
+	if ctx.IsSet(LightServeFlag.Name) && cfg.TransactionHistory != 0 {
+		log.Warn("LES server cannot serve old transaction status and cannot connect below les/4 protocol version if transaction lookup index is limited")
 	}
 	if ctx.IsSet(CacheFlag.Name) || ctx.IsSet(CacheTrieFlag.Name) {
 		cfg.TrieCleanCache = ctx.Int(CacheFlag.Name) * ctx.Int(CacheTrieFlag.Name) / 100
@@ -1814,15 +1852,7 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *ethconfig.Config) {
 		// Create a new developer genesis block or reuse existing one
 		cfg.Genesis = core.DeveloperGenesisBlock(ctx.Uint64(DeveloperGasLimitFlag.Name), developer.Address)
 		if ctx.IsSet(DataDirFlag.Name) {
-			// If datadir doesn't exist we need to open db in write-mode
+			chaindb := tryMakeReadOnlyDatabase(ctx, stack)
-			// so leveldb can create files.
-			readonly := true
-			if !common.FileExist(stack.ResolvePath("chaindata")) {
-				readonly = false
-			}
-			// Check if we have an already initialized chain and fall back to
-			// that if so. Otherwise we need to generate a new genesis spec.
-			chaindb := MakeChainDatabase(ctx, stack, readonly)
 			if rawdb.ReadCanonicalHash(chaindb, 0) != (common.Hash{}) {
 				cfg.Genesis = nil // fallback to db content
 			}
@@ -1930,7 +1960,7 @@ func RegisterFullSyncTester(stack *node.Node, eth *eth.Ethereum, path string) {
 	if err := rlp.DecodeBytes(rlpBlob, &block); err != nil {
 		Fatalf("Failed to decode block: %v", err)
 	}
-	ethcatalyst.RegisterFullSyncTester(stack, eth, &block)
+	catalyst.RegisterFullSyncTester(stack, eth, &block)
 	log.Info("Registered full-sync tester", "number", block.NumberU64(), "hash", block.Hash())
 }
 
@@ -2040,6 +2070,18 @@ func MakeChainDatabase(ctx *cli.Context, stack *node.Node, readonly bool) ethdb.
 	return chainDb
 }
 
+// tryMakeReadOnlyDatabase try to open the chain database in read-only mode,
+// or fallback to write mode if the database is not initialized.
+func tryMakeReadOnlyDatabase(ctx *cli.Context, stack *node.Node) ethdb.Database {
+	// If datadir doesn't exist we need to open db in write-mode
+	// so database engine can create files.
+	readonly := true
+	if !common.FileExist(stack.ResolvePath("chaindata")) {
+		readonly = false
+	}
+	return MakeChainDatabase(ctx, stack, readonly)
+}
+
 func IsNetworkPreset(ctx *cli.Context) bool {
 	for _, flag := range NetworkFlags {
 		bFlag, _ := flag.(*cli.BoolFlag)
@@ -2106,6 +2148,10 @@ func MakeChain(ctx *cli.Context, stack *node.Node, readonly bool) (*core.BlockCh
 	if gcmode := ctx.String(GCModeFlag.Name); gcmode != "full" && gcmode != "archive" {
 		Fatalf("--%s must be either 'full' or 'archive'", GCModeFlag.Name)
 	}
+	scheme, err := ParseStateScheme(ctx, chainDb)
+	if err != nil {
+		Fatalf("%v", err)
+	}
 	cache := &core.CacheConfig{
 		TrieCleanLimit:      ethconfig.Defaults.TrieCleanCache,
 		TrieCleanNoPrefetch: ctx.Bool(CacheNoPrefetchFlag.Name),
@@ -2114,6 +2160,8 @@ func MakeChain(ctx *cli.Context, stack *node.Node, readonly bool) (*core.BlockCh
 		TrieTimeLimit:       ethconfig.Defaults.TrieTimeout,
 		SnapshotLimit:       ethconfig.Defaults.SnapshotCache,
 		Preimages:           ctx.Bool(CachePreimagesFlag.Name),
+		StateScheme:         scheme,
+		StateHistory:        ctx.Uint64(StateHistoryFlag.Name),
 	}
 	if cache.TrieDirtyDisabled && !cache.Preimages {
 		cache.Preimages = true
@@ -2158,3 +2206,62 @@ func MakeConsolePreloads(ctx *cli.Context) []string {
 	}
 	return preloads
 }
+
+// ParseStateScheme resolves scheme identifier from CLI flag. If the provided
+// state scheme is not compatible with the one of persistent scheme, an error
+// will be returned.
+//
+//   - none: use the scheme consistent with persistent state, or fallback
+//     to hash-based scheme if state is empty.
+//   - hash: use hash-based scheme or error out if not compatible with
+//     persistent state scheme.
+//   - path: use path-based scheme or error out if not compatible with
+//     persistent state scheme.
+func ParseStateScheme(ctx *cli.Context, disk ethdb.Database) (string, error) {
+	// If state scheme is not specified, use the scheme consistent
+	// with persistent state, or fallback to hash mode if database
+	// is empty.
+	stored := rawdb.ReadStateScheme(disk)
+	if !ctx.IsSet(StateSchemeFlag.Name) {
+		if stored == "" {
+			// use default scheme for empty database, flip it when
+			// path mode is chosen as default
+			log.Info("State schema set to default", "scheme", "hash")
+			return rawdb.HashScheme, nil
+		}
+		log.Info("State scheme set to already existing", "scheme", stored)
+		return stored, nil // reuse scheme of persistent scheme
+	}
+	// If state scheme is specified, ensure it's compatible with
+	// persistent state.
+	scheme := ctx.String(StateSchemeFlag.Name)
+	if stored == "" || scheme == stored {
+		log.Info("State scheme set by user", "scheme", scheme)
+		return scheme, nil
+	}
+	return "", fmt.Errorf("incompatible state scheme, stored: %s, provided: %s", stored, scheme)
+}
+
+// MakeTrieDatabase constructs a trie database based on the configured scheme.
+func MakeTrieDatabase(ctx *cli.Context, disk ethdb.Database, preimage bool, readOnly bool) *trie.Database {
+	config := &trie.Config{
+		Preimages: preimage,
+	}
+	scheme, err := ParseStateScheme(ctx, disk)
+	if err != nil {
+		Fatalf("%v", err)
+	}
+	if scheme == rawdb.HashScheme {
+		// Read-only mode is not implemented in hash mode,
+		// ignore the parameter silently. TODO(rjl493456442)
+		// please config it if read mode is implemented.
+		config.HashDB = hashdb.Defaults
+		return trie.NewDatabase(disk, config)
+	}
+	if readOnly {
+		config.PathDB = pathdb.ReadOnly
+	} else {
+		config.PathDB = pathdb.Defaults
+	}
+	return trie.NewDatabase(disk, config)
+}

cmd/utils/flags_legacy.go

@@ -19,6 +19,7 @@ package utils
 import (
 	"fmt"
 
+	"github.com/ethereum/go-ethereum/eth/ethconfig"
 	"github.com/ethereum/go-ethereum/internal/flags"
 	"github.com/urfave/cli/v2"
 )
@@ -37,6 +38,7 @@ var DeprecatedFlags = []cli.Flag{
 	CacheTrieJournalFlag,
 	CacheTrieRejournalFlag,
 	LegacyDiscoveryV5Flag,
+	TxLookupLimitFlag,
 }
 
 var (
@@ -68,6 +70,13 @@ var (
 		Usage:    "Enables the experimental RLPx V5 (Topic Discovery) mechanism (deprecated, use --discv5 instead)",
 		Category: flags.DeprecatedCategory,
 	}
+	// Deprecated August 2023
+	TxLookupLimitFlag = &cli.Uint64Flag{
+		Name:     "txlookuplimit",
+		Usage:    "Number of recent blocks to maintain transactions index for (default = about one year, 0 = entire chain) (deprecated, use history.transactions instead)",
+		Value:    ethconfig.Defaults.TransactionHistory,
+		Category: flags.DeprecatedCategory,
+	}
 )
 
 // showDeprecated displays deprecated flags that will be soon removed from the codebase.

core/block_validator_test.go

@@ -35,6 +35,11 @@ import (
 
 // Tests that simple header verification works, for both good and bad blocks.
 func TestHeaderVerification(t *testing.T) {
+	testHeaderVerification(t, rawdb.HashScheme)
+	testHeaderVerification(t, rawdb.PathScheme)
+}
+
+func testHeaderVerification(t *testing.T, scheme string) {
 	// Create a simple chain to verify
 	var (
 		gspec        = &Genesis{Config: params.TestChainConfig}
@@ -45,7 +50,7 @@ func TestHeaderVerification(t *testing.T) {
 		headers[i] = block.Header()
 	}
 	// Run the header checker for blocks one-by-one, checking for both valid and invalid nonces
-	chain, _ := NewBlockChain(rawdb.NewMemoryDatabase(), nil, gspec, nil, ethash.NewFaker(), vm.Config{}, nil, nil)
+	chain, _ := NewBlockChain(rawdb.NewMemoryDatabase(), DefaultCacheConfigWithScheme(scheme), gspec, nil, ethash.NewFaker(), vm.Config{}, nil, nil)
 	defer chain.Stop()
 
 	for i := 0; i < len(blocks); i++ {

core/blockchain.go

@@ -48,6 +48,8 @@ import (
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 	"golang.org/x/exp/slices"
 )
 
@@ -128,7 +130,7 @@ const (
 )
 
 // CacheConfig contains the configuration values for the trie database
-// that's resident in a blockchain.
+// and state snapshot these are resident in a blockchain.
 type CacheConfig struct {
 	TrieCleanLimit      int           // Memory allowance (MB) to use for caching trie nodes in memory
 	TrieCleanNoPrefetch bool          // Whether to disable heuristic state prefetching for followup blocks
@@ -137,11 +139,31 @@ type CacheConfig struct {
 	TrieTimeLimit       time.Duration // Time limit after which to flush the current in-memory trie to disk
 	SnapshotLimit       int           // Memory allowance (MB) to use for caching snapshot entries in memory
 	Preimages           bool          // Whether to store preimage of trie key to the disk
+	StateHistory        uint64        // Number of blocks from head whose state histories are reserved.
+	StateScheme         string        // Scheme used to store ethereum states and merkle tree nodes on top
 
 	SnapshotNoBuild bool // Whether the background generation is allowed
 	SnapshotWait    bool // Wait for snapshot construction on startup. TODO(karalabe): This is a dirty hack for testing, nuke it
 }
 
+// triedbConfig derives the configures for trie database.
+func (c *CacheConfig) triedbConfig() *trie.Config {
+	config := &trie.Config{Preimages: c.Preimages}
+	if c.StateScheme == rawdb.HashScheme {
+		config.HashDB = &hashdb.Config{
+			CleanCacheSize: c.TrieCleanLimit * 1024 * 1024,
+		}
+	}
+	if c.StateScheme == rawdb.PathScheme {
+		config.PathDB = &pathdb.Config{
+			StateHistory:   c.StateHistory,
+			CleanCacheSize: c.TrieCleanLimit * 1024 * 1024,
+			DirtyCacheSize: c.TrieDirtyLimit * 1024 * 1024,
+		}
+	}
+	return config
+}
+
 // defaultCacheConfig are the default caching values if none are specified by the
 // user (also used during testing).
 var defaultCacheConfig = &CacheConfig{
@@ -150,6 +172,15 @@ var defaultCacheConfig = &CacheConfig{
 	TrieTimeLimit:  5 * time.Minute,
 	SnapshotLimit:  256,
 	SnapshotWait:   true,
+	StateScheme:    rawdb.HashScheme,
+}
+
+// DefaultCacheConfigWithScheme returns a deep copied default cache config with
+// a provided trie node scheme.
+func DefaultCacheConfigWithScheme(scheme string) *CacheConfig {
+	config := *defaultCacheConfig
+	config.StateScheme = scheme
+	return &config
 }
 
 // BlockChain represents the canonical chain given a database with a genesis
@@ -235,10 +266,8 @@ func NewBlockChain(db ethdb.Database, cacheConfig *CacheConfig, genesis *Genesis
 		cacheConfig = defaultCacheConfig
 	}
 	// Open trie database with provided config
-	triedb := trie.NewDatabaseWithConfig(db, &trie.Config{
+	triedb := trie.NewDatabase(db, cacheConfig.triedbConfig())
-		Cache:     cacheConfig.TrieCleanLimit,
+
-		Preimages: cacheConfig.Preimages,
-	})
 	// Setup the genesis block, commit the provided genesis specification
 	// to database if the genesis block is not present yet, or load the
 	// stored one from database.
@@ -612,6 +641,25 @@ func (bc *BlockChain) setHeadBeyondRoot(head uint64, time uint64, root common.Ha
 	pivot := rawdb.ReadLastPivotNumber(bc.db)
 	frozen, _ := bc.db.Ancients()
 
+	// resetState resets the persistent state to genesis if it's not available.
+	resetState := func() {
+		// Short circuit if the genesis state is already present.
+		if bc.HasState(bc.genesisBlock.Root()) {
+			return
+		}
+		// Reset the state database to empty for committing genesis state.
+		// Note, it should only happen in path-based scheme and Reset function
+		// is also only call-able in this mode.
+		if bc.triedb.Scheme() == rawdb.PathScheme {
+			if err := bc.triedb.Reset(types.EmptyRootHash); err != nil {
+				log.Crit("Failed to clean state", "err", err) // Shouldn't happen
+			}
+		}
+		// Write genesis state into database.
+		if err := CommitGenesisState(bc.db, bc.triedb, bc.genesisBlock.Hash()); err != nil {
+			log.Crit("Failed to commit genesis state", "err", err)
+		}
+	}
 	updateFn := func(db ethdb.KeyValueWriter, header *types.Header) (*types.Header, bool) {
 		// Rewind the blockchain, ensuring we don't end up with a stateless head
 		// block. Note, depth equality is permitted to allow using SetHead as a
@@ -621,6 +669,7 @@ func (bc *BlockChain) setHeadBeyondRoot(head uint64, time uint64, root common.Ha
 			if newHeadBlock == nil {
 				log.Error("Gap in the chain, rewinding to genesis", "number", header.Number, "hash", header.Hash())
 				newHeadBlock = bc.genesisBlock
+				resetState()
 			} else {
 				// Block exists, keep rewinding until we find one with state,
 				// keeping rewinding until we exceed the optional threshold
@@ -632,7 +681,7 @@ func (bc *BlockChain) setHeadBeyondRoot(head uint64, time uint64, root common.Ha
 					if root != (common.Hash{}) && !beyondRoot && newHeadBlock.Root() == root {
 						beyondRoot, rootNumber = true, newHeadBlock.NumberU64()
 					}
-					if !bc.HasState(newHeadBlock.Root()) {
+					if !bc.HasState(newHeadBlock.Root()) && !bc.stateRecoverable(newHeadBlock.Root()) {
 						log.Trace("Block state missing, rewinding further", "number", newHeadBlock.NumberU64(), "hash", newHeadBlock.Hash())
 						if pivot == nil || newHeadBlock.NumberU64() > *pivot {
 							parent := bc.GetBlock(newHeadBlock.ParentHash(), newHeadBlock.NumberU64()-1)
@@ -649,16 +698,12 @@ func (bc *BlockChain) setHeadBeyondRoot(head uint64, time uint64, root common.Ha
 					}
 					if beyondRoot || newHeadBlock.NumberU64() == 0 {
 						if newHeadBlock.NumberU64() == 0 {
-							// Recommit the genesis state into disk in case the rewinding destination
+							resetState()
-							// is genesis block and the relevant state is gone. In the future this
+						} else if !bc.HasState(newHeadBlock.Root()) {
-							// rewinding destination can be the earliest block stored in the chain
+							// Rewind to a block with recoverable state. If the state is
-							// if the historical chain pruning is enabled. In that case the logic
+							// missing, run the state recovery here.
-							// needs to be improved here.
+							if err := bc.triedb.Recover(newHeadBlock.Root()); err != nil {
-							if !bc.HasState(bc.genesisBlock.Root()) {
+								log.Crit("Failed to rollback state", "err", err) // Shouldn't happen
-								if err := CommitGenesisState(bc.db, bc.triedb, bc.genesisBlock.Hash()); err != nil {
-									log.Crit("Failed to commit genesis state", "err", err)
-								}
-								log.Debug("Recommitted genesis state to disk")
 							}
 						}
 						log.Debug("Rewound to block with state", "number", newHeadBlock.NumberU64(), "hash", newHeadBlock.Hash())
@@ -772,7 +817,13 @@ func (bc *BlockChain) SnapSyncCommitHead(hash common.Hash) error {
 	if block == nil {
 		return fmt.Errorf("non existent block [%x..]", hash[:4])
 	}
+	// Reset the trie database with the fresh snap synced state.
 	root := block.Root()
+	if bc.triedb.Scheme() == rawdb.PathScheme {
+		if err := bc.triedb.Reset(root); err != nil {
+			return err
+		}
+	}
 	if !bc.HasState(root) {
 		return fmt.Errorf("non existent state [%x..]", root[:4])
 	}
@@ -937,7 +988,12 @@ func (bc *BlockChain) Stop() {
 			log.Error("Failed to journal state snapshot", "err", err)
 		}
 	}
-
+	if bc.triedb.Scheme() == rawdb.PathScheme {
+		// Ensure that the in-memory trie nodes are journaled to disk properly.
+		if err := bc.triedb.Journal(bc.CurrentBlock().Root); err != nil {
+			log.Info("Failed to journal in-memory trie nodes", "err", err)
+		}
+	} else {
 		// Ensure the state of a recent block is also stored to disk before exiting.
 		// We're writing three different states to catch different restart scenarios:
 		//  - HEAD:     So we don't need to reprocess any blocks in the general case
@@ -969,9 +1025,10 @@ func (bc *BlockChain) Stop() {
 				log.Error("Dangling trie nodes after full cleanup")
 			}
 		}
-	// Flush the collected preimages to disk
+	}
-	if err := bc.stateCache.TrieDB().Close(); err != nil {
+	// Close the trie database, release all the held resources as the last step.
-		log.Error("Failed to close trie db", "err", err)
+	if err := bc.triedb.Close(); err != nil {
+		log.Error("Failed to close trie database", "err", err)
 	}
 	log.Info("Blockchain stopped")
 }
@@ -1341,6 +1398,11 @@ func (bc *BlockChain) writeBlockWithState(block *types.Block, receipts []*types.
 	if err != nil {
 		return err
 	}
+	// If node is running in path mode, skip explicit gc operation
+	// which is unnecessary in this mode.
+	if bc.triedb.Scheme() == rawdb.PathScheme {
+		return nil
+	}
 	// If we're running an archive node, always flush
 	if bc.cacheConfig.TrieDirtyDisabled {
 		return bc.triedb.Commit(root, false)
@@ -1349,8 +1411,8 @@ func (bc *BlockChain) writeBlockWithState(block *types.Block, receipts []*types.
 	bc.triedb.Reference(root, common.Hash{}) // metadata reference to keep trie alive
 	bc.triegc.Push(root, -int64(block.NumberU64()))
 
-	current := block.NumberU64()
 	// Flush limits are not considered for the first TriesInMemory blocks.
+	current := block.NumberU64()
 	if current <= TriesInMemory {
 		return nil
 	}
@@ -1936,6 +1998,12 @@ func (bc *BlockChain) insertSideChain(block *types.Block, it *insertIterator) (i
 	)
 	parent := it.previous()
 	for parent != nil && !bc.HasState(parent.Root) {
+		if bc.stateRecoverable(parent.Root) {
+			if err := bc.triedb.Recover(parent.Root); err != nil {
+				return 0, err
+			}
+			break
+		}
 		hashes = append(hashes, parent.Hash())
 		numbers = append(numbers, parent.Number.Uint64())
 
@@ -1992,6 +2060,12 @@ func (bc *BlockChain) recoverAncestors(block *types.Block) (common.Hash, error)
 		parent  = block
 	)
 	for parent != nil && !bc.HasState(parent.Root()) {
+		if bc.stateRecoverable(parent.Root()) {
+			if err := bc.triedb.Recover(parent.Root()); err != nil {
+				return common.Hash{}, err
+			}
+			break
+		}
 		hashes = append(hashes, parent.Hash())
 		numbers = append(numbers, parent.NumberU64())
 		parent = bc.GetBlock(parent.ParentHash(), parent.NumberU64()-1)
@@ -2393,6 +2467,7 @@ func (bc *BlockChain) maintainTxIndex() {
 		return
 	}
 	defer sub.Unsubscribe()
+	log.Info("Initialized transaction indexer", "limit", bc.TxLookupLimit())
 
 	for {
 		select {

core/blockchain_reader.go

@@ -293,10 +293,16 @@ func (bc *BlockChain) HasBlockAndState(hash common.Hash, number uint64) bool {
 	return bc.HasState(block.Root())
 }
 
-// TrieNode retrieves a blob of data associated with a trie node
+// stateRecoverable checks if the specified state is recoverable.
-// either from ephemeral in-memory cache, or from persistent storage.
+// Note, this function assumes the state is not present, because
-func (bc *BlockChain) TrieNode(hash common.Hash) ([]byte, error) {
+// state is not treated as recoverable if it's available, thus
-	return bc.stateCache.TrieDB().Node(hash)
+// false will be returned in this case.
+func (bc *BlockChain) stateRecoverable(root common.Hash) bool {
+	if bc.triedb.Scheme() == rawdb.HashScheme {
+		return false
+	}
+	result, _ := bc.triedb.Recoverable(root)
+	return result
 }
 
 // ContractCodeWithPrefix retrieves a blob of data associated with a contract

core/blockchain_repair_test.go

@@ -22,6 +22,7 @@ package core
 
 import (
 	"math/big"
+	"path"
 	"testing"
 	"time"
 
@@ -1749,16 +1750,23 @@ func testLongReorgedSnapSyncingDeepRepair(t *testing.T, snapshots bool) {
 }
 
 func testRepair(t *testing.T, tt *rewindTest, snapshots bool) {
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
+		testRepairWithScheme(t, tt, snapshots, scheme)
+	}
+}
+
+func testRepairWithScheme(t *testing.T, tt *rewindTest, snapshots bool, scheme string) {
 	// It's hard to follow the test case, visualize the input
 	//log.Root().SetHandler(log.LvlFilterHandler(log.LvlTrace, log.StreamHandler(os.Stderr, log.TerminalFormat(true))))
 	// fmt.Println(tt.dump(true))
 
 	// Create a temporary persistent database
 	datadir := t.TempDir()
+	ancient := path.Join(datadir, "ancient")
 
 	db, err := rawdb.Open(rawdb.OpenOptions{
 		Directory:         datadir,
-		AncientsDirectory: datadir,
+		AncientsDirectory: ancient,
 	})
 	if err != nil {
 		t.Fatalf("Failed to create persistent database: %v", err)
@@ -1777,6 +1785,7 @@ func testRepair(t *testing.T, tt *rewindTest, snapshots bool) {
 			TrieDirtyLimit: 256,
 			TrieTimeLimit:  5 * time.Minute,
 			SnapshotLimit:  0, // Disable snapshot by default
+			StateScheme:    scheme,
 		}
 	)
 	defer engine.Close()
@@ -1806,7 +1815,9 @@ func testRepair(t *testing.T, tt *rewindTest, snapshots bool) {
 		t.Fatalf("Failed to import canonical chain start: %v", err)
 	}
 	if tt.commitBlock > 0 {
-		chain.stateCache.TrieDB().Commit(canonblocks[tt.commitBlock-1].Root(), false)
+		if err := chain.triedb.Commit(canonblocks[tt.commitBlock-1].Root(), false); err != nil {
+			t.Fatalf("Failed to flush trie state: %v", err)
+		}
 		if snapshots {
 			if err := chain.snaps.Cap(canonblocks[tt.commitBlock-1].Root(), 0); err != nil {
 				t.Fatalf("Failed to flatten snapshots: %v", err)
@@ -1828,21 +1839,21 @@ func testRepair(t *testing.T, tt *rewindTest, snapshots bool) {
 		rawdb.WriteLastPivotNumber(db, *tt.pivotBlock)
 	}
 	// Pull the plug on the database, simulating a hard crash
+	chain.triedb.Close()
 	db.Close()
 	chain.stopWithoutSaving()
 
 	// Start a new blockchain back up and see where the repair leads us
 	db, err = rawdb.Open(rawdb.OpenOptions{
 		Directory:         datadir,
-		AncientsDirectory: datadir,
+		AncientsDirectory: ancient,
 	})
-
 	if err != nil {
 		t.Fatalf("Failed to reopen persistent database: %v", err)
 	}
 	defer db.Close()
 
-	newChain, err := NewBlockChain(db, nil, gspec, nil, engine, vm.Config{}, nil, nil)
+	newChain, err := NewBlockChain(db, config, gspec, nil, engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
@@ -1885,17 +1896,22 @@ func testRepair(t *testing.T, tt *rewindTest, snapshots bool) {
 // In this case the snapshot layer of B3 is not created because of existent
 // state.
 func TestIssue23496(t *testing.T) {
+	testIssue23496(t, rawdb.HashScheme)
+	testIssue23496(t, rawdb.PathScheme)
+}
+
+func testIssue23496(t *testing.T, scheme string) {
 	// It's hard to follow the test case, visualize the input
 	//log.Root().SetHandler(log.LvlFilterHandler(log.LvlTrace, log.StreamHandler(os.Stderr, log.TerminalFormat(true))))
 
 	// Create a temporary persistent database
 	datadir := t.TempDir()
+	ancient := path.Join(datadir, "ancient")
 
 	db, err := rawdb.Open(rawdb.OpenOptions{
 		Directory:         datadir,
-		AncientsDirectory: datadir,
+		AncientsDirectory: ancient,
 	})
-
 	if err != nil {
 		t.Fatalf("Failed to create persistent database: %v", err)
 	}
@@ -1908,15 +1924,8 @@ func TestIssue23496(t *testing.T) {
 			BaseFee: big.NewInt(params.InitialBaseFee),
 		}
 		engine = ethash.NewFullFaker()
-		config = &CacheConfig{
-			TrieCleanLimit: 256,
-			TrieDirtyLimit: 256,
-			TrieTimeLimit:  5 * time.Minute,
-			SnapshotLimit:  256,
-			SnapshotWait:   true,
-		}
 	)
-	chain, err := NewBlockChain(db, config, gspec, nil, engine, vm.Config{}, nil, nil)
+	chain, err := NewBlockChain(db, DefaultCacheConfigWithScheme(scheme), gspec, nil, engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to create chain: %v", err)
 	}
@@ -1929,7 +1938,7 @@ func TestIssue23496(t *testing.T) {
 	if _, err := chain.InsertChain(blocks[:1]); err != nil {
 		t.Fatalf("Failed to import canonical chain start: %v", err)
 	}
-	chain.stateCache.TrieDB().Commit(blocks[0].Root(), false)
+	chain.triedb.Commit(blocks[0].Root(), false)
 
 	// Insert block B2 and commit the snapshot into disk
 	if _, err := chain.InsertChain(blocks[1:2]); err != nil {
@@ -1943,7 +1952,7 @@ func TestIssue23496(t *testing.T) {
 	if _, err := chain.InsertChain(blocks[2:3]); err != nil {
 		t.Fatalf("Failed to import canonical chain start: %v", err)
 	}
-	chain.stateCache.TrieDB().Commit(blocks[2].Root(), false)
+	chain.triedb.Commit(blocks[2].Root(), false)
 
 	// Insert the remaining blocks
 	if _, err := chain.InsertChain(blocks[3:]); err != nil {
@@ -1951,20 +1960,21 @@ func TestIssue23496(t *testing.T) {
 	}
 
 	// Pull the plug on the database, simulating a hard crash
+	chain.triedb.Close()
 	db.Close()
 	chain.stopWithoutSaving()
 
 	// Start a new blockchain back up and see where the repair leads us
 	db, err = rawdb.Open(rawdb.OpenOptions{
 		Directory:         datadir,
-		AncientsDirectory: datadir,
+		AncientsDirectory: ancient,
 	})
 	if err != nil {
 		t.Fatalf("Failed to reopen persistent database: %v", err)
 	}
 	defer db.Close()
 
-	chain, err = NewBlockChain(db, nil, gspec, nil, engine, vm.Config{}, nil, nil)
+	chain, err = NewBlockChain(db, DefaultCacheConfigWithScheme(scheme), gspec, nil, engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
@@ -1976,8 +1986,12 @@ func TestIssue23496(t *testing.T) {
 	if head := chain.CurrentSnapBlock(); head.Number.Uint64() != uint64(4) {
 		t.Errorf("Head fast block mismatch: have %d, want %d", head.Number, uint64(4))
 	}
-	if head := chain.CurrentBlock(); head.Number.Uint64() != uint64(1) {
+	expHead := uint64(1)
-		t.Errorf("Head block mismatch: have %d, want %d", head.Number, uint64(1))
+	if scheme == rawdb.PathScheme {
+		expHead = uint64(2)
+	}
+	if head := chain.CurrentBlock(); head.Number.Uint64() != expHead {
+		t.Errorf("Head block mismatch: have %d, want %d", head.Number, expHead)
 	}
 
 	// Reinsert B2-B4

core/blockchain_sethead_test.go

@@ -22,6 +22,7 @@ package core
 import (
 	"fmt"
 	"math/big"
+	"path"
 	"strings"
 	"testing"
 	"time"
@@ -29,9 +30,13 @@ import (
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/consensus/ethash"
 	"github.com/ethereum/go-ethereum/core/rawdb"
+	"github.com/ethereum/go-ethereum/core/state"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/params"
+	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 )
 
 // rewindTest is a test case for chain rollback upon user request.
@@ -1949,16 +1954,23 @@ func testLongReorgedSnapSyncingDeepSetHead(t *testing.T, snapshots bool) {
 }
 
 func testSetHead(t *testing.T, tt *rewindTest, snapshots bool) {
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
+		testSetHeadWithScheme(t, tt, snapshots, scheme)
+	}
+}
+
+func testSetHeadWithScheme(t *testing.T, tt *rewindTest, snapshots bool, scheme string) {
 	// It's hard to follow the test case, visualize the input
 	// log.Root().SetHandler(log.LvlFilterHandler(log.LvlTrace, log.StreamHandler(os.Stderr, log.TerminalFormat(true))))
 	// fmt.Println(tt.dump(false))
 
 	// Create a temporary persistent database
 	datadir := t.TempDir()
+	ancient := path.Join(datadir, "ancient")
 
 	db, err := rawdb.Open(rawdb.OpenOptions{
 		Directory:         datadir,
-		AncientsDirectory: datadir,
+		AncientsDirectory: ancient,
 	})
 	if err != nil {
 		t.Fatalf("Failed to create persistent database: %v", err)
@@ -1977,6 +1989,7 @@ func testSetHead(t *testing.T, tt *rewindTest, snapshots bool) {
 			TrieDirtyLimit: 256,
 			TrieTimeLimit:  5 * time.Minute,
 			SnapshotLimit:  0, // Disable snapshot
+			StateScheme:    scheme,
 		}
 	)
 	if snapshots {
@@ -2007,7 +2020,7 @@ func testSetHead(t *testing.T, tt *rewindTest, snapshots bool) {
 		t.Fatalf("Failed to import canonical chain start: %v", err)
 	}
 	if tt.commitBlock > 0 {
-		chain.stateCache.TrieDB().Commit(canonblocks[tt.commitBlock-1].Root(), false)
+		chain.triedb.Commit(canonblocks[tt.commitBlock-1].Root(), false)
 		if snapshots {
 			if err := chain.snaps.Cap(canonblocks[tt.commitBlock-1].Root(), 0); err != nil {
 				t.Fatalf("Failed to flatten snapshots: %v", err)
@@ -2017,13 +2030,17 @@ func testSetHead(t *testing.T, tt *rewindTest, snapshots bool) {
 	if _, err := chain.InsertChain(canonblocks[tt.commitBlock:]); err != nil {
 		t.Fatalf("Failed to import canonical chain tail: %v", err)
 	}
-	// Manually dereference anything not committed to not have to work with 128+ tries
+	// Reopen the trie database without persisting in-memory dirty nodes.
-	for _, block := range sideblocks {
+	chain.triedb.Close()
-		chain.stateCache.TrieDB().Dereference(block.Root())
+	dbconfig := &trie.Config{}
-	}
+	if scheme == rawdb.PathScheme {
-	for _, block := range canonblocks {
+		dbconfig.PathDB = pathdb.Defaults
-		chain.stateCache.TrieDB().Dereference(block.Root())
+	} else {
+		dbconfig.HashDB = hashdb.Defaults
 	}
+	chain.triedb = trie.NewDatabase(chain.db, dbconfig)
+	chain.stateCache = state.NewDatabaseWithNodeDB(chain.db, chain.triedb)
+
 	// Force run a freeze cycle
 	type freezer interface {
 		Freeze(threshold uint64) error

core/blockchain_snapshot_test.go

@@ -24,6 +24,7 @@ import (
 	"fmt"
 	"math/big"
 	"os"
+	"path"
 	"strings"
 	"testing"
 	"time"
@@ -39,6 +40,7 @@ import (
 
 // snapshotTestBasic wraps the common testing fields in the snapshot tests.
 type snapshotTestBasic struct {
+	scheme        string // Disk scheme used for storing trie nodes
 	chainBlocks   int    // Number of blocks to generate for the canonical chain
 	snapshotBlock uint64 // Block number of the relevant snapshot disk layer
 	commitBlock   uint64 // Block number for which to commit the state to disk
@@ -51,6 +53,7 @@ type snapshotTestBasic struct {
 
 	// share fields, set in runtime
 	datadir string
+	ancient string
 	db      ethdb.Database
 	genDb   ethdb.Database
 	engine  consensus.Engine
@@ -60,10 +63,11 @@ type snapshotTestBasic struct {
 func (basic *snapshotTestBasic) prepare(t *testing.T) (*BlockChain, []*types.Block) {
 	// Create a temporary persistent database
 	datadir := t.TempDir()
+	ancient := path.Join(datadir, "ancient")
 
 	db, err := rawdb.Open(rawdb.OpenOptions{
 		Directory:         datadir,
-		AncientsDirectory: datadir,
+		AncientsDirectory: ancient,
 	})
 	if err != nil {
 		t.Fatalf("Failed to create persistent database: %v", err)
@@ -75,13 +79,8 @@ func (basic *snapshotTestBasic) prepare(t *testing.T) (*BlockChain, []*types.Blo
 			Config:  params.AllEthashProtocolChanges,
 		}
 		engine = ethash.NewFullFaker()
-
-		// Snapshot is enabled, the first snapshot is created from the Genesis.
-		// The snapshot memory allowance is 256MB, it means no snapshot flush
-		// will happen during the block insertion.
-		cacheConfig = defaultCacheConfig
 	)
-	chain, err := NewBlockChain(db, cacheConfig, gspec, nil, engine, vm.Config{}, nil, nil)
+	chain, err := NewBlockChain(db, DefaultCacheConfigWithScheme(basic.scheme), gspec, nil, engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to create chain: %v", err)
 	}
@@ -102,7 +101,7 @@ func (basic *snapshotTestBasic) prepare(t *testing.T) (*BlockChain, []*types.Blo
 		startPoint = point
 
 		if basic.commitBlock > 0 && basic.commitBlock == point {
-			chain.stateCache.TrieDB().Commit(blocks[point-1].Root(), false)
+			chain.TrieDB().Commit(blocks[point-1].Root(), false)
 		}
 		if basic.snapshotBlock > 0 && basic.snapshotBlock == point {
 			// Flushing the entire snap tree into the disk, the
@@ -121,6 +120,7 @@ func (basic *snapshotTestBasic) prepare(t *testing.T) (*BlockChain, []*types.Blo
 
 	// Set runtime fields
 	basic.datadir = datadir
+	basic.ancient = ancient
 	basic.db = db
 	basic.genDb = genDb
 	basic.engine = engine
@@ -210,6 +210,7 @@ func (basic *snapshotTestBasic) teardown() {
 	basic.db.Close()
 	basic.genDb.Close()
 	os.RemoveAll(basic.datadir)
+	os.RemoveAll(basic.ancient)
 }
 
 // snapshotTest is a test case type for normal snapshot recovery.
@@ -226,7 +227,7 @@ func (snaptest *snapshotTest) test(t *testing.T) {
 
 	// Restart the chain normally
 	chain.Stop()
-	newchain, err := NewBlockChain(snaptest.db, nil, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
+	newchain, err := NewBlockChain(snaptest.db, DefaultCacheConfigWithScheme(snaptest.scheme), snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
@@ -235,7 +236,7 @@ func (snaptest *snapshotTest) test(t *testing.T) {
 	snaptest.verify(t, newchain, blocks)
 }
 
-// crashSnapshotTest is a test case type for innormal snapshot recovery.
+// crashSnapshotTest is a test case type for irregular snapshot recovery.
 // It can be used for testing that restart Geth after the crash.
 type crashSnapshotTest struct {
 	snapshotTestBasic
@@ -251,13 +252,13 @@ func (snaptest *crashSnapshotTest) test(t *testing.T) {
 	db := chain.db
 	db.Close()
 	chain.stopWithoutSaving()
+	chain.triedb.Close()
 
 	// Start a new blockchain back up and see where the repair leads us
 	newdb, err := rawdb.Open(rawdb.OpenOptions{
 		Directory:         snaptest.datadir,
-		AncientsDirectory: snaptest.datadir,
+		AncientsDirectory: snaptest.ancient,
 	})
-
 	if err != nil {
 		t.Fatalf("Failed to reopen persistent database: %v", err)
 	}
@@ -267,13 +268,13 @@ func (snaptest *crashSnapshotTest) test(t *testing.T) {
 	// the crash, we do restart twice here: one after the crash and one
 	// after the normal stop. It's used to ensure the broken snapshot
 	// can be detected all the time.
-	newchain, err := NewBlockChain(newdb, nil, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
+	newchain, err := NewBlockChain(newdb, DefaultCacheConfigWithScheme(snaptest.scheme), snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
 	newchain.Stop()
 
-	newchain, err = NewBlockChain(newdb, nil, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
+	newchain, err = NewBlockChain(newdb, DefaultCacheConfigWithScheme(snaptest.scheme), snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
@@ -300,7 +301,7 @@ func (snaptest *gappedSnapshotTest) test(t *testing.T) {
 
 	// Insert blocks without enabling snapshot if gapping is required.
 	chain.Stop()
-	gappedBlocks, _ := GenerateChain(params.TestChainConfig, blocks[len(blocks)-1], snaptest.engine, snaptest.genDb, snaptest.gapped, func(i int, b *BlockGen) {})
+	gappedBlocks, _ := GenerateChain(snaptest.gspec.Config, blocks[len(blocks)-1], snaptest.engine, snaptest.genDb, snaptest.gapped, func(i int, b *BlockGen) {})
 
 	// Insert a few more blocks without enabling snapshot
 	var cacheConfig = &CacheConfig{
@@ -308,6 +309,7 @@ func (snaptest *gappedSnapshotTest) test(t *testing.T) {
 		TrieDirtyLimit: 256,
 		TrieTimeLimit:  5 * time.Minute,
 		SnapshotLimit:  0,
+		StateScheme:    snaptest.scheme,
 	}
 	newchain, err := NewBlockChain(snaptest.db, cacheConfig, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
@@ -317,7 +319,7 @@ func (snaptest *gappedSnapshotTest) test(t *testing.T) {
 	newchain.Stop()
 
 	// Restart the chain with enabling the snapshot
-	newchain, err = NewBlockChain(snaptest.db, nil, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
+	newchain, err = NewBlockChain(snaptest.db, DefaultCacheConfigWithScheme(snaptest.scheme), snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
@@ -345,7 +347,7 @@ func (snaptest *setHeadSnapshotTest) test(t *testing.T) {
 	chain.SetHead(snaptest.setHead)
 	chain.Stop()
 
-	newchain, err := NewBlockChain(snaptest.db, nil, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
+	newchain, err := NewBlockChain(snaptest.db, DefaultCacheConfigWithScheme(snaptest.scheme), snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
@@ -379,22 +381,24 @@ func (snaptest *wipeCrashSnapshotTest) test(t *testing.T) {
 		TrieDirtyLimit: 256,
 		TrieTimeLimit:  5 * time.Minute,
 		SnapshotLimit:  0,
+		StateScheme:    snaptest.scheme,
 	}
 	newchain, err := NewBlockChain(snaptest.db, config, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
-	newBlocks, _ := GenerateChain(params.TestChainConfig, blocks[len(blocks)-1], snaptest.engine, snaptest.genDb, snaptest.newBlocks, func(i int, b *BlockGen) {})
+	newBlocks, _ := GenerateChain(snaptest.gspec.Config, blocks[len(blocks)-1], snaptest.engine, snaptest.genDb, snaptest.newBlocks, func(i int, b *BlockGen) {})
 	newchain.InsertChain(newBlocks)
 	newchain.Stop()
 
-	// Restart the chain, the wiper should starts working
+	// Restart the chain, the wiper should start working
 	config = &CacheConfig{
 		TrieCleanLimit: 256,
 		TrieDirtyLimit: 256,
 		TrieTimeLimit:  5 * time.Minute,
 		SnapshotLimit:  256,
 		SnapshotWait:   false, // Don't wait rebuild
+		StateScheme:    snaptest.scheme,
 	}
 	tmp, err := NewBlockChain(snaptest.db, config, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
@@ -402,14 +406,15 @@ func (snaptest *wipeCrashSnapshotTest) test(t *testing.T) {
 	}
 
 	// Simulate the blockchain crash.
+	tmp.triedb.Close()
 	tmp.stopWithoutSaving()
 
-	newchain, err = NewBlockChain(snaptest.db, nil, snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
+	newchain, err = NewBlockChain(snaptest.db, DefaultCacheConfigWithScheme(snaptest.scheme), snaptest.gspec, nil, snaptest.engine, vm.Config{}, nil, nil)
 	if err != nil {
 		t.Fatalf("Failed to recreate chain: %v", err)
 	}
-	defer newchain.Stop()
 	snaptest.verify(t, newchain, blocks)
+	newchain.Stop()
 }
 
 // Tests a Geth restart with valid snapshot. Before the shutdown, all snapshot
@@ -433,8 +438,10 @@ func TestRestartWithNewSnapshot(t *testing.T) {
 	// Expected head fast block: C8
 	// Expected head block     : C8
 	// Expected snapshot disk  : G
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
 		test := &snapshotTest{
 			snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      0,
 				commitBlock:        0,
@@ -447,6 +454,7 @@ func TestRestartWithNewSnapshot(t *testing.T) {
 		}
 		test.test(t)
 		test.teardown()
+	}
 }
 
 // Tests a Geth was crashed and restarts with a broken snapshot. In this case the
@@ -472,8 +480,10 @@ func TestNoCommitCrashWithNewSnapshot(t *testing.T) {
 	// Expected head fast block: C8
 	// Expected head block     : G
 	// Expected snapshot disk  : C4
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
 		test := &crashSnapshotTest{
 			snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      4,
 				commitBlock:        0,
@@ -486,6 +496,7 @@ func TestNoCommitCrashWithNewSnapshot(t *testing.T) {
 		}
 		test.test(t)
 		test.teardown()
+	}
 }
 
 // Tests a Geth was crashed and restarts with a broken snapshot. In this case the
@@ -511,8 +522,10 @@ func TestLowCommitCrashWithNewSnapshot(t *testing.T) {
 	// Expected head fast block: C8
 	// Expected head block     : C2
 	// Expected snapshot disk  : C4
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
 		test := &crashSnapshotTest{
 			snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      4,
 				commitBlock:        2,
@@ -525,6 +538,7 @@ func TestLowCommitCrashWithNewSnapshot(t *testing.T) {
 		}
 		test.test(t)
 		test.teardown()
+	}
 }
 
 // Tests a Geth was crashed and restarts with a broken snapshot. In this case
@@ -550,20 +564,27 @@ func TestHighCommitCrashWithNewSnapshot(t *testing.T) {
 	// Expected head fast block: C8
 	// Expected head block     : G
 	// Expected snapshot disk  : C4
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
+		expHead := uint64(0)
+		if scheme == rawdb.PathScheme {
+			expHead = uint64(4)
+		}
 		test := &crashSnapshotTest{
 			snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      4,
 				commitBlock:        6,
 				expCanonicalBlocks: 8,
 				expHeadHeader:      8,
 				expHeadFastBlock:   8,
-			expHeadBlock:       0,
+				expHeadBlock:       expHead,
 				expSnapshotBottom:  4, // Last committed disk layer, wait recovery
 			},
 		}
 		test.test(t)
 		test.teardown()
+	}
 }
 
 // Tests a Geth was running with snapshot enabled. Then restarts without
@@ -587,8 +608,10 @@ func TestGappedNewSnapshot(t *testing.T) {
 	// Expected head fast block: C10
 	// Expected head block     : C10
 	// Expected snapshot disk  : C10
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
 		test := &gappedSnapshotTest{
 			snapshotTestBasic: snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      0,
 				commitBlock:        0,
@@ -602,6 +625,7 @@ func TestGappedNewSnapshot(t *testing.T) {
 		}
 		test.test(t)
 		test.teardown()
+	}
 }
 
 // Tests the Geth was running with snapshot enabled and resetHead is applied.
@@ -625,8 +649,10 @@ func TestSetHeadWithNewSnapshot(t *testing.T) {
 	// Expected head fast block: C4
 	// Expected head block     : C4
 	// Expected snapshot disk  : G
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
 		test := &setHeadSnapshotTest{
 			snapshotTestBasic: snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      0,
 				commitBlock:        0,
@@ -640,6 +666,7 @@ func TestSetHeadWithNewSnapshot(t *testing.T) {
 		}
 		test.test(t)
 		test.teardown()
+	}
 }
 
 // Tests the Geth was running with a complete snapshot and then imports a few
@@ -663,8 +690,10 @@ func TestRecoverSnapshotFromWipingCrash(t *testing.T) {
 	// Expected head fast block: C10
 	// Expected head block     : C8
 	// Expected snapshot disk  : C10
+	for _, scheme := range []string{rawdb.HashScheme, rawdb.PathScheme} {
 		test := &wipeCrashSnapshotTest{
 			snapshotTestBasic: snapshotTestBasic{
+				scheme:             scheme,
 				chainBlocks:        8,
 				snapshotBlock:      4,
 				commitBlock:        0,
@@ -678,4 +707,5 @@ func TestRecoverSnapshotFromWipingCrash(t *testing.T) {
 		}
 		test.test(t)
 		test.teardown()
+	}
 }

core/chain_makers.go

@@ -283,7 +283,7 @@ func GenerateChain(config *params.ChainConfig, parent *types.Block, engine conse
 	}
 	blocks, receipts := make(types.Blocks, n), make([]types.Receipts, n)
 	chainreader := &fakeChainReader{config: config}
-	genblock := func(i int, parent *types.Block, statedb *state.StateDB) (*types.Block, types.Receipts) {
+	genblock := func(i int, parent *types.Block, triedb *trie.Database, statedb *state.StateDB) (*types.Block, types.Receipts) {
 		b := &BlockGen{i: i, chain: blocks, parent: parent, statedb: statedb, config: config, engine: engine}
 		b.header = makeHeader(chainreader, parent, statedb, b.engine)
 
@@ -326,19 +326,23 @@ func GenerateChain(config *params.ChainConfig, parent *types.Block, engine conse
 			if err != nil {
 				panic(fmt.Sprintf("state write error: %v", err))
 			}
-			if err := statedb.Database().TrieDB().Commit(root, false); err != nil {
+			if err = triedb.Commit(root, false); err != nil {
 				panic(fmt.Sprintf("trie write error: %v", err))
 			}
 			return block, b.receipts
 		}
 		return nil, nil
 	}
+	// Forcibly use hash-based state scheme for retaining all nodes in disk.
+	triedb := trie.NewDatabase(db, trie.HashDefaults)
+	defer triedb.Close()
+
 	for i := 0; i < n; i++ {
-		statedb, err := state.New(parent.Root(), state.NewDatabase(db), nil)
+		statedb, err := state.New(parent.Root(), state.NewDatabaseWithNodeDB(db, triedb), nil)
 		if err != nil {
 			panic(err)
 		}
-		block, receipt := genblock(i, parent, statedb)
+		block, receipt := genblock(i, parent, triedb, statedb)
 		blocks[i] = block
 		receipts[i] = receipt
 		parent = block
@@ -351,7 +355,9 @@ func GenerateChain(config *params.ChainConfig, parent *types.Block, engine conse
 // then generate chain on top.
 func GenerateChainWithGenesis(genesis *Genesis, engine consensus.Engine, n int, gen func(int, *BlockGen)) (ethdb.Database, []*types.Block, []types.Receipts) {
 	db := rawdb.NewMemoryDatabase()
-	_, err := genesis.Commit(db, trie.NewDatabase(db))
+	triedb := trie.NewDatabase(db, trie.HashDefaults)
+	defer triedb.Close()
+	_, err := genesis.Commit(db, triedb)
 	if err != nil {
 		panic(err)
 	}

core/chain_makers_test.go

@@ -29,6 +29,7 @@ import (
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/params"
+	"github.com/ethereum/go-ethereum/trie"
 )
 
 func TestGenerateWithdrawalChain(t *testing.T) {
@@ -74,8 +75,7 @@ func TestGenerateWithdrawalChain(t *testing.T) {
 		Storage: storage,
 		Code:    common.Hex2Bytes("600154600354"),
 	}
-
+	genesis := gspec.MustCommit(gendb, trie.NewDatabase(gendb, trie.HashDefaults))
-	genesis := gspec.MustCommit(gendb)
 
 	chain, _ := GenerateChain(gspec.Config, genesis, beacon.NewFaker(), gendb, 4, func(i int, gen *BlockGen) {
 		tx, _ := types.SignTx(types.NewTransaction(gen.TxNonce(address), address, big.NewInt(1000), params.TxGas, new(big.Int).Add(gen.BaseFee(), common.Big1), nil), signer, key)
@@ -146,6 +146,7 @@ func ExampleGenerateChain() {
 		addr2   = crypto.PubkeyToAddress(key2.PublicKey)
 		addr3   = crypto.PubkeyToAddress(key3.PublicKey)
 		db      = rawdb.NewMemoryDatabase()
+		genDb   = rawdb.NewMemoryDatabase()
 	)
 
 	// Ensure that key1 has some funds in the genesis block.
@@ -153,13 +154,13 @@ func ExampleGenerateChain() {
 		Config: &params.ChainConfig{HomesteadBlock: new(big.Int)},
 		Alloc:  GenesisAlloc{addr1: {Balance: big.NewInt(1000000)}},
 	}
-	genesis := gspec.MustCommit(db)
+	genesis := gspec.MustCommit(genDb, trie.NewDatabase(genDb, trie.HashDefaults))
 
 	// This call generates a chain of 5 blocks. The function runs for
 	// each block and adds different features to gen based on the
 	// block index.
 	signer := types.HomesteadSigner{}
-	chain, _ := GenerateChain(gspec.Config, genesis, ethash.NewFaker(), db, 5, func(i int, gen *BlockGen) {
+	chain, _ := GenerateChain(gspec.Config, genesis, ethash.NewFaker(), genDb, 5, func(i int, gen *BlockGen) {
 		switch i {
 		case 0:
 			// In block 1, addr1 sends addr2 some ether.
@@ -188,7 +189,7 @@ func ExampleGenerateChain() {
 	})
 
 	// Import the chain. This runs all block validation rules.
-	blockchain, _ := NewBlockChain(db, nil, gspec, nil, ethash.NewFaker(), vm.Config{}, nil, nil)
+	blockchain, _ := NewBlockChain(db, DefaultCacheConfigWithScheme(rawdb.HashScheme), gspec, nil, ethash.NewFaker(), vm.Config{}, nil, nil)
 	defer blockchain.Stop()
 
 	if i, err := blockchain.InsertChain(chain); err != nil {

core/dao_test.go

@@ -83,7 +83,7 @@ func TestDAOForkRangeExtradata(t *testing.T) {
 		if _, err := bc.InsertChain(blocks); err != nil {
 			t.Fatalf("failed to import contra-fork chain for expansion: %v", err)
 		}
-		if err := bc.stateCache.TrieDB().Commit(bc.CurrentHeader().Root, false); err != nil {
+		if err := bc.triedb.Commit(bc.CurrentHeader().Root, false); err != nil {
 			t.Fatalf("failed to commit contra-fork head for expansion: %v", err)
 		}
 		bc.Stop()
@@ -106,7 +106,7 @@ func TestDAOForkRangeExtradata(t *testing.T) {
 		if _, err := bc.InsertChain(blocks); err != nil {
 			t.Fatalf("failed to import pro-fork chain for expansion: %v", err)
 		}
-		if err := bc.stateCache.TrieDB().Commit(bc.CurrentHeader().Root, false); err != nil {
+		if err := bc.triedb.Commit(bc.CurrentHeader().Root, false); err != nil {
 			t.Fatalf("failed to commit pro-fork head for expansion: %v", err)
 		}
 		bc.Stop()
@@ -131,7 +131,7 @@ func TestDAOForkRangeExtradata(t *testing.T) {
 	if _, err := bc.InsertChain(blocks); err != nil {
 		t.Fatalf("failed to import contra-fork chain for expansion: %v", err)
 	}
-	if err := bc.stateCache.TrieDB().Commit(bc.CurrentHeader().Root, false); err != nil {
+	if err := bc.triedb.Commit(bc.CurrentHeader().Root, false); err != nil {
 		t.Fatalf("failed to commit contra-fork head for expansion: %v", err)
 	}
 	blocks, _ = GenerateChain(&proConf, conBc.GetBlockByHash(conBc.CurrentBlock().Hash()), ethash.NewFaker(), genDb, 1, func(i int, gen *BlockGen) {})
@@ -149,7 +149,7 @@ func TestDAOForkRangeExtradata(t *testing.T) {
 	if _, err := bc.InsertChain(blocks); err != nil {
 		t.Fatalf("failed to import pro-fork chain for expansion: %v", err)
 	}
-	if err := bc.stateCache.TrieDB().Commit(bc.CurrentHeader().Root, false); err != nil {
+	if err := bc.triedb.Commit(bc.CurrentHeader().Root, false); err != nil {
 		t.Fatalf("failed to commit pro-fork head for expansion: %v", err)
 	}
 	blocks, _ = GenerateChain(&conConf, proBc.GetBlockByHash(proBc.CurrentBlock().Hash()), ethash.NewFaker(), genDb, 1, func(i int, gen *BlockGen) {})

core/genesis.go

@@ -323,10 +323,12 @@ func SetupGenesisBlockWithOverride(db ethdb.Database, triedb *trie.Database, gen
 		applyOverrides(genesis.Config)
 		return genesis.Config, block.Hash(), nil
 	}
-	// We have the genesis block in database(perhaps in ancient database)
+	// The genesis block is present(perhaps in ancient database) while the
-	// but the corresponding state is missing.
+	// state database is not initialized yet. It can happen that the node
+	// is initialized with an external ancient store. Commit genesis state
+	// in this case.
 	header := rawdb.ReadHeader(db, stored, 0)
-	if header.Root != types.EmptyRootHash && !rawdb.HasLegacyTrieNode(db, header.Root) {
+	if header.Root != types.EmptyRootHash && !triedb.Initialized(header.Root) {
 		if genesis == nil {
 			genesis = DefaultGenesisBlock()
 		}
@@ -526,10 +528,8 @@ func (g *Genesis) Commit(db ethdb.Database, triedb *trie.Database) (*types.Block
 
 // MustCommit writes the genesis block and state to db, panicking on error.
 // The block is committed as the canonical head block.
-// Note the state changes will be committed in hash-based scheme, use Commit
+func (g *Genesis) MustCommit(db ethdb.Database, triedb *trie.Database) *types.Block {
-// if path-scheme is preferred.
+	block, err := g.Commit(db, triedb)
-func (g *Genesis) MustCommit(db ethdb.Database) *types.Block {
-	block, err := g.Commit(db, trie.NewDatabase(db))
 	if err != nil {
 		panic(err)
 	}

core/genesis_test.go

@@ -30,18 +30,24 @@ import (
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 )
 
 func TestInvalidCliqueConfig(t *testing.T) {
 	block := DefaultGoerliGenesisBlock()
 	block.ExtraData = []byte{}
 	db := rawdb.NewMemoryDatabase()
-	if _, err := block.Commit(db, trie.NewDatabase(db)); err == nil {
+	if _, err := block.Commit(db, trie.NewDatabase(db, nil)); err == nil {
 		t.Fatal("Expected error on invalid clique config")
 	}
 }
 
 func TestSetupGenesis(t *testing.T) {
+	testSetupGenesis(t, rawdb.HashScheme)
+	testSetupGenesis(t, rawdb.PathScheme)
+}
+
+func testSetupGenesis(t *testing.T, scheme string) {
 	var (
 		customghash = common.HexToHash("0x89c99d90b79719238d2645c7642f2c9295246e80775b38cfd162b696817fbd50")
 		customg     = Genesis{
@@ -53,6 +59,7 @@ func TestSetupGenesis(t *testing.T) {
 		oldcustomg = customg
 	)
 	oldcustomg.Config = &params.ChainConfig{HomesteadBlock: big.NewInt(2)}
+
 	tests := []struct {
 		name       string
 		fn         func(ethdb.Database) (*params.ChainConfig, common.Hash, error)
@@ -63,7 +70,7 @@ func TestSetupGenesis(t *testing.T) {
 		{
 			name: "genesis without ChainConfig",
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
-				return SetupGenesisBlock(db, trie.NewDatabase(db), new(Genesis))
+				return SetupGenesisBlock(db, trie.NewDatabase(db, newDbConfig(scheme)), new(Genesis))
 			},
 			wantErr:    errGenesisNoConfig,
 			wantConfig: params.AllEthashProtocolChanges,
@@ -71,7 +78,7 @@ func TestSetupGenesis(t *testing.T) {
 		{
 			name: "no block in DB, genesis == nil",
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
-				return SetupGenesisBlock(db, trie.NewDatabase(db), nil)
+				return SetupGenesisBlock(db, trie.NewDatabase(db, newDbConfig(scheme)), nil)
 			},
 			wantHash:   params.MainnetGenesisHash,
 			wantConfig: params.MainnetChainConfig,
@@ -79,8 +86,8 @@ func TestSetupGenesis(t *testing.T) {
 		{
 			name: "mainnet block in DB, genesis == nil",
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
-				DefaultGenesisBlock().MustCommit(db)
+				DefaultGenesisBlock().MustCommit(db, trie.NewDatabase(db, newDbConfig(scheme)))
-				return SetupGenesisBlock(db, trie.NewDatabase(db), nil)
+				return SetupGenesisBlock(db, trie.NewDatabase(db, newDbConfig(scheme)), nil)
 			},
 			wantHash:   params.MainnetGenesisHash,
 			wantConfig: params.MainnetChainConfig,
@@ -88,8 +95,9 @@ func TestSetupGenesis(t *testing.T) {
 		{
 			name: "custom block in DB, genesis == nil",
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
-				customg.MustCommit(db)
+				tdb := trie.NewDatabase(db, newDbConfig(scheme))
-				return SetupGenesisBlock(db, trie.NewDatabase(db), nil)
+				customg.Commit(db, tdb)
+				return SetupGenesisBlock(db, tdb, nil)
 			},
 			wantHash:   customghash,
 			wantConfig: customg.Config,
@@ -97,8 +105,9 @@ func TestSetupGenesis(t *testing.T) {
 		{
 			name: "custom block in DB, genesis == goerli",
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
-				customg.MustCommit(db)
+				tdb := trie.NewDatabase(db, newDbConfig(scheme))
-				return SetupGenesisBlock(db, trie.NewDatabase(db), DefaultGoerliGenesisBlock())
+				customg.Commit(db, tdb)
+				return SetupGenesisBlock(db, tdb, DefaultGoerliGenesisBlock())
 			},
 			wantErr:    &GenesisMismatchError{Stored: customghash, New: params.GoerliGenesisHash},
 			wantHash:   params.GoerliGenesisHash,
@@ -107,8 +116,9 @@ func TestSetupGenesis(t *testing.T) {
 		{
 			name: "compatible config in DB",
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
-				oldcustomg.MustCommit(db)
+				tdb := trie.NewDatabase(db, newDbConfig(scheme))
-				return SetupGenesisBlock(db, trie.NewDatabase(db), &customg)
+				oldcustomg.Commit(db, tdb)
+				return SetupGenesisBlock(db, tdb, &customg)
 			},
 			wantHash:   customghash,
 			wantConfig: customg.Config,
@@ -118,16 +128,17 @@ func TestSetupGenesis(t *testing.T) {
 			fn: func(db ethdb.Database) (*params.ChainConfig, common.Hash, error) {
 				// Commit the 'old' genesis block with Homestead transition at #2.
 				// Advance to block #4, past the homestead transition block of customg.
-				genesis := oldcustomg.MustCommit(db)
+				tdb := trie.NewDatabase(db, newDbConfig(scheme))
+				oldcustomg.Commit(db, tdb)
 
-				bc, _ := NewBlockChain(db, nil, &oldcustomg, nil, ethash.NewFullFaker(), vm.Config{}, nil, nil)
+				bc, _ := NewBlockChain(db, DefaultCacheConfigWithScheme(scheme), &oldcustomg, nil, ethash.NewFullFaker(), vm.Config{}, nil, nil)
 				defer bc.Stop()
 
-				blocks, _ := GenerateChain(oldcustomg.Config, genesis, ethash.NewFaker(), db, 4, nil)
+				_, blocks, _ := GenerateChainWithGenesis(&oldcustomg, ethash.NewFaker(), 4, nil)
 				bc.InsertChain(blocks)
 
 				// This should return a compatibility error.
-				return SetupGenesisBlock(db, trie.NewDatabase(db), &customg)
+				return SetupGenesisBlock(db, tdb, &customg)
 			},
 			wantHash:   customghash,
 			wantConfig: customg.Config,
@@ -175,7 +186,8 @@ func TestGenesisHashes(t *testing.T) {
 		{DefaultSepoliaGenesisBlock(), params.SepoliaGenesisHash},
 	} {
 		// Test via MustCommit
-		if have := c.genesis.MustCommit(rawdb.NewMemoryDatabase()).Hash(); have != c.want {
+		db := rawdb.NewMemoryDatabase()
+		if have := c.genesis.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults)).Hash(); have != c.want {
 			t.Errorf("case: %d a), want: %s, got: %s", i, c.want.Hex(), have.Hex())
 		}
 		// Test via ToBlock
@@ -193,7 +205,7 @@ func TestGenesis_Commit(t *testing.T) {
 	}
 
 	db := rawdb.NewMemoryDatabase()
-	genesisBlock := genesis.MustCommit(db)
+	genesisBlock := genesis.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults))
 
 	if genesis.Difficulty != nil {
 		t.Fatalf("assumption wrong")
@@ -242,3 +254,10 @@ func TestReadWriteGenesisAlloc(t *testing.T) {
 		}
 	}
 }
+
+func newDbConfig(scheme string) *trie.Config {
+	if scheme == rawdb.HashScheme {
+		return trie.HashDefaults
+	}
+	return &trie.Config{PathDB: pathdb.Defaults}
+}

core/headerchain_test.go

@@ -73,7 +73,7 @@ func TestHeaderInsertion(t *testing.T) {
 		db    = rawdb.NewMemoryDatabase()
 		gspec = &Genesis{BaseFee: big.NewInt(params.InitialBaseFee), Config: params.AllEthashProtocolChanges}
 	)
-	gspec.Commit(db, trie.NewDatabase(db))
+	gspec.Commit(db, trie.NewDatabase(db, nil))
 	hc, err := NewHeaderChain(db, gspec.Config, ethash.NewFaker(), func() bool { return false })
 	if err != nil {
 		t.Fatal(err)

core/rawdb/accessors_chain_test.go

@@ -435,12 +435,12 @@ func checkReceiptsRLP(have, want types.Receipts) error {
 func TestAncientStorage(t *testing.T) {
 	// Freezer style fast import the chain.
 	frdir := t.TempDir()
-
 	db, err := NewDatabaseWithFreezer(NewMemoryDatabase(), frdir, "", false)
 	if err != nil {
 		t.Fatalf("failed to create database with ancient backend")
 	}
 	defer db.Close()
+
 	// Create a test block
 	block := types.NewBlockWithHeader(&types.Header{
 		Number:      big.NewInt(0),

core/rawdb/accessors_trie.go

@@ -36,7 +36,7 @@ import (
 //
 // Now this scheme is still kept for backward compatibility, and it will be used
 // for archive node and some other tries(e.g. light trie).
-const HashScheme = "hashScheme"
+const HashScheme = "hash"
 
 // PathScheme is the new path-based state scheme with which trie nodes are stored
 // in the disk with node path as the database key. This scheme will only store one
@@ -44,7 +44,7 @@ const HashScheme = "hashScheme"
 // is native. At the same time, this scheme will put adjacent trie nodes in the same
 // area of the disk with good data locality property. But this scheme needs to rely
 // on extra state diffs to survive deep reorg.
-const PathScheme = "pathScheme"
+const PathScheme = "path"
 
 // hasher is used to compute the sha256 hash of the provided data.
 type hasher struct{ sha crypto.KeccakState }
@@ -263,3 +263,25 @@ func DeleteTrieNode(db ethdb.KeyValueWriter, owner common.Hash, path []byte, has
 		panic(fmt.Sprintf("Unknown scheme %v", scheme))
 	}
 }
+
+// ReadStateScheme reads the state scheme of persistent state, or none
+// if the state is not present in database.
+func ReadStateScheme(db ethdb.Reader) string {
+	// Check if state in path-based scheme is present
+	blob, _ := ReadAccountTrieNode(db, nil)
+	if len(blob) != 0 {
+		return PathScheme
+	}
+	// In a hash-based scheme, the genesis state is consistently stored
+	// on the disk. To assess the scheme of the persistent state, it
+	// suffices to inspect the scheme of the genesis state.
+	header := ReadHeader(db, ReadCanonicalHash(db, 0), 0)
+	if header == nil {
+		return "" // empty datadir
+	}
+	blob = ReadLegacyTrieNode(db, header.Root)
+	if len(blob) == 0 {
+		return "" // no state in disk
+	}
+	return HashScheme
+}

core/rawdb/ancient_scheme.go

@@ -58,7 +58,7 @@ const (
 	stateHistoryStorageData  = "storage.data"
 )
 
-var stateHistoryFreezerNoSnappy = map[string]bool{
+var stateFreezerNoSnappy = map[string]bool{
 	stateHistoryMeta:         true,
 	stateHistoryAccountIndex: false,
 	stateHistoryStorageIndex: false,
@@ -75,7 +75,7 @@ var (
 // freezers the collections of all builtin freezers.
 var freezers = []string{chainFreezerName, stateFreezerName}
 
-// NewStateHistoryFreezer initializes the freezer for state history.
+// NewStateFreezer initializes the freezer for state history.
-func NewStateHistoryFreezer(ancientDir string, readOnly bool) (*ResettableFreezer, error) {
+func NewStateFreezer(ancientDir string, readOnly bool) (*ResettableFreezer, error) {
-	return NewResettableFreezer(filepath.Join(ancientDir, stateFreezerName), "eth/db/state", readOnly, stateHistoryTableSize, stateHistoryFreezerNoSnappy)
+	return NewResettableFreezer(filepath.Join(ancientDir, stateFreezerName), "eth/db/state", readOnly, stateHistoryTableSize, stateFreezerNoSnappy)
 }

core/rawdb/ancient_utils.go

@@ -50,36 +50,58 @@ func (info *freezerInfo) size() common.StorageSize {
 	return total
 }
 
+func inspect(name string, order map[string]bool, reader ethdb.AncientReader) (freezerInfo, error) {
+	info := freezerInfo{name: name}
+	for t := range order {
+		size, err := reader.AncientSize(t)
+		if err != nil {
+			return freezerInfo{}, err
+		}
+		info.sizes = append(info.sizes, tableSize{name: t, size: common.StorageSize(size)})
+	}
+	// Retrieve the number of last stored item
+	ancients, err := reader.Ancients()
+	if err != nil {
+		return freezerInfo{}, err
+	}
+	info.head = ancients - 1
+
+	// Retrieve the number of first stored item
+	tail, err := reader.Tail()
+	if err != nil {
+		return freezerInfo{}, err
+	}
+	info.tail = tail
+	return info, nil
+}
+
 // inspectFreezers inspects all freezers registered in the system.
 func inspectFreezers(db ethdb.Database) ([]freezerInfo, error) {
 	var infos []freezerInfo
 	for _, freezer := range freezers {
 		switch freezer {
 		case chainFreezerName:
-			// Chain ancient store is a bit special. It's always opened along
+			info, err := inspect(chainFreezerName, chainFreezerNoSnappy, db)
-			// with the key-value store, inspect the chain store directly.
-			info := freezerInfo{name: freezer}
-			// Retrieve storage size of every contained table.
-			for table := range chainFreezerNoSnappy {
-				size, err := db.AncientSize(table)
 			if err != nil {
 				return nil, err
 			}
-				info.sizes = append(info.sizes, tableSize{name: table, size: common.StorageSize(size)})
+			infos = append(infos, info)
-			}
-			// Retrieve the number of last stored item
-			ancients, err := db.Ancients()
-			if err != nil {
-				return nil, err
-			}
-			info.head = ancients - 1
 
-			// Retrieve the number of first stored item
+		case stateFreezerName:
-			tail, err := db.Tail()
+			datadir, err := db.AncientDatadir()
+			if err != nil {
+				return nil, err
+			}
+			f, err := NewStateFreezer(datadir, true)
+			if err != nil {
+				return nil, err
+			}
+			defer f.Close()
+
+			info, err := inspect(stateFreezerName, stateFreezerNoSnappy, f)
 			if err != nil {
 				return nil, err
 			}
-			info.tail = tail
 			infos = append(infos, info)
 
 		default:

core/rawdb/database.go

@@ -463,7 +463,10 @@ func InspectDatabase(db ethdb.Database, keyPrefix, keyStart []byte) error {
 		tds             stat
 		numHashPairings stat
 		hashNumPairings stat
-		tries           stat
+		legacyTries     stat
+		stateLookups    stat
+		accountTries    stat
+		storageTries    stat
 		codes           stat
 		txLookups       stat
 		accountSnaps    stat
@@ -504,8 +507,14 @@ func InspectDatabase(db ethdb.Database, keyPrefix, keyStart []byte) error {
 			numHashPairings.Add(size)
 		case bytes.HasPrefix(key, headerNumberPrefix) && len(key) == (len(headerNumberPrefix)+common.HashLength):
 			hashNumPairings.Add(size)
-		case len(key) == common.HashLength:
+		case IsLegacyTrieNode(key, it.Value()):
-			tries.Add(size)
+			legacyTries.Add(size)
+		case bytes.HasPrefix(key, stateIDPrefix) && len(key) == len(stateIDPrefix)+common.HashLength:
+			stateLookups.Add(size)
+		case IsAccountTrieNode(key):
+			accountTries.Add(size)
+		case IsStorageTrieNode(key):
+			storageTries.Add(size)
 		case bytes.HasPrefix(key, CodePrefix) && len(key) == len(CodePrefix)+common.HashLength:
 			codes.Add(size)
 		case bytes.HasPrefix(key, txLookupPrefix) && len(key) == (len(txLookupPrefix)+common.HashLength):
@@ -543,6 +552,7 @@ func InspectDatabase(db ethdb.Database, keyPrefix, keyStart []byte) error {
 				lastPivotKey, fastTrieProgressKey, snapshotDisabledKey, SnapshotRootKey, snapshotJournalKey,
 				snapshotGeneratorKey, snapshotRecoveryKey, txIndexTailKey, fastTxLookupLimitKey,
 				uncleanShutdownKey, badBlockKey, transitionStatusKey, skeletonSyncStatusKey,
+				persistentStateIDKey, trieJournalKey, snapshotSyncStatusKey,
 			} {
 				if bytes.Equal(key, meta) {
 					metadata.Add(size)
@@ -571,7 +581,10 @@ func InspectDatabase(db ethdb.Database, keyPrefix, keyStart []byte) error {
 		{"Key-Value store", "Transaction index", txLookups.Size(), txLookups.Count()},
 		{"Key-Value store", "Bloombit index", bloomBits.Size(), bloomBits.Count()},
 		{"Key-Value store", "Contract codes", codes.Size(), codes.Count()},
-		{"Key-Value store", "Trie nodes", tries.Size(), tries.Count()},
+		{"Key-Value store", "Hash trie nodes", legacyTries.Size(), legacyTries.Count()},
+		{"Key-Value store", "Path trie state lookups", stateLookups.Size(), stateLookups.Count()},
+		{"Key-Value store", "Path trie account nodes", accountTries.Size(), accountTries.Count()},
+		{"Key-Value store", "Path trie storage nodes", storageTries.Size(), storageTries.Count()},
 		{"Key-Value store", "Trie preimages", preimages.Size(), preimages.Count()},
 		{"Key-Value store", "Account snapshot", accountSnaps.Size(), accountSnaps.Count()},
 		{"Key-Value store", "Storage snapshot", storageSnaps.Size(), storageSnaps.Count()},

core/rawdb/schema.go

@@ -273,9 +273,10 @@ func IsLegacyTrieNode(key []byte, val []byte) bool {
 	return bytes.Equal(key, crypto.Keccak256(val))
 }
 
-// IsAccountTrieNode reports whether a provided database entry is an account
+// ResolveAccountTrieNodeKey reports whether a provided database entry is an
-// trie node in path-based state scheme.
+// account trie node in path-based state scheme, and returns the resolved
-func IsAccountTrieNode(key []byte) (bool, []byte) {
+// node path if so.
+func ResolveAccountTrieNodeKey(key []byte) (bool, []byte) {
 	if !bytes.HasPrefix(key, trieNodeAccountPrefix) {
 		return false, nil
 	}
@@ -288,9 +289,17 @@ func IsAccountTrieNode(key []byte) (bool, []byte) {
 	return true, key[len(trieNodeAccountPrefix):]
 }
 
-// IsStorageTrieNode reports whether a provided database entry is a storage
+// IsAccountTrieNode reports whether a provided database entry is an account
 // trie node in path-based state scheme.
-func IsStorageTrieNode(key []byte) (bool, common.Hash, []byte) {
+func IsAccountTrieNode(key []byte) bool {
+	ok, _ := ResolveAccountTrieNodeKey(key)
+	return ok
+}
+
+// ResolveStorageTrieNode reports whether a provided database entry is a storage
+// trie node in path-based state scheme, and returns the resolved account hash
+// and node path if so.
+func ResolveStorageTrieNode(key []byte) (bool, common.Hash, []byte) {
 	if !bytes.HasPrefix(key, trieNodeStoragePrefix) {
 		return false, common.Hash{}, nil
 	}
@@ -306,3 +315,10 @@ func IsStorageTrieNode(key []byte) (bool, common.Hash, []byte) {
 	accountHash := common.BytesToHash(key[len(trieNodeStoragePrefix) : len(trieNodeStoragePrefix)+common.HashLength])
 	return true, accountHash, key[len(trieNodeStoragePrefix)+common.HashLength:]
 }
+
+// IsStorageTrieNode reports whether a provided database entry is a storage
+// trie node in path-based state scheme.
+func IsStorageTrieNode(key []byte) bool {
+	ok, _, _ := ResolveStorageTrieNode(key)
+	return ok
+}

core/state/database.go

@@ -58,7 +58,7 @@ type Database interface {
 	// DiskDB returns the underlying key-value disk database.
 	DiskDB() ethdb.KeyValueStore
 
-	// TrieDB retrieves the low level trie database used for data storage.
+	// TrieDB returns the underlying trie database for managing trie nodes.
 	TrieDB() *trie.Database
 }
 
@@ -147,7 +147,7 @@ func NewDatabaseWithConfig(db ethdb.Database, config *trie.Config) Database {
 		disk:          db,
 		codeSizeCache: lru.NewCache[common.Hash, int](codeSizeCacheSize),
 		codeCache:     lru.NewSizeConstrainedCache[common.Hash, []byte](codeCacheSize),
-		triedb:        trie.NewDatabaseWithConfig(db, config),
+		triedb:        trie.NewDatabase(db, config),
 	}
 }
 

core/state/iterator_test.go

@@ -26,9 +26,14 @@ import (
 
 // Tests that the node iterator indeed walks over the entire database contents.
 func TestNodeIteratorCoverage(t *testing.T) {
+	testNodeIteratorCoverage(t, rawdb.HashScheme)
+	testNodeIteratorCoverage(t, rawdb.PathScheme)
+}
+
+func testNodeIteratorCoverage(t *testing.T, scheme string) {
 	// Create some arbitrary test state to iterate
-	db, sdb, root, _ := makeTestState()
+	db, sdb, ndb, root, _ := makeTestState(scheme)
-	sdb.TrieDB().Commit(root, false)
+	ndb.Commit(root, false)
 
 	state, err := New(root, sdb, nil)
 	if err != nil {
@@ -48,7 +53,7 @@ func TestNodeIteratorCoverage(t *testing.T) {
 	)
 	it := db.NewIterator(nil, nil)
 	for it.Next() {
-		ok, hash := isTrieNode(sdb.TrieDB().Scheme(), it.Key(), it.Value())
+		ok, hash := isTrieNode(scheme, it.Key(), it.Value())
 		if !ok {
 			continue
 		}
@@ -90,11 +95,11 @@ func isTrieNode(scheme string, key, val []byte) (bool, common.Hash) {
 			return true, common.BytesToHash(key)
 		}
 	} else {
-		ok, _ := rawdb.IsAccountTrieNode(key)
+		ok := rawdb.IsAccountTrieNode(key)
 		if ok {
 			return true, crypto.Keccak256Hash(val)
 		}
-		ok, _, _ = rawdb.IsStorageTrieNode(key)
+		ok = rawdb.IsStorageTrieNode(key)
 		if ok {
 			return true, crypto.Keccak256Hash(val)
 		}

core/state/pruner/pruner.go

@@ -85,13 +85,16 @@ func NewPruner(db ethdb.Database, config Config) (*Pruner, error) {
 	if headBlock == nil {
 		return nil, errors.New("failed to load head block")
 	}
+	// Offline pruning is only supported in legacy hash based scheme.
+	triedb := trie.NewDatabase(db, trie.HashDefaults)
+
 	snapconfig := snapshot.Config{
 		CacheSize:  256,
 		Recovery:   false,
 		NoBuild:    true,
 		AsyncBuild: false,
 	}
-	snaptree, err := snapshot.New(snapconfig, db, trie.NewDatabase(db), headBlock.Root())
+	snaptree, err := snapshot.New(snapconfig, db, triedb, headBlock.Root())
 	if err != nil {
 		return nil, err // The relevant snapshot(s) might not exist
 	}
@@ -361,7 +364,9 @@ func RecoverPruning(datadir string, db ethdb.Database) error {
 		NoBuild:    true,
 		AsyncBuild: false,
 	}
-	snaptree, err := snapshot.New(snapconfig, db, trie.NewDatabase(db), headBlock.Root())
+	// Offline pruning is only supported in legacy hash based scheme.
+	triedb := trie.NewDatabase(db, trie.HashDefaults)
+	snaptree, err := snapshot.New(snapconfig, db, triedb, headBlock.Root())
 	if err != nil {
 		return err // The relevant snapshot(s) might not exist
 	}
@@ -403,7 +408,7 @@ func extractGenesis(db ethdb.Database, stateBloom *stateBloom) error {
 	if genesis == nil {
 		return errors.New("missing genesis block")
 	}
-	t, err := trie.NewStateTrie(trie.StateTrieID(genesis.Root()), trie.NewDatabase(db))
+	t, err := trie.NewStateTrie(trie.StateTrieID(genesis.Root()), trie.NewDatabase(db, trie.HashDefaults))
 	if err != nil {
 		return err
 	}
@@ -427,7 +432,7 @@ func extractGenesis(db ethdb.Database, stateBloom *stateBloom) error {
 			}
 			if acc.Root != types.EmptyRootHash {
 				id := trie.StorageTrieID(genesis.Root(), common.BytesToHash(accIter.LeafKey()), acc.Root)
-				storageTrie, err := trie.NewStateTrie(id, trie.NewDatabase(db))
+				storageTrie, err := trie.NewStateTrie(id, trie.NewDatabase(db, trie.HashDefaults))
 				if err != nil {
 					return err
 				}

core/state/snapshot/generate.go

@@ -356,7 +356,8 @@ func (dl *diskLayer) generateRange(ctx *generatorContext, trieId *trie.ID, prefi
 	var resolver trie.NodeResolver
 	if len(result.keys) > 0 {
 		mdb := rawdb.NewMemoryDatabase()
-		tdb := trie.NewDatabase(mdb)
+		tdb := trie.NewDatabase(mdb, trie.HashDefaults)
+		defer tdb.Close()
 		snapTrie := trie.NewEmpty(tdb)
 		for i, key := range result.keys {
 			snapTrie.Update(key, result.vals[i])

core/state/snapshot/generate_test.go

@@ -30,6 +30,8 @@ import (
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 	"github.com/ethereum/go-ethereum/trie/trienode"
 	"golang.org/x/crypto/sha3"
 )
@@ -45,10 +47,15 @@ func hashData(input []byte) common.Hash {
 
 // Tests that snapshot generation from an empty database.
 func TestGeneration(t *testing.T) {
+	testGeneration(t, rawdb.HashScheme)
+	testGeneration(t, rawdb.PathScheme)
+}
+
+func testGeneration(t *testing.T, scheme string) {
 	// We can't use statedb to make a test trie (circular dependency), so make
 	// a fake one manually. We're going with a small account trie of 3 accounts,
 	// two of which also has the same 3-slot storage trie attached.
-	var helper = newHelper()
+	var helper = newHelper(scheme)
 	stRoot := helper.makeStorageTrie(common.Hash{}, []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, false)
 
 	helper.addTrieAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: stRoot, CodeHash: types.EmptyCodeHash.Bytes()})
@@ -79,10 +86,15 @@ func TestGeneration(t *testing.T) {
 
 // Tests that snapshot generation with existent flat state.
 func TestGenerateExistentState(t *testing.T) {
+	testGenerateExistentState(t, rawdb.HashScheme)
+	testGenerateExistentState(t, rawdb.PathScheme)
+}
+
+func testGenerateExistentState(t *testing.T, scheme string) {
 	// We can't use statedb to make a test trie (circular dependency), so make
 	// a fake one manually. We're going with a small account trie of 3 accounts,
 	// two of which also has the same 3-slot storage trie attached.
-	var helper = newHelper()
+	var helper = newHelper(scheme)
 
 	stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)
 	helper.addTrieAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: stRoot, CodeHash: types.EmptyCodeHash.Bytes()})
@@ -148,9 +160,15 @@ type testHelper struct {
 	nodes   *trienode.MergedNodeSet
 }
 
-func newHelper() *testHelper {
+func newHelper(scheme string) *testHelper {
 	diskdb := rawdb.NewMemoryDatabase()
-	triedb := trie.NewDatabase(diskdb)
+	config := &trie.Config{}
+	if scheme == rawdb.PathScheme {
+		config.PathDB = &pathdb.Config{} // disable caching
+	} else {
+		config.HashDB = &hashdb.Config{} // disable caching
+	}
+	triedb := trie.NewDatabase(diskdb, config)
 	accTrie, _ := trie.NewStateTrie(trie.StateTrieID(types.EmptyRootHash), triedb)
 	return &testHelper{
 		diskdb:  diskdb,
@@ -233,7 +251,12 @@ func (t *testHelper) CommitAndGenerate() (common.Hash, *diskLayer) {
 //   - extra slots in the middle
 //   - extra slots in the end
 func TestGenerateExistentStateWithWrongStorage(t *testing.T) {
-	helper := newHelper()
+	testGenerateExistentStateWithWrongStorage(t, rawdb.HashScheme)
+	testGenerateExistentStateWithWrongStorage(t, rawdb.PathScheme)
+}
+
+func testGenerateExistentStateWithWrongStorage(t *testing.T, scheme string) {
+	helper := newHelper(scheme)
 
 	// Account one, empty root but non-empty database
 	helper.addAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: types.EmptyRootHash, CodeHash: types.EmptyCodeHash.Bytes()})
@@ -325,7 +348,12 @@ func TestGenerateExistentStateWithWrongStorage(t *testing.T) {
 // - wrong accounts
 // - extra accounts
 func TestGenerateExistentStateWithWrongAccounts(t *testing.T) {
-	helper := newHelper()
+	testGenerateExistentStateWithWrongAccounts(t, rawdb.HashScheme)
+	testGenerateExistentStateWithWrongAccounts(t, rawdb.PathScheme)
+}
+
+func testGenerateExistentStateWithWrongAccounts(t *testing.T, scheme string) {
+	helper := newHelper(scheme)
 
 	helper.makeStorageTrie(hashData([]byte("acc-1")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)
 	helper.makeStorageTrie(hashData([]byte("acc-2")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)
@@ -380,10 +408,15 @@ func TestGenerateExistentStateWithWrongAccounts(t *testing.T) {
 // Tests that snapshot generation errors out correctly in case of a missing trie
 // node in the account trie.
 func TestGenerateCorruptAccountTrie(t *testing.T) {
+	testGenerateCorruptAccountTrie(t, rawdb.HashScheme)
+	testGenerateCorruptAccountTrie(t, rawdb.PathScheme)
+}
+
+func testGenerateCorruptAccountTrie(t *testing.T, scheme string) {
 	// We can't use statedb to make a test trie (circular dependency), so make
 	// a fake one manually. We're going with a small account trie of 3 accounts,
 	// without any storage slots to keep the test smaller.
-	helper := newHelper()
+	helper := newHelper(scheme)
 
 	helper.addTrieAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: types.EmptyRootHash, CodeHash: types.EmptyCodeHash.Bytes()}) // 0xc7a30f39aff471c95d8a837497ad0e49b65be475cc0953540f80cfcdbdcd9074
 	helper.addTrieAccount("acc-2", &types.StateAccount{Balance: big.NewInt(2), Root: types.EmptyRootHash, CodeHash: types.EmptyCodeHash.Bytes()}) // 0x65145f923027566669a1ae5ccac66f945b55ff6eaeb17d2ea8e048b7d381f2d7
@@ -391,9 +424,11 @@ func TestGenerateCorruptAccountTrie(t *testing.T) {
 
 	root := helper.Commit() // Root: 0xa04693ea110a31037fb5ee814308a6f1d76bdab0b11676bdf4541d2de55ba978
 
-	// Delete an account trie leaf and ensure the generator chokes
+	// Delete an account trie node and ensure the generator chokes
-	helper.triedb.Commit(root, false)
+	targetPath := []byte{0xc}
-	helper.diskdb.Delete(common.HexToHash("0x65145f923027566669a1ae5ccac66f945b55ff6eaeb17d2ea8e048b7d381f2d7").Bytes())
+	targetHash := common.HexToHash("0x65145f923027566669a1ae5ccac66f945b55ff6eaeb17d2ea8e048b7d381f2d7")
+
+	rawdb.DeleteTrieNode(helper.diskdb, common.Hash{}, targetPath, targetHash, scheme)
 
 	snap := generateSnapshot(helper.diskdb, helper.triedb, 16, root)
 	select {
@@ -414,11 +449,19 @@ func TestGenerateCorruptAccountTrie(t *testing.T) {
 // trie node for a storage trie. It's similar to internal corruption but it is
 // handled differently inside the generator.
 func TestGenerateMissingStorageTrie(t *testing.T) {
+	testGenerateMissingStorageTrie(t, rawdb.HashScheme)
+	testGenerateMissingStorageTrie(t, rawdb.PathScheme)
+}
+
+func testGenerateMissingStorageTrie(t *testing.T, scheme string) {
 	// We can't use statedb to make a test trie (circular dependency), so make
 	// a fake one manually. We're going with a small account trie of 3 accounts,
 	// two of which also has the same 3-slot storage trie attached.
-	helper := newHelper()
+	var (
-
+		acc1   = hashData([]byte("acc-1"))
+		acc3   = hashData([]byte("acc-3"))
+		helper = newHelper(scheme)
+	)
 	stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)   // 0xddefcd9376dd029653ef384bd2f0a126bb755fe84fdcc9e7cf421ba454f2bc67
 	helper.addTrieAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: stRoot, CodeHash: types.EmptyCodeHash.Bytes()})              // 0x9250573b9c18c664139f3b6a7a8081b7d8f8916a8fcc5d94feec6c29f5fd4e9e
 	helper.addTrieAccount("acc-2", &types.StateAccount{Balance: big.NewInt(2), Root: types.EmptyRootHash, CodeHash: types.EmptyCodeHash.Bytes()}) // 0x65145f923027566669a1ae5ccac66f945b55ff6eaeb17d2ea8e048b7d381f2d7
@@ -427,8 +470,9 @@ func TestGenerateMissingStorageTrie(t *testing.T) {
 
 	root := helper.Commit()
 
-	// Delete a storage trie root and ensure the generator chokes
+	// Delete storage trie root of account one and three.
-	helper.diskdb.Delete(stRoot.Bytes())
+	rawdb.DeleteTrieNode(helper.diskdb, acc1, nil, stRoot, scheme)
+	rawdb.DeleteTrieNode(helper.diskdb, acc3, nil, stRoot, scheme)
 
 	snap := generateSnapshot(helper.diskdb, helper.triedb, 16, root)
 	select {
@@ -448,10 +492,15 @@ func TestGenerateMissingStorageTrie(t *testing.T) {
 // Tests that snapshot generation errors out correctly in case of a missing trie
 // node in a storage trie.
 func TestGenerateCorruptStorageTrie(t *testing.T) {
+	testGenerateCorruptStorageTrie(t, rawdb.HashScheme)
+	testGenerateCorruptStorageTrie(t, rawdb.PathScheme)
+}
+
+func testGenerateCorruptStorageTrie(t *testing.T, scheme string) {
 	// We can't use statedb to make a test trie (circular dependency), so make
 	// a fake one manually. We're going with a small account trie of 3 accounts,
 	// two of which also has the same 3-slot storage trie attached.
-	helper := newHelper()
+	helper := newHelper(scheme)
 
 	stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)   // 0xddefcd9376dd029653ef384bd2f0a126bb755fe84fdcc9e7cf421ba454f2bc67
 	helper.addTrieAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: stRoot, CodeHash: types.EmptyCodeHash.Bytes()})              // 0x9250573b9c18c664139f3b6a7a8081b7d8f8916a8fcc5d94feec6c29f5fd4e9e
@@ -461,8 +510,11 @@ func TestGenerateCorruptStorageTrie(t *testing.T) {
 
 	root := helper.Commit()
 
-	// Delete a storage trie leaf and ensure the generator chokes
+	// Delete a node in the storage trie.
-	helper.diskdb.Delete(common.HexToHash("0x18a0f4d79cff4459642dd7604f303886ad9d77c30cf3d7d7cedb3a693ab6d371").Bytes())
+	targetPath := []byte{0x4}
+	targetHash := common.HexToHash("0x18a0f4d79cff4459642dd7604f303886ad9d77c30cf3d7d7cedb3a693ab6d371")
+	rawdb.DeleteTrieNode(helper.diskdb, hashData([]byte("acc-1")), targetPath, targetHash, scheme)
+	rawdb.DeleteTrieNode(helper.diskdb, hashData([]byte("acc-3")), targetPath, targetHash, scheme)
 
 	snap := generateSnapshot(helper.diskdb, helper.triedb, 16, root)
 	select {
@@ -481,7 +533,12 @@ func TestGenerateCorruptStorageTrie(t *testing.T) {
 
 // Tests that snapshot generation when an extra account with storage exists in the snap state.
 func TestGenerateWithExtraAccounts(t *testing.T) {
-	helper := newHelper()
+	testGenerateWithExtraAccounts(t, rawdb.HashScheme)
+	testGenerateWithExtraAccounts(t, rawdb.PathScheme)
+}
+
+func testGenerateWithExtraAccounts(t *testing.T, scheme string) {
+	helper := newHelper(scheme)
 	{
 		// Account one in the trie
 		stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")),
@@ -549,10 +606,15 @@ func enableLogging() {
 
 // Tests that snapshot generation when an extra account with storage exists in the snap state.
 func TestGenerateWithManyExtraAccounts(t *testing.T) {
+	testGenerateWithManyExtraAccounts(t, rawdb.HashScheme)
+	testGenerateWithManyExtraAccounts(t, rawdb.PathScheme)
+}
+
+func testGenerateWithManyExtraAccounts(t *testing.T, scheme string) {
 	if false {
 		enableLogging()
 	}
-	helper := newHelper()
+	helper := newHelper(scheme)
 	{
 		// Account one in the trie
 		stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")),
@@ -605,11 +667,16 @@ func TestGenerateWithManyExtraAccounts(t *testing.T) {
 // So in trie, we iterate 2 entries 0x03, 0x07. We create the 0x07 in the database and abort the procedure, because the trie is exhausted.
 // But in the database, we still have the stale storage slots 0x04, 0x05. They are not iterated yet, but the procedure is finished.
 func TestGenerateWithExtraBeforeAndAfter(t *testing.T) {
+	testGenerateWithExtraBeforeAndAfter(t, rawdb.HashScheme)
+	testGenerateWithExtraBeforeAndAfter(t, rawdb.PathScheme)
+}
+
+func testGenerateWithExtraBeforeAndAfter(t *testing.T, scheme string) {
 	accountCheckRange = 3
 	if false {
 		enableLogging()
 	}
-	helper := newHelper()
+	helper := newHelper(scheme)
 	{
 		acc := &types.StateAccount{Balance: big.NewInt(1), Root: types.EmptyRootHash, CodeHash: types.EmptyCodeHash.Bytes()}
 		val, _ := rlp.EncodeToBytes(acc)
@@ -642,11 +709,16 @@ func TestGenerateWithExtraBeforeAndAfter(t *testing.T) {
 // TestGenerateWithMalformedSnapdata tests what happes if we have some junk
 // in the snapshot database, which cannot be parsed back to an account
 func TestGenerateWithMalformedSnapdata(t *testing.T) {
+	testGenerateWithMalformedSnapdata(t, rawdb.HashScheme)
+	testGenerateWithMalformedSnapdata(t, rawdb.PathScheme)
+}
+
+func testGenerateWithMalformedSnapdata(t *testing.T, scheme string) {
 	accountCheckRange = 3
 	if false {
 		enableLogging()
 	}
-	helper := newHelper()
+	helper := newHelper(scheme)
 	{
 		acc := &types.StateAccount{Balance: big.NewInt(1), Root: types.EmptyRootHash, CodeHash: types.EmptyCodeHash.Bytes()}
 		val, _ := rlp.EncodeToBytes(acc)
@@ -679,10 +751,15 @@ func TestGenerateWithMalformedSnapdata(t *testing.T) {
 }
 
 func TestGenerateFromEmptySnap(t *testing.T) {
+	testGenerateFromEmptySnap(t, rawdb.HashScheme)
+	testGenerateFromEmptySnap(t, rawdb.PathScheme)
+}
+
+func testGenerateFromEmptySnap(t *testing.T, scheme string) {
 	//enableLogging()
 	accountCheckRange = 10
 	storageCheckRange = 20
-	helper := newHelper()
+	helper := newHelper(scheme)
 	// Add 1K accounts to the trie
 	for i := 0; i < 400; i++ {
 		stRoot := helper.makeStorageTrie(hashData([]byte(fmt.Sprintf("acc-%d", i))), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)
@@ -714,8 +791,13 @@ func TestGenerateFromEmptySnap(t *testing.T) {
 // This hits a case where the snap verification passes, but there are more elements in the trie
 // which we must also add.
 func TestGenerateWithIncompleteStorage(t *testing.T) {
+	testGenerateWithIncompleteStorage(t, rawdb.HashScheme)
+	testGenerateWithIncompleteStorage(t, rawdb.PathScheme)
+}
+
+func testGenerateWithIncompleteStorage(t *testing.T, scheme string) {
 	storageCheckRange = 4
-	helper := newHelper()
+	helper := newHelper(scheme)
 	stKeys := []string{"1", "2", "3", "4", "5", "6", "7", "8"}
 	stVals := []string{"v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8"}
 	// We add 8 accounts, each one is missing exactly one of the storage slots. This means
@@ -813,7 +895,12 @@ func populateDangling(disk ethdb.KeyValueStore) {
 //
 // This test will populate some dangling storages to see if they can be cleaned up.
 func TestGenerateCompleteSnapshotWithDanglingStorage(t *testing.T) {
-	var helper = newHelper()
+	testGenerateCompleteSnapshotWithDanglingStorage(t, rawdb.HashScheme)
+	testGenerateCompleteSnapshotWithDanglingStorage(t, rawdb.PathScheme)
+}
+
+func testGenerateCompleteSnapshotWithDanglingStorage(t *testing.T, scheme string) {
+	var helper = newHelper(scheme)
 
 	stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)
 	helper.addAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: stRoot, CodeHash: types.EmptyCodeHash.Bytes()})
@@ -848,7 +935,12 @@ func TestGenerateCompleteSnapshotWithDanglingStorage(t *testing.T) {
 //
 // This test will populate some dangling storages to see if they can be cleaned up.
 func TestGenerateBrokenSnapshotWithDanglingStorage(t *testing.T) {
-	var helper = newHelper()
+	testGenerateBrokenSnapshotWithDanglingStorage(t, rawdb.HashScheme)
+	testGenerateBrokenSnapshotWithDanglingStorage(t, rawdb.PathScheme)
+}
+
+func testGenerateBrokenSnapshotWithDanglingStorage(t *testing.T, scheme string) {
+	var helper = newHelper(scheme)
 
 	stRoot := helper.makeStorageTrie(hashData([]byte("acc-1")), []string{"key-1", "key-2", "key-3"}, []string{"val-1", "val-2", "val-3"}, true)
 	helper.addTrieAccount("acc-1", &types.StateAccount{Balance: big.NewInt(1), Root: stRoot, CodeHash: types.EmptyCodeHash.Bytes()})

core/state/statedb_fuzz_test.go

@@ -179,7 +179,7 @@ func (test *stateTest) run() bool {
 			storageList = append(storageList, copy2DSet(states.Storages))
 		}
 		disk      = rawdb.NewMemoryDatabase()
-		tdb       = trie.NewDatabaseWithConfig(disk, &trie.Config{OnCommit: onCommit})
+		tdb       = trie.NewDatabase(disk, &trie.Config{OnCommit: onCommit})
 		sdb       = NewDatabaseWithNodeDB(disk, tdb)
 		byzantium = rand.Intn(2) == 0
 	)

core/state/statedb_test.go

@@ -36,14 +36,19 @@ import (
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 )
 
 // Tests that updating a state trie does not leak any database writes prior to
 // actually committing the state.
 func TestUpdateLeaks(t *testing.T) {
 	// Create an empty state database
-	db := rawdb.NewMemoryDatabase()
+	var (
-	state, _ := New(types.EmptyRootHash, NewDatabase(db), nil)
+		db  = rawdb.NewMemoryDatabase()
+		tdb = trie.NewDatabase(db, nil)
+	)
+	state, _ := New(types.EmptyRootHash, NewDatabaseWithNodeDB(db, tdb), nil)
 
 	// Update it with some accounts
 	for i := byte(0); i < 255; i++ {
@@ -59,7 +64,7 @@ func TestUpdateLeaks(t *testing.T) {
 	}
 
 	root := state.IntermediateRoot(false)
-	if err := state.Database().TrieDB().Commit(root, false); err != nil {
+	if err := tdb.Commit(root, false); err != nil {
 		t.Errorf("can not commit trie %v to persistent database", root.Hex())
 	}
 
@@ -77,8 +82,10 @@ func TestIntermediateLeaks(t *testing.T) {
 	// Create two state databases, one transitioning to the final state, the other final from the beginning
 	transDb := rawdb.NewMemoryDatabase()
 	finalDb := rawdb.NewMemoryDatabase()
-	transState, _ := New(types.EmptyRootHash, NewDatabase(transDb), nil)
+	transNdb := trie.NewDatabase(transDb, nil)
-	finalState, _ := New(types.EmptyRootHash, NewDatabase(finalDb), nil)
+	finalNdb := trie.NewDatabase(finalDb, nil)
+	transState, _ := New(types.EmptyRootHash, NewDatabaseWithNodeDB(transDb, transNdb), nil)
+	finalState, _ := New(types.EmptyRootHash, NewDatabaseWithNodeDB(finalDb, finalNdb), nil)
 
 	modify := func(state *StateDB, addr common.Address, i, tweak byte) {
 		state.SetBalance(addr, big.NewInt(int64(11*i)+int64(tweak)))
@@ -110,7 +117,7 @@ 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); err != nil {
+	if err = transNdb.Commit(transRoot, false); err != nil {
 		t.Errorf("can not commit trie %v to persistent database", transRoot.Hex())
 	}
 
@@ -118,7 +125,7 @@ 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); err != nil {
+	if err = finalNdb.Commit(finalRoot, false); err != nil {
 		t.Errorf("can not commit trie %v to persistent database", finalRoot.Hex())
 	}
 
@@ -747,9 +754,28 @@ func TestDeleteCreateRevert(t *testing.T) {
 // the Commit operation fails with an error
 // If we are missing trie nodes, we should not continue writing to the trie
 func TestMissingTrieNodes(t *testing.T) {
+	testMissingTrieNodes(t, rawdb.HashScheme)
+	testMissingTrieNodes(t, rawdb.PathScheme)
+}
+
+func testMissingTrieNodes(t *testing.T, scheme string) {
 	// Create an initial state with a few accounts
-	memDb := rawdb.NewMemoryDatabase()
+	var (
-	db := NewDatabase(memDb)
+		triedb *trie.Database
+		memDb  = rawdb.NewMemoryDatabase()
+	)
+	if scheme == rawdb.PathScheme {
+		triedb = trie.NewDatabase(memDb, &trie.Config{PathDB: &pathdb.Config{
+			CleanCacheSize: 0,
+			DirtyCacheSize: 0,
+		}}) // disable caching
+	} else {
+		triedb = trie.NewDatabase(memDb, &trie.Config{HashDB: &hashdb.Config{
+			CleanCacheSize: 0,
+		}}) // disable caching
+	}
+	db := NewDatabaseWithNodeDB(memDb, triedb)
+
 	var root common.Hash
 	state, _ := New(types.EmptyRootHash, db, nil)
 	addr := common.BytesToAddress([]byte("so"))
@@ -762,7 +788,7 @@ func TestMissingTrieNodes(t *testing.T) {
 		root, _ = state.Commit(0, false)
 		t.Logf("root: %x", root)
 		// force-flush
-		state.Database().TrieDB().Cap(0)
+		triedb.Commit(root, false)
 	}
 	// Create a new state on the old root
 	state, _ = New(root, db, nil)
@@ -969,7 +995,8 @@ func TestFlushOrderDataLoss(t *testing.T) {
 	// Create a state trie with many accounts and slots
 	var (
 		memdb    = rawdb.NewMemoryDatabase()
-		statedb  = NewDatabase(memdb)
+		triedb   = trie.NewDatabase(memdb, nil)
+		statedb  = NewDatabaseWithNodeDB(memdb, triedb)
 		state, _ = New(types.EmptyRootHash, statedb, nil)
 	)
 	for a := byte(0); a < 10; a++ {
@@ -982,11 +1009,11 @@ func TestFlushOrderDataLoss(t *testing.T) {
 	if err != nil {
 		t.Fatalf("failed to commit state trie: %v", err)
 	}
-	statedb.TrieDB().Reference(root, common.Hash{})
+	triedb.Reference(root, common.Hash{})
-	if err := statedb.TrieDB().Cap(1024); err != nil {
+	if err := triedb.Cap(1024); err != nil {
 		t.Fatalf("failed to cap trie dirty cache: %v", err)
 	}
-	if err := statedb.TrieDB().Commit(root, false); err != nil {
+	if err := 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
@@ -1040,7 +1067,7 @@ func TestStateDBTransientStorage(t *testing.T) {
 func TestResetObject(t *testing.T) {
 	var (
 		disk     = rawdb.NewMemoryDatabase()
-		tdb      = trie.NewDatabase(disk)
+		tdb      = trie.NewDatabase(disk, nil)
 		db       = NewDatabaseWithNodeDB(disk, tdb)
 		snaps, _ = snapshot.New(snapshot.Config{CacheSize: 10}, disk, tdb, types.EmptyRootHash)
 		state, _ = New(types.EmptyRootHash, db, snaps)

core/state/sync_test.go

@@ -28,6 +28,8 @@ import (
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 )
 
 // testAccount is the data associated with an account used by the state tests.
@@ -39,10 +41,17 @@ type testAccount struct {
 }
 
 // makeTestState create a sample test state to test node-wise reconstruction.
-func makeTestState() (ethdb.Database, Database, common.Hash, []*testAccount) {
+func makeTestState(scheme string) (ethdb.Database, Database, *trie.Database, common.Hash, []*testAccount) {
 	// Create an empty state
+	config := &trie.Config{Preimages: true}
+	if scheme == rawdb.PathScheme {
+		config.PathDB = pathdb.Defaults
+	} else {
+		config.HashDB = hashdb.Defaults
+	}
 	db := rawdb.NewMemoryDatabase()
-	sdb := NewDatabaseWithConfig(db, &trie.Config{Preimages: true})
+	nodeDb := trie.NewDatabase(db, config)
+	sdb := NewDatabaseWithNodeDB(db, nodeDb)
 	state, _ := New(types.EmptyRootHash, sdb, nil)
 
 	// Fill it with some arbitrary data
@@ -67,24 +76,27 @@ func makeTestState() (ethdb.Database, Database, common.Hash, []*testAccount) {
 				obj.SetState(hash, hash)
 			}
 		}
-		state.updateStateObject(obj)
 		accounts = append(accounts, acc)
 	}
 	root, _ := state.Commit(0, false)
 
 	// Return the generated state
-	return db, sdb, root, accounts
+	return db, sdb, nodeDb, root, accounts
 }
 
 // checkStateAccounts cross references a reconstructed state with an expected
 // account array.
-func checkStateAccounts(t *testing.T, db ethdb.Database, root common.Hash, accounts []*testAccount) {
+func checkStateAccounts(t *testing.T, db ethdb.Database, scheme string, root common.Hash, accounts []*testAccount) {
+	var config trie.Config
+	if scheme == rawdb.PathScheme {
+		config.PathDB = pathdb.Defaults
+	}
 	// Check root availability and state contents
-	state, err := New(root, NewDatabase(db), nil)
+	state, err := New(root, NewDatabaseWithConfig(db, &config), nil)
 	if err != nil {
 		t.Fatalf("failed to create state trie at %x: %v", root, err)
 	}
-	if err := checkStateConsistency(db, root); err != nil {
+	if err := checkStateConsistency(db, scheme, root); err != nil {
 		t.Fatalf("inconsistent state trie at %x: %v", root, err)
 	}
 	for i, acc := range accounts {
@@ -101,8 +113,12 @@ func checkStateAccounts(t *testing.T, db ethdb.Database, root common.Hash, accou
 }
 
 // checkStateConsistency checks that all data of a state root is present.
-func checkStateConsistency(db ethdb.Database, root common.Hash) error {
+func checkStateConsistency(db ethdb.Database, scheme string, root common.Hash) error {
-	state, err := New(root, NewDatabaseWithConfig(db, &trie.Config{Preimages: true}), nil)
+	config := &trie.Config{Preimages: true}
+	if scheme == rawdb.PathScheme {
+		config.PathDB = pathdb.Defaults
+	}
+	state, err := New(root, NewDatabaseWithConfig(db, config), nil)
 	if err != nil {
 		return err
 	}
@@ -114,8 +130,14 @@ 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())
+	dbA := trie.NewDatabase(rawdb.NewMemoryDatabase(), nil)
-	sync := NewStateSync(types.EmptyRootHash, rawdb.NewMemoryDatabase(), nil, db.Scheme())
+	dbB := trie.NewDatabase(rawdb.NewMemoryDatabase(), &trie.Config{PathDB: pathdb.Defaults})
+
+	sync := NewStateSync(types.EmptyRootHash, rawdb.NewMemoryDatabase(), nil, dbA.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)
+	}
+	sync = NewStateSync(types.EmptyRootHash, rawdb.NewMemoryDatabase(), nil, dbB.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)
 	}
@@ -124,22 +146,28 @@ func TestEmptyStateSync(t *testing.T) {
 // Tests that given a root hash, a state can sync iteratively on a single thread,
 // requesting retrieval tasks and returning all of them in one go.
 func TestIterativeStateSyncIndividual(t *testing.T) {
-	testIterativeStateSync(t, 1, false, false)
+	testIterativeStateSync(t, 1, false, false, rawdb.HashScheme)
+	testIterativeStateSync(t, 1, false, false, rawdb.PathScheme)
 }
 func TestIterativeStateSyncBatched(t *testing.T) {
-	testIterativeStateSync(t, 100, false, false)
+	testIterativeStateSync(t, 100, false, false, rawdb.HashScheme)
+	testIterativeStateSync(t, 100, false, false, rawdb.PathScheme)
 }
 func TestIterativeStateSyncIndividualFromDisk(t *testing.T) {
-	testIterativeStateSync(t, 1, true, false)
+	testIterativeStateSync(t, 1, true, false, rawdb.HashScheme)
+	testIterativeStateSync(t, 1, true, false, rawdb.PathScheme)
 }
 func TestIterativeStateSyncBatchedFromDisk(t *testing.T) {
-	testIterativeStateSync(t, 100, true, false)
+	testIterativeStateSync(t, 100, true, false, rawdb.HashScheme)
+	testIterativeStateSync(t, 100, true, false, rawdb.PathScheme)
 }
 func TestIterativeStateSyncIndividualByPath(t *testing.T) {
-	testIterativeStateSync(t, 1, false, true)
+	testIterativeStateSync(t, 1, false, true, rawdb.HashScheme)
+	testIterativeStateSync(t, 1, false, true, rawdb.PathScheme)
 }
 func TestIterativeStateSyncBatchedByPath(t *testing.T) {
-	testIterativeStateSync(t, 100, false, true)
+	testIterativeStateSync(t, 100, false, true, rawdb.HashScheme)
+	testIterativeStateSync(t, 100, false, true, rawdb.PathScheme)
 }
 
 // stateElement represents the element in the state trie(bytecode or trie node).
@@ -150,17 +178,17 @@ type stateElement struct {
 	syncPath trie.SyncPath
 }
 
-func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
+func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool, scheme string) {
 	// Create a random state to copy
-	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, ndb, srcRoot, srcAccounts := makeTestState(scheme)
 	if commit {
-		srcDb.TrieDB().Commit(srcRoot, false)
+		ndb.Commit(srcRoot, false)
 	}
-	srcTrie, _ := trie.New(trie.StateTrieID(srcRoot), srcDb.TrieDB())
+	srcTrie, _ := trie.New(trie.StateTrieID(srcRoot), ndb)
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
+	sched := NewStateSync(srcRoot, dstDb, nil, ndb.Scheme())
 
 	var (
 		nodeElements []stateElement
@@ -175,9 +203,11 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 		})
 	}
 	for i := 0; i < len(codes); i++ {
-		codeElements = append(codeElements, stateElement{
+		codeElements = append(codeElements, stateElement{code: codes[i]})
-			code: codes[i],
+	}
-		})
+	reader, err := ndb.Reader(srcRoot)
+	if err != nil {
+		t.Fatalf("state is not existent, %#x", srcRoot)
 	}
 	for len(nodeElements)+len(codeElements) > 0 {
 		var (
@@ -205,7 +235,7 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 						t.Fatalf("failed to decode account on path %x: %v", node.syncPath[0], err)
 					}
 					id := trie.StorageTrieID(srcRoot, common.BytesToHash(node.syncPath[0]), acc.Root)
-					stTrie, err := trie.New(id, srcDb.TrieDB())
+					stTrie, err := trie.New(id, ndb)
 					if err != nil {
 						t.Fatalf("failed to retriev storage trie for path %x: %v", node.syncPath[1], err)
 					}
@@ -216,7 +246,8 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 					nodeResults[i] = trie.NodeSyncResult{Path: node.path, Data: data}
 				}
 			} else {
-				data, err := srcDb.TrieDB().Node(node.hash)
+				owner, inner := trie.ResolvePath([]byte(node.path))
+				data, err := reader.Node(owner, inner, node.hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for key %v", []byte(node.path))
 				}
@@ -260,18 +291,23 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 	copyPreimages(srcDisk, dstDb)
 
 	// Cross check that the two states are in sync
-	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
+	checkStateAccounts(t, dstDb, ndb.Scheme(), srcRoot, srcAccounts)
 }
 
 // Tests that the trie scheduler can correctly reconstruct the state even if only
 // partial results are returned, and the others sent only later.
 func TestIterativeDelayedStateSync(t *testing.T) {
+	testIterativeDelayedStateSync(t, rawdb.HashScheme)
+	testIterativeDelayedStateSync(t, rawdb.PathScheme)
+}
+
+func testIterativeDelayedStateSync(t *testing.T, scheme string) {
 	// Create a random state to copy
-	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, ndb, srcRoot, srcAccounts := makeTestState(scheme)
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
+	sched := NewStateSync(srcRoot, dstDb, nil, ndb.Scheme())
 
 	var (
 		nodeElements []stateElement
@@ -286,9 +322,11 @@ func TestIterativeDelayedStateSync(t *testing.T) {
 		})
 	}
 	for i := 0; i < len(codes); i++ {
-		codeElements = append(codeElements, stateElement{
+		codeElements = append(codeElements, stateElement{code: codes[i]})
-			code: codes[i],
+	}
-		})
+	reader, err := ndb.Reader(srcRoot)
+	if err != nil {
+		t.Fatalf("state is not existent, %#x", srcRoot)
 	}
 	for len(nodeElements)+len(codeElements) > 0 {
 		// Sync only half of the scheduled nodes
@@ -313,7 +351,8 @@ func TestIterativeDelayedStateSync(t *testing.T) {
 		if len(nodeElements) > 0 {
 			nodeResults := make([]trie.NodeSyncResult, len(nodeElements)/2+1)
 			for i, element := range nodeElements[:len(nodeResults)] {
-				data, err := srcDb.TrieDB().Node(element.hash)
+				owner, inner := trie.ResolvePath([]byte(element.path))
+				data, err := reader.Node(owner, inner, element.hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve contract bytecode for %x", element.code)
 				}
@@ -353,22 +392,28 @@ func TestIterativeDelayedStateSync(t *testing.T) {
 	copyPreimages(srcDisk, dstDb)
 
 	// Cross check that the two states are in sync
-	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
+	checkStateAccounts(t, dstDb, ndb.Scheme(), srcRoot, srcAccounts)
 }
 
 // Tests that given a root hash, a trie can sync iteratively on a single thread,
 // requesting retrieval tasks and returning all of them in one go, however in a
 // random order.
-func TestIterativeRandomStateSyncIndividual(t *testing.T) { testIterativeRandomStateSync(t, 1) }
+func TestIterativeRandomStateSyncIndividual(t *testing.T) {
-func TestIterativeRandomStateSyncBatched(t *testing.T)    { testIterativeRandomStateSync(t, 100) }
+	testIterativeRandomStateSync(t, 1, rawdb.HashScheme)
+	testIterativeRandomStateSync(t, 1, rawdb.PathScheme)
+}
+func TestIterativeRandomStateSyncBatched(t *testing.T) {
+	testIterativeRandomStateSync(t, 100, rawdb.HashScheme)
+	testIterativeRandomStateSync(t, 100, rawdb.PathScheme)
+}
 
-func testIterativeRandomStateSync(t *testing.T, count int) {
+func testIterativeRandomStateSync(t *testing.T, count int, scheme string) {
 	// Create a random state to copy
-	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, ndb, srcRoot, srcAccounts := makeTestState(scheme)
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
+	sched := NewStateSync(srcRoot, dstDb, nil, ndb.Scheme())
 
 	nodeQueue := make(map[string]stateElement)
 	codeQueue := make(map[common.Hash]struct{})
@@ -383,6 +428,10 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 	for _, hash := range codes {
 		codeQueue[hash] = struct{}{}
 	}
+	reader, err := ndb.Reader(srcRoot)
+	if err != nil {
+		t.Fatalf("state is not existent, %#x", srcRoot)
+	}
 	for len(nodeQueue)+len(codeQueue) > 0 {
 		// Fetch all the queued nodes in a random order
 		if len(codeQueue) > 0 {
@@ -403,7 +452,8 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 		if len(nodeQueue) > 0 {
 			results := make([]trie.NodeSyncResult, 0, len(nodeQueue))
 			for path, element := range nodeQueue {
-				data, err := srcDb.TrieDB().Node(element.hash)
+				owner, inner := trie.ResolvePath([]byte(element.path))
+				data, err := reader.Node(owner, inner, element.hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for %x %v %v", element.hash, []byte(element.path), element.path)
 				}
@@ -415,7 +465,6 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 				}
 			}
 		}
-		// Feed the retrieved results back and queue new tasks
 		batch := dstDb.NewBatch()
 		if err := sched.Commit(batch); err != nil {
 			t.Fatalf("failed to commit data: %v", err)
@@ -441,18 +490,23 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 	copyPreimages(srcDisk, dstDb)
 
 	// Cross check that the two states are in sync
-	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
+	checkStateAccounts(t, dstDb, ndb.Scheme(), srcRoot, srcAccounts)
 }
 
 // Tests that the trie scheduler can correctly reconstruct the state even if only
 // partial results are returned (Even those randomly), others sent only later.
 func TestIterativeRandomDelayedStateSync(t *testing.T) {
+	testIterativeRandomDelayedStateSync(t, rawdb.HashScheme)
+	testIterativeRandomDelayedStateSync(t, rawdb.PathScheme)
+}
+
+func testIterativeRandomDelayedStateSync(t *testing.T, scheme string) {
 	// Create a random state to copy
-	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, ndb, srcRoot, srcAccounts := makeTestState(scheme)
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
+	sched := NewStateSync(srcRoot, dstDb, nil, ndb.Scheme())
 
 	nodeQueue := make(map[string]stateElement)
 	codeQueue := make(map[common.Hash]struct{})
@@ -467,6 +521,10 @@ func TestIterativeRandomDelayedStateSync(t *testing.T) {
 	for _, hash := range codes {
 		codeQueue[hash] = struct{}{}
 	}
+	reader, err := ndb.Reader(srcRoot)
+	if err != nil {
+		t.Fatalf("state is not existent, %#x", srcRoot)
+	}
 	for len(nodeQueue)+len(codeQueue) > 0 {
 		// Sync only half of the scheduled nodes, even those in random order
 		if len(codeQueue) > 0 {
@@ -495,7 +553,8 @@ func TestIterativeRandomDelayedStateSync(t *testing.T) {
 			for path, element := range nodeQueue {
 				delete(nodeQueue, path)
 
-				data, err := srcDb.TrieDB().Node(element.hash)
+				owner, inner := trie.ResolvePath([]byte(element.path))
+				data, err := reader.Node(owner, inner, element.hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for %x", element.hash)
 				}
@@ -535,14 +594,19 @@ func TestIterativeRandomDelayedStateSync(t *testing.T) {
 	copyPreimages(srcDisk, dstDb)
 
 	// Cross check that the two states are in sync
-	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
+	checkStateAccounts(t, dstDb, ndb.Scheme(), srcRoot, srcAccounts)
 }
 
 // Tests that at any point in time during a sync, only complete sub-tries are in
 // the database.
 func TestIncompleteStateSync(t *testing.T) {
+	testIncompleteStateSync(t, rawdb.HashScheme)
+	testIncompleteStateSync(t, rawdb.PathScheme)
+}
+
+func testIncompleteStateSync(t *testing.T, scheme string) {
 	// Create a random state to copy
-	db, srcDb, srcRoot, srcAccounts := makeTestState()
+	db, srcDb, ndb, srcRoot, srcAccounts := makeTestState(scheme)
 
 	// isCodeLookup to save some hashing
 	var isCode = make(map[common.Hash]struct{})
@@ -555,14 +619,14 @@ func TestIncompleteStateSync(t *testing.T) {
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
-	sched := NewStateSync(srcRoot, dstDb, nil, srcDb.TrieDB().Scheme())
+	sched := NewStateSync(srcRoot, dstDb, nil, ndb.Scheme())
 
 	var (
 		addedCodes  []common.Hash
 		addedPaths  []string
 		addedHashes []common.Hash
 	)
-	reader, err := srcDb.TrieDB().Reader(srcRoot)
+	reader, err := ndb.Reader(srcRoot)
 	if err != nil {
 		t.Fatalf("state is not available %x", srcRoot)
 	}
@@ -649,12 +713,11 @@ func TestIncompleteStateSync(t *testing.T) {
 	for _, node := range addedCodes {
 		val := rawdb.ReadCode(dstDb, node)
 		rawdb.DeleteCode(dstDb, node)
-		if err := checkStateConsistency(dstDb, srcRoot); err == nil {
+		if err := checkStateConsistency(dstDb, ndb.Scheme(), srcRoot); err == nil {
 			t.Errorf("trie inconsistency not caught, missing: %x", node)
 		}
 		rawdb.WriteCode(dstDb, node, val)
 	}
-	scheme := srcDb.TrieDB().Scheme()
 	for i, path := range addedPaths {
 		owner, inner := trie.ResolvePath([]byte(path))
 		hash := addedHashes[i]
@@ -663,7 +726,7 @@ func TestIncompleteStateSync(t *testing.T) {
 			t.Error("missing trie node")
 		}
 		rawdb.DeleteTrieNode(dstDb, owner, inner, hash, scheme)
-		if err := checkStateConsistency(dstDb, srcRoot); err == nil {
+		if err := checkStateConsistency(dstDb, scheme, srcRoot); err == nil {
 			t.Errorf("trie inconsistency not caught, missing: %v", path)
 		}
 		rawdb.WriteTrieNode(dstDb, owner, inner, hash, val, scheme)

core/types/hashing_test.go

@@ -39,7 +39,7 @@ func TestDeriveSha(t *testing.T) {
 		t.Fatal(err)
 	}
 	for len(txs) < 1000 {
-		exp := types.DeriveSha(txs, trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase())))
+		exp := types.DeriveSha(txs, trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), nil)))
 		got := types.DeriveSha(txs, trie.NewStackTrie(nil))
 		if !bytes.Equal(got[:], exp[:]) {
 			t.Fatalf("%d txs: got %x exp %x", len(txs), got, exp)
@@ -86,7 +86,7 @@ func BenchmarkDeriveSha200(b *testing.B) {
 		b.ResetTimer()
 		b.ReportAllocs()
 		for i := 0; i < b.N; i++ {
-			exp = types.DeriveSha(txs, trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase())))
+			exp = types.DeriveSha(txs, trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), nil)))
 		}
 	})
 
@@ -107,7 +107,7 @@ func TestFuzzDeriveSha(t *testing.T) {
 	rndSeed := mrand.Int()
 	for i := 0; i < 10; i++ {
 		seed := rndSeed + i
-		exp := types.DeriveSha(newDummy(i), trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase())))
+		exp := types.DeriveSha(newDummy(i), trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), nil)))
 		got := types.DeriveSha(newDummy(i), trie.NewStackTrie(nil))
 		if !bytes.Equal(got[:], exp[:]) {
 			printList(newDummy(seed))
@@ -135,7 +135,7 @@ func TestDerivableList(t *testing.T) {
 		},
 	}
 	for i, tc := range tcs[1:] {
-		exp := types.DeriveSha(flatList(tc), trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase())))
+		exp := types.DeriveSha(flatList(tc), trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), nil)))
 		got := types.DeriveSha(flatList(tc), trie.NewStackTrie(nil))
 		if !bytes.Equal(got[:], exp[:]) {
 			t.Fatalf("case %d: got %x exp %x", i, got, exp)

eth/api_backend.go

@@ -418,7 +418,7 @@ func (b *EthAPIBackend) StartMining() error {
 }
 
 func (b *EthAPIBackend) StateAtBlock(ctx context.Context, block *types.Block, reexec uint64, base *state.StateDB, readOnly bool, preferDisk bool) (*state.StateDB, tracers.StateReleaseFunc, error) {
-	return b.eth.StateAtBlock(ctx, block, reexec, base, readOnly, preferDisk)
+	return b.eth.stateAtBlock(ctx, block, reexec, base, readOnly, preferDisk)
 }
 
 func (b *EthAPIBackend) StateAtTransaction(ctx context.Context, block *types.Block, txIndex int, reexec uint64) (*core.Message, vm.BlockContext, *state.StateDB, tracers.StateReleaseFunc, error) {

eth/api_debug.go

@@ -322,7 +322,7 @@ func (api *DebugAPI) getModifiedAccounts(startBlock, endBlock *types.Block) ([]c
 	if startBlock.Number().Uint64() >= endBlock.Number().Uint64() {
 		return nil, fmt.Errorf("start block height (%d) must be less than end block height (%d)", startBlock.Number().Uint64(), endBlock.Number().Uint64())
 	}
-	triedb := api.eth.BlockChain().StateCache().TrieDB()
+	triedb := api.eth.BlockChain().TrieDB()
 
 	oldTrie, err := trie.NewStateTrie(trie.StateTrieID(startBlock.Root()), triedb)
 	if err != nil {

eth/backend.go

@@ -133,9 +133,12 @@ func New(stack *node.Node, config *ethconfig.Config) (*Ethereum, error) {
 	if err != nil {
 		return nil, err
 	}
+	// Try to recover offline state pruning only in hash-based.
+	if config.StateScheme == rawdb.HashScheme {
 		if err := pruner.RecoverPruning(stack.ResolvePath(""), chainDb); err != nil {
 			log.Error("Failed to recover state", "error", err)
 		}
+	}
 	// Transfer mining-related config to the ethash config.
 	chainConfig, err := core.LoadChainConfig(chainDb, config.Genesis)
 	if err != nil {
@@ -161,7 +164,6 @@ func New(stack *node.Node, config *ethconfig.Config) (*Ethereum, error) {
 		p2pServer:         stack.Server(),
 		shutdownTracker:   shutdowncheck.NewShutdownTracker(chainDb),
 	}
-
 	bcVersion := rawdb.ReadDatabaseVersion(chainDb)
 	var dbVer = "<nil>"
 	if bcVersion != nil {
@@ -191,6 +193,8 @@ func New(stack *node.Node, config *ethconfig.Config) (*Ethereum, error) {
 			TrieTimeLimit:       config.TrieTimeout,
 			SnapshotLimit:       config.SnapshotCache,
 			Preimages:           config.Preimages,
+			StateHistory:        config.StateHistory,
+			StateScheme:         config.StateScheme,
 		}
 	)
 	// Override the chain config with provided settings.
@@ -201,7 +205,7 @@ func New(stack *node.Node, config *ethconfig.Config) (*Ethereum, error) {
 	if config.OverrideVerkle != nil {
 		overrides.OverrideVerkle = config.OverrideVerkle
 	}
-	eth.blockchain, err = core.NewBlockChain(chainDb, cacheConfig, config.Genesis, &overrides, eth.engine, vmConfig, eth.shouldPreserve, &config.TxLookupLimit)
+	eth.blockchain, err = core.NewBlockChain(chainDb, cacheConfig, config.Genesis, &overrides, eth.engine, vmConfig, eth.shouldPreserve, &config.TransactionHistory)
 	if err != nil {
 		return nil, err
 	}
@@ -438,7 +442,7 @@ func (s *Ethereum) StartMining() error {
 		}
 		// If mining is started, we can disable the transaction rejection mechanism
 		// introduced to speed sync times.
-		s.handler.acceptTxs.Store(true)
+		s.handler.enableSyncedFeatures()
 
 		go s.miner.Start()
 	}
@@ -471,7 +475,7 @@ func (s *Ethereum) ChainDb() ethdb.Database            { return s.chainDb }
 func (s *Ethereum) IsListening() bool                  { return true } // Always listening
 func (s *Ethereum) Downloader() *downloader.Downloader { return s.handler.downloader }
 func (s *Ethereum) Synced() bool                       { return s.handler.acceptTxs.Load() }
-func (s *Ethereum) SetSynced()                         { s.handler.acceptTxs.Store(true) }
+func (s *Ethereum) SetSynced()                         { s.handler.enableSyncedFeatures() }
 func (s *Ethereum) ArchiveMode() bool                  { return s.config.NoPruning }
 func (s *Ethereum) BloomIndexer() *core.ChainIndexer   { return s.bloomIndexer }
 func (s *Ethereum) Merger() *consensus.Merger          { return s.merger }

eth/downloader/testchain_test.go

@@ -30,6 +30,7 @@ import (
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/params"
+	"github.com/ethereum/go-ethereum/trie"
 )
 
 // Test chain parameters.
@@ -43,7 +44,7 @@ var (
 		Alloc:   core.GenesisAlloc{testAddress: {Balance: big.NewInt(1000000000000000)}},
 		BaseFee: big.NewInt(params.InitialBaseFee),
 	}
-	testGenesis = testGspec.MustCommit(testDB)
+	testGenesis = testGspec.MustCommit(testDB, trie.NewDatabase(testDB, trie.HashDefaults))
 )
 
 // The common prefix of all test chains:

eth/ethconfig/config.go

@@ -27,6 +27,7 @@ import (
 	"github.com/ethereum/go-ethereum/consensus/clique"
 	"github.com/ethereum/go-ethereum/consensus/ethash"
 	"github.com/ethereum/go-ethereum/core"
+	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/txpool/blobpool"
 	"github.com/ethereum/go-ethereum/core/txpool/legacypool"
 	"github.com/ethereum/go-ethereum/eth/downloader"
@@ -61,6 +62,9 @@ var Defaults = Config{
 	SyncMode:           downloader.SnapSync,
 	NetworkId:          1,
 	TxLookupLimit:      2350000,
+	TransactionHistory: 2350000,
+	StateHistory:       params.FullImmutabilityThreshold,
+	StateScheme:        rawdb.HashScheme,
 	LightPeers:         100,
 	DatabaseCache:      512,
 	TrieCleanCache:     154,
@@ -97,7 +101,11 @@ type Config struct {
 	NoPruning  bool // Whether to disable pruning and flush everything to disk
 	NoPrefetch bool // Whether to disable prefetching and only load state on demand
 
+	// Deprecated, use 'TransactionHistory' instead.
 	TxLookupLimit      uint64 `toml:",omitempty"` // The maximum number of blocks from head whose tx indices are reserved.
+	TransactionHistory uint64 `toml:",omitempty"` // The maximum number of blocks from head whose tx indices are reserved.
+	StateHistory       uint64 `toml:",omitempty"` // The maximum number of blocks from head whose state histories are reserved.
+	StateScheme        string `toml:",omitempty"` // State scheme used to store ethereum state and merkle trie nodes on top
 
 	// RequiredBlocks is a set of block number -> hash mappings which must be in the
 	// canonical chain of all remote peers. Setting the option makes geth verify the

eth/ethconfig/gen_config.go

@@ -25,6 +25,9 @@ func (c Config) MarshalTOML() (interface{}, error) {
 		NoPruning               bool
 		NoPrefetch              bool
 		TxLookupLimit           uint64                 `toml:",omitempty"`
+		TransactionHistory      uint64                 `toml:",omitempty"`
+		StateHistory            uint64                 `toml:",omitempty"`
+		StateScheme             string                 `toml:",omitempty"`
 		RequiredBlocks          map[uint64]common.Hash `toml:"-"`
 		LightServ               int                    `toml:",omitempty"`
 		LightIngress            int                    `toml:",omitempty"`
@@ -63,6 +66,9 @@ func (c Config) MarshalTOML() (interface{}, error) {
 	enc.NoPruning = c.NoPruning
 	enc.NoPrefetch = c.NoPrefetch
 	enc.TxLookupLimit = c.TxLookupLimit
+	enc.TransactionHistory = c.TransactionHistory
+	enc.StateHistory = c.StateHistory
+	enc.StateScheme = c.StateScheme
 	enc.RequiredBlocks = c.RequiredBlocks
 	enc.LightServ = c.LightServ
 	enc.LightIngress = c.LightIngress
@@ -105,6 +111,9 @@ func (c *Config) UnmarshalTOML(unmarshal func(interface{}) error) error {
 		NoPruning               *bool
 		NoPrefetch              *bool
 		TxLookupLimit           *uint64                `toml:",omitempty"`
+		TransactionHistory      *uint64                `toml:",omitempty"`
+		StateHistory            *uint64                `toml:",omitempty"`
+		StateScheme             *string                `toml:",omitempty"`
 		RequiredBlocks          map[uint64]common.Hash `toml:"-"`
 		LightServ               *int                   `toml:",omitempty"`
 		LightIngress            *int                   `toml:",omitempty"`
@@ -162,6 +171,15 @@ func (c *Config) UnmarshalTOML(unmarshal func(interface{}) error) error {
 	if dec.TxLookupLimit != nil {
 		c.TxLookupLimit = *dec.TxLookupLimit
 	}
+	if dec.TransactionHistory != nil {
+		c.TransactionHistory = *dec.TransactionHistory
+	}
+	if dec.StateHistory != nil {
+		c.StateHistory = *dec.StateHistory
+	}
+	if dec.StateScheme != nil {
+		c.StateScheme = *dec.StateScheme
+	}
 	if dec.RequiredBlocks != nil {
 		c.RequiredBlocks = dec.RequiredBlocks
 	}

eth/fetcher/block_fetcher_test.go

@@ -44,7 +44,7 @@ var (
 		Alloc:   core.GenesisAlloc{testAddress: {Balance: big.NewInt(1000000000000000)}},
 		BaseFee: big.NewInt(params.InitialBaseFee),
 	}
-	genesis      = gspec.MustCommit(testdb)
+	genesis      = gspec.MustCommit(testdb, trie.NewDatabase(testdb, trie.HashDefaults))
 	unknownBlock = types.NewBlock(&types.Header{Root: types.EmptyRootHash, GasLimit: params.GenesisGasLimit, BaseFee: big.NewInt(params.InitialBaseFee)}, nil, nil, nil, trie.NewStackTrie(nil))
 )
 

eth/filters/filter_test.go

@@ -86,7 +86,7 @@ func BenchmarkFilters(b *testing.B) {
 	// The test txs are not properly signed, can't simply create a chain
 	// and then import blocks. TODO(rjl493456442) try to get rid of the
 	// manual database writes.
-	gspec.MustCommit(db)
+	gspec.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults))
 
 	for i, block := range chain {
 		rawdb.WriteBlock(db, block)
@@ -180,7 +180,7 @@ func TestFilters(t *testing.T) {
 
 	// Hack: GenerateChainWithGenesis creates a new db.
 	// Commit the genesis manually and use GenerateChain.
-	_, err = gspec.Commit(db, trie.NewDatabase(db))
+	_, err = gspec.Commit(db, trie.NewDatabase(db, nil))
 	if err != nil {
 		t.Fatal(err)
 	}

eth/handler.go

@@ -29,6 +29,7 @@ import (
 	"github.com/ethereum/go-ethereum/consensus/beacon"
 	"github.com/ethereum/go-ethereum/core"
 	"github.com/ethereum/go-ethereum/core/forkid"
+	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/txpool"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/eth/downloader"
@@ -40,6 +41,7 @@ import (
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/metrics"
 	"github.com/ethereum/go-ethereum/p2p"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 )
 
 const (
@@ -183,7 +185,7 @@ func newHandler(config *handlerConfig) (*handler, error) {
 		}
 		// If we've successfully finished a sync cycle, accept transactions from
 		// the network
-		h.acceptTxs.Store(true)
+		h.enableSyncedFeatures()
 	}
 	// Construct the downloader (long sync)
 	h.downloader = downloader.New(config.Database, h.eventMux, h.chain, nil, h.removePeer, success)
@@ -272,7 +274,7 @@ func newHandler(config *handlerConfig) (*handler, error) {
 		}
 		n, err := h.chain.InsertChain(blocks)
 		if err == nil {
-			h.acceptTxs.Store(true) // Mark initial sync done on any fetcher import
+			h.enableSyncedFeatures() // Mark initial sync done on any fetcher import
 		}
 		return n, err
 	}
@@ -674,3 +676,12 @@ func (h *handler) txBroadcastLoop() {
 		}
 	}
 }
+
+// enableSyncedFeatures enables the post-sync functionalities when the initial
+// sync is finished.
+func (h *handler) enableSyncedFeatures() {
+	h.acceptTxs.Store(true)
+	if h.chain.TrieDB().Scheme() == rawdb.PathScheme {
+		h.chain.TrieDB().SetBufferSize(pathdb.DefaultBufferSize)
+	}
+}

eth/protocols/eth/handler_test.go

@@ -112,7 +112,7 @@ func newTestBackendWithGenerator(blocks int, shanghai bool, generator func(int,
 		panic(err)
 	}
 	for _, block := range bs {
-		chain.StateCache().TrieDB().Commit(block.Root(), false)
+		chain.TrieDB().Commit(block.Root(), false)
 	}
 	txconfig := legacypool.DefaultConfig
 	txconfig.Journal = "" // Don't litter the disk with test journals

eth/protocols/eth/handlers.go

@@ -22,6 +22,7 @@ import (
 
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/core"
+	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/rlp"
@@ -246,6 +247,10 @@ func handleGetNodeData66(backend Backend, msg Decoder, peer *Peer) error {
 // ServiceGetNodeDataQuery assembles the response to a node data query. It is
 // exposed to allow external packages to test protocol behavior.
 func ServiceGetNodeDataQuery(chain *core.BlockChain, query GetNodeDataPacket) [][]byte {
+	// Request nodes by hash is not supported in path-based scheme.
+	if chain.TrieDB().Scheme() == rawdb.PathScheme {
+		return nil
+	}
 	// Gather state data until the fetch or network limits is reached
 	var (
 		bytes int
@@ -257,7 +262,7 @@ func ServiceGetNodeDataQuery(chain *core.BlockChain, query GetNodeDataPacket) []
 			break
 		}
 		// Retrieve the requested state entry
-		entry, err := chain.TrieNode(hash)
+		entry, err := chain.TrieDB().Node(hash)
 		if len(entry) == 0 || err != nil {
 			// Read the contract code with prefix only to save unnecessary lookups.
 			entry, err = chain.ContractCodeWithPrefix(hash)

eth/protocols/snap/handler.go

@@ -284,7 +284,7 @@ func ServiceGetAccountRangeQuery(chain *core.BlockChain, req *GetAccountRangePac
 		req.Bytes = softResponseLimit
 	}
 	// Retrieve the requested state and bail out if non existent
-	tr, err := trie.New(trie.StateTrieID(req.Root), chain.StateCache().TrieDB())
+	tr, err := trie.New(trie.StateTrieID(req.Root), chain.TrieDB())
 	if err != nil {
 		return nil, nil
 	}
@@ -414,7 +414,7 @@ func ServiceGetStorageRangesQuery(chain *core.BlockChain, req *GetStorageRangesP
 		if origin != (common.Hash{}) || (abort && len(storage) > 0) {
 			// Request started at a non-zero hash or was capped prematurely, add
 			// the endpoint Merkle proofs
-			accTrie, err := trie.NewStateTrie(trie.StateTrieID(req.Root), chain.StateCache().TrieDB())
+			accTrie, err := trie.NewStateTrie(trie.StateTrieID(req.Root), chain.TrieDB())
 			if err != nil {
 				return nil, nil
 			}
@@ -423,7 +423,7 @@ func ServiceGetStorageRangesQuery(chain *core.BlockChain, req *GetStorageRangesP
 				return nil, nil
 			}
 			id := trie.StorageTrieID(req.Root, account, acc.Root)
-			stTrie, err := trie.NewStateTrie(id, chain.StateCache().TrieDB())
+			stTrie, err := trie.NewStateTrie(id, chain.TrieDB())
 			if err != nil {
 				return nil, nil
 			}
@@ -487,7 +487,7 @@ func ServiceGetTrieNodesQuery(chain *core.BlockChain, req *GetTrieNodesPacket, s
 		req.Bytes = softResponseLimit
 	}
 	// Make sure we have the state associated with the request
-	triedb := chain.StateCache().TrieDB()
+	triedb := chain.TrieDB()
 
 	accTrie, err := trie.NewStateTrie(trie.StateTrieID(req.Root), triedb)
 	if err != nil {

eth/protocols/snap/sync_test.go

@@ -35,6 +35,7 @@ import (
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 	"github.com/ethereum/go-ethereum/trie/trienode"
 	"golang.org/x/crypto/sha3"
 	"golang.org/x/exp/slices"
@@ -561,6 +562,11 @@ func noProofStorageRequestHandler(t *testPeer, requestId uint64, root common.Has
 func TestSyncBloatedProof(t *testing.T) {
 	t.Parallel()
 
+	testSyncBloatedProof(t, rawdb.HashScheme)
+	testSyncBloatedProof(t, rawdb.PathScheme)
+}
+
+func testSyncBloatedProof(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -570,7 +576,7 @@ func TestSyncBloatedProof(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100, scheme)
 	source := newTestPeer("source", t, term)
 	source.accountTrie = sourceAccountTrie.Copy()
 	source.accountValues = elems
@@ -638,6 +644,11 @@ func setupSyncer(scheme string, peers ...*testPeer) *Syncer {
 func TestSync(t *testing.T) {
 	t.Parallel()
 
+	testSync(t, rawdb.HashScheme)
+	testSync(t, rawdb.PathScheme)
+}
+
+func testSync(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -647,7 +658,7 @@ func TestSync(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100, scheme)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -659,7 +670,7 @@ func TestSync(t *testing.T) {
 	if err := syncer.Sync(sourceAccountTrie.Hash(), cancel); err != nil {
 		t.Fatalf("sync failed: %v", err)
 	}
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncTinyTriePanic tests a basic sync with one peer, and a tiny trie. This caused a
@@ -667,6 +678,11 @@ func TestSync(t *testing.T) {
 func TestSyncTinyTriePanic(t *testing.T) {
 	t.Parallel()
 
+	testSyncTinyTriePanic(t, rawdb.HashScheme)
+	testSyncTinyTriePanic(t, rawdb.PathScheme)
+}
+
+func testSyncTinyTriePanic(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -676,7 +692,7 @@ func TestSyncTinyTriePanic(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(1)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(1, scheme)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -690,13 +706,18 @@ func TestSyncTinyTriePanic(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestMultiSync tests a basic sync with multiple peers
 func TestMultiSync(t *testing.T) {
 	t.Parallel()
 
+	testMultiSync(t, rawdb.HashScheme)
+	testMultiSync(t, rawdb.PathScheme)
+}
+
+func testMultiSync(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -706,7 +727,7 @@ func TestMultiSync(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100, scheme)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -720,13 +741,18 @@ func TestMultiSync(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncWithStorage tests  basic sync using accounts + storage + code
 func TestSyncWithStorage(t *testing.T) {
 	t.Parallel()
 
+	testSyncWithStorage(t, rawdb.HashScheme)
+	testSyncWithStorage(t, rawdb.PathScheme)
+}
+
+func testSyncWithStorage(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -736,7 +762,7 @@ func TestSyncWithStorage(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(3, 3000, true, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 3, 3000, true, false)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -752,13 +778,18 @@ func TestSyncWithStorage(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestMultiSyncManyUseless contains one good peer, and many which doesn't return anything valuable at all
 func TestMultiSyncManyUseless(t *testing.T) {
 	t.Parallel()
 
+	testMultiSyncManyUseless(t, rawdb.HashScheme)
+	testMultiSyncManyUseless(t, rawdb.PathScheme)
+}
+
+func testMultiSyncManyUseless(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -768,7 +799,7 @@ func TestMultiSyncManyUseless(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(100, 3000, true, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 100, 3000, true, false)
 
 	mkSource := func(name string, noAccount, noStorage, noTrieNode bool) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -801,11 +832,18 @@ func TestMultiSyncManyUseless(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestMultiSyncManyUseless contains one good peer, and many which doesn't return anything valuable at all
 func TestMultiSyncManyUselessWithLowTimeout(t *testing.T) {
+	t.Parallel()
+
+	testMultiSyncManyUselessWithLowTimeout(t, rawdb.HashScheme)
+	testMultiSyncManyUselessWithLowTimeout(t, rawdb.PathScheme)
+}
+
+func testMultiSyncManyUselessWithLowTimeout(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -815,7 +853,7 @@ func TestMultiSyncManyUselessWithLowTimeout(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(100, 3000, true, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 100, 3000, true, false)
 
 	mkSource := func(name string, noAccount, noStorage, noTrieNode bool) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -853,11 +891,18 @@ func TestMultiSyncManyUselessWithLowTimeout(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestMultiSyncManyUnresponsive contains one good peer, and many which doesn't respond at all
 func TestMultiSyncManyUnresponsive(t *testing.T) {
+	t.Parallel()
+
+	testMultiSyncManyUnresponsive(t, rawdb.HashScheme)
+	testMultiSyncManyUnresponsive(t, rawdb.PathScheme)
+}
+
+func testMultiSyncManyUnresponsive(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -867,7 +912,7 @@ func TestMultiSyncManyUnresponsive(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(100, 3000, true, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 100, 3000, true, false)
 
 	mkSource := func(name string, noAccount, noStorage, noTrieNode bool) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -903,7 +948,7 @@ func TestMultiSyncManyUnresponsive(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 func checkStall(t *testing.T, term func()) chan struct{} {
@@ -925,6 +970,11 @@ func checkStall(t *testing.T, term func()) chan struct{} {
 func TestSyncBoundaryAccountTrie(t *testing.T) {
 	t.Parallel()
 
+	testSyncBoundaryAccountTrie(t, rawdb.HashScheme)
+	testSyncBoundaryAccountTrie(t, rawdb.PathScheme)
+}
+
+func testSyncBoundaryAccountTrie(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -934,7 +984,7 @@ func TestSyncBoundaryAccountTrie(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeBoundaryAccountTrie(3000)
+	nodeScheme, sourceAccountTrie, elems := makeBoundaryAccountTrie(scheme, 3000)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -952,7 +1002,7 @@ func TestSyncBoundaryAccountTrie(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncNoStorageAndOneCappedPeer tests sync using accounts and no storage, where one peer is
@@ -960,6 +1010,11 @@ func TestSyncBoundaryAccountTrie(t *testing.T) {
 func TestSyncNoStorageAndOneCappedPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncNoStorageAndOneCappedPeer(t, rawdb.HashScheme)
+	testSyncNoStorageAndOneCappedPeer(t, rawdb.PathScheme)
+}
+
+func testSyncNoStorageAndOneCappedPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -969,7 +1024,7 @@ func TestSyncNoStorageAndOneCappedPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000, scheme)
 
 	mkSource := func(name string, slow bool) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -994,7 +1049,7 @@ func TestSyncNoStorageAndOneCappedPeer(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncNoStorageAndOneCodeCorruptPeer has one peer which doesn't deliver
@@ -1002,6 +1057,11 @@ func TestSyncNoStorageAndOneCappedPeer(t *testing.T) {
 func TestSyncNoStorageAndOneCodeCorruptPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncNoStorageAndOneCodeCorruptPeer(t, rawdb.HashScheme)
+	testSyncNoStorageAndOneCodeCorruptPeer(t, rawdb.PathScheme)
+}
+
+func testSyncNoStorageAndOneCodeCorruptPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1011,7 +1071,7 @@ func TestSyncNoStorageAndOneCodeCorruptPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000, scheme)
 
 	mkSource := func(name string, codeFn codeHandlerFunc) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1034,12 +1094,17 @@ func TestSyncNoStorageAndOneCodeCorruptPeer(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 func TestSyncNoStorageAndOneAccountCorruptPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncNoStorageAndOneAccountCorruptPeer(t, rawdb.HashScheme)
+	testSyncNoStorageAndOneAccountCorruptPeer(t, rawdb.PathScheme)
+}
+
+func testSyncNoStorageAndOneAccountCorruptPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1049,7 +1114,7 @@ func TestSyncNoStorageAndOneAccountCorruptPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000, scheme)
 
 	mkSource := func(name string, accFn accountHandlerFunc) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1072,7 +1137,7 @@ func TestSyncNoStorageAndOneAccountCorruptPeer(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncNoStorageAndOneCodeCappedPeer has one peer which delivers code hashes
@@ -1080,6 +1145,11 @@ func TestSyncNoStorageAndOneAccountCorruptPeer(t *testing.T) {
 func TestSyncNoStorageAndOneCodeCappedPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncNoStorageAndOneCodeCappedPeer(t, rawdb.HashScheme)
+	testSyncNoStorageAndOneCodeCappedPeer(t, rawdb.PathScheme)
+}
+
+func testSyncNoStorageAndOneCodeCappedPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1089,7 +1159,7 @@ func TestSyncNoStorageAndOneCodeCappedPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(3000, scheme)
 
 	mkSource := func(name string, codeFn codeHandlerFunc) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1123,7 +1193,7 @@ func TestSyncNoStorageAndOneCodeCappedPeer(t *testing.T) {
 	if threshold := 100; counter > threshold {
 		t.Logf("Error, expected < %d invocations, got %d", threshold, counter)
 	}
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncBoundaryStorageTrie tests sync against a few normal peers, but the
@@ -1131,6 +1201,11 @@ func TestSyncNoStorageAndOneCodeCappedPeer(t *testing.T) {
 func TestSyncBoundaryStorageTrie(t *testing.T) {
 	t.Parallel()
 
+	testSyncBoundaryStorageTrie(t, rawdb.HashScheme)
+	testSyncBoundaryStorageTrie(t, rawdb.PathScheme)
+}
+
+func testSyncBoundaryStorageTrie(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1140,7 +1215,7 @@ func TestSyncBoundaryStorageTrie(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(10, 1000, false, true)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 10, 1000, false, true)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1160,7 +1235,7 @@ func TestSyncBoundaryStorageTrie(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncWithStorageAndOneCappedPeer tests sync using accounts + storage, where one peer is
@@ -1168,6 +1243,11 @@ func TestSyncBoundaryStorageTrie(t *testing.T) {
 func TestSyncWithStorageAndOneCappedPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncWithStorageAndOneCappedPeer(t, rawdb.HashScheme)
+	testSyncWithStorageAndOneCappedPeer(t, rawdb.PathScheme)
+}
+
+func testSyncWithStorageAndOneCappedPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1177,7 +1257,7 @@ func TestSyncWithStorageAndOneCappedPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(300, 1000, false, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 300, 1000, false, false)
 
 	mkSource := func(name string, slow bool) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1202,7 +1282,7 @@ func TestSyncWithStorageAndOneCappedPeer(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncWithStorageAndCorruptPeer tests sync using accounts + storage, where one peer is
@@ -1210,6 +1290,11 @@ func TestSyncWithStorageAndOneCappedPeer(t *testing.T) {
 func TestSyncWithStorageAndCorruptPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncWithStorageAndCorruptPeer(t, rawdb.HashScheme)
+	testSyncWithStorageAndCorruptPeer(t, rawdb.PathScheme)
+}
+
+func testSyncWithStorageAndCorruptPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1219,7 +1304,7 @@ func TestSyncWithStorageAndCorruptPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(100, 3000, true, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 100, 3000, true, false)
 
 	mkSource := func(name string, handler storageHandlerFunc) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1243,12 +1328,17 @@ func TestSyncWithStorageAndCorruptPeer(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 func TestSyncWithStorageAndNonProvingPeer(t *testing.T) {
 	t.Parallel()
 
+	testSyncWithStorageAndNonProvingPeer(t, rawdb.HashScheme)
+	testSyncWithStorageAndNonProvingPeer(t, rawdb.PathScheme)
+}
+
+func testSyncWithStorageAndNonProvingPeer(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1258,7 +1348,7 @@ func TestSyncWithStorageAndNonProvingPeer(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(100, 3000, true, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorage(scheme, 100, 3000, true, false)
 
 	mkSource := func(name string, handler storageHandlerFunc) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1281,7 +1371,7 @@ func TestSyncWithStorageAndNonProvingPeer(t *testing.T) {
 		t.Fatalf("sync failed: %v", err)
 	}
 	close(done)
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 // TestSyncWithStorage tests  basic sync using accounts + storage + code, against
@@ -1290,6 +1380,12 @@ func TestSyncWithStorageAndNonProvingPeer(t *testing.T) {
 // did not mark the account for healing.
 func TestSyncWithStorageMisbehavingProve(t *testing.T) {
 	t.Parallel()
+
+	testSyncWithStorageMisbehavingProve(t, rawdb.HashScheme)
+	testSyncWithStorageMisbehavingProve(t, rawdb.PathScheme)
+}
+
+func testSyncWithStorageMisbehavingProve(t *testing.T, scheme string) {
 	var (
 		once   sync.Once
 		cancel = make(chan struct{})
@@ -1299,7 +1395,7 @@ func TestSyncWithStorageMisbehavingProve(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorageWithUniqueStorage(10, 30, false)
+	nodeScheme, sourceAccountTrie, elems, storageTries, storageElems := makeAccountTrieWithStorageWithUniqueStorage(scheme, 10, 30, false)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1314,7 +1410,7 @@ func TestSyncWithStorageMisbehavingProve(t *testing.T) {
 	if err := syncer.Sync(sourceAccountTrie.Hash(), cancel); err != nil {
 		t.Fatalf("sync failed: %v", err)
 	}
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 }
 
 type kv struct {
@@ -1364,9 +1460,9 @@ func getCodeByHash(hash common.Hash) []byte {
 }
 
 // makeAccountTrieNoStorage spits out a trie, along with the leafs
-func makeAccountTrieNoStorage(n int) (string, *trie.Trie, []*kv) {
+func makeAccountTrieNoStorage(n int, scheme string) (string, *trie.Trie, []*kv) {
 	var (
-		db      = trie.NewDatabase(rawdb.NewMemoryDatabase())
+		db      = trie.NewDatabase(rawdb.NewMemoryDatabase(), newDbConfig(scheme))
 		accTrie = trie.NewEmpty(db)
 		entries []*kv
 	)
@@ -1396,12 +1492,12 @@ func makeAccountTrieNoStorage(n int) (string, *trie.Trie, []*kv) {
 // makeBoundaryAccountTrie constructs an account trie. Instead of filling
 // accounts normally, this function will fill a few accounts which have
 // boundary hash.
-func makeBoundaryAccountTrie(n int) (string, *trie.Trie, []*kv) {
+func makeBoundaryAccountTrie(scheme string, n int) (string, *trie.Trie, []*kv) {
 	var (
 		entries    []*kv
 		boundaries []common.Hash
 
-		db      = trie.NewDatabase(rawdb.NewMemoryDatabase())
+		db      = trie.NewDatabase(rawdb.NewMemoryDatabase(), newDbConfig(scheme))
 		accTrie = trie.NewEmpty(db)
 	)
 	// Initialize boundaries
@@ -1457,9 +1553,9 @@ func makeBoundaryAccountTrie(n int) (string, *trie.Trie, []*kv) {
 
 // makeAccountTrieWithStorageWithUniqueStorage creates an account trie where each accounts
 // has a unique storage set.
-func makeAccountTrieWithStorageWithUniqueStorage(accounts, slots int, code bool) (string, *trie.Trie, []*kv, map[common.Hash]*trie.Trie, map[common.Hash][]*kv) {
+func makeAccountTrieWithStorageWithUniqueStorage(scheme string, accounts, slots int, code bool) (string, *trie.Trie, []*kv, map[common.Hash]*trie.Trie, map[common.Hash][]*kv) {
 	var (
-		db             = trie.NewDatabase(rawdb.NewMemoryDatabase())
+		db             = trie.NewDatabase(rawdb.NewMemoryDatabase(), newDbConfig(scheme))
 		accTrie        = trie.NewEmpty(db)
 		entries        []*kv
 		storageRoots   = make(map[common.Hash]common.Hash)
@@ -1512,9 +1608,9 @@ func makeAccountTrieWithStorageWithUniqueStorage(accounts, slots int, code bool)
 }
 
 // makeAccountTrieWithStorage spits out a trie, along with the leafs
-func makeAccountTrieWithStorage(accounts, slots int, code, boundary bool) (string, *trie.Trie, []*kv, map[common.Hash]*trie.Trie, map[common.Hash][]*kv) {
+func makeAccountTrieWithStorage(scheme string, accounts, slots int, code, boundary bool) (string, *trie.Trie, []*kv, map[common.Hash]*trie.Trie, map[common.Hash][]*kv) {
 	var (
-		db             = trie.NewDatabase(rawdb.NewMemoryDatabase())
+		db             = trie.NewDatabase(rawdb.NewMemoryDatabase(), newDbConfig(scheme))
 		accTrie        = trie.NewEmpty(db)
 		entries        []*kv
 		storageRoots   = make(map[common.Hash]common.Hash)
@@ -1656,9 +1752,9 @@ func makeBoundaryStorageTrie(owner common.Hash, n int, db *trie.Database) (commo
 	return root, nodes, entries
 }
 
-func verifyTrie(db ethdb.KeyValueStore, root common.Hash, t *testing.T) {
+func verifyTrie(scheme string, db ethdb.KeyValueStore, root common.Hash, t *testing.T) {
 	t.Helper()
-	triedb := trie.NewDatabase(rawdb.NewDatabase(db))
+	triedb := trie.NewDatabase(rawdb.NewDatabase(db), newDbConfig(scheme))
 	accTrie, err := trie.New(trie.StateTrieID(root), triedb)
 	if err != nil {
 		t.Fatal(err)
@@ -1700,6 +1796,13 @@ func verifyTrie(db ethdb.KeyValueStore, root common.Hash, t *testing.T) {
 // TestSyncAccountPerformance tests how efficient the snap algo is at minimizing
 // state healing
 func TestSyncAccountPerformance(t *testing.T) {
+	t.Parallel()
+
+	testSyncAccountPerformance(t, rawdb.HashScheme)
+	testSyncAccountPerformance(t, rawdb.PathScheme)
+}
+
+func testSyncAccountPerformance(t *testing.T, scheme string) {
 	// Set the account concurrency to 1. This _should_ result in the
 	// range root to become correct, and there should be no healing needed
 	defer func(old int) { accountConcurrency = old }(accountConcurrency)
@@ -1714,7 +1817,7 @@ func TestSyncAccountPerformance(t *testing.T) {
 			})
 		}
 	)
-	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100)
+	nodeScheme, sourceAccountTrie, elems := makeAccountTrieNoStorage(100, scheme)
 
 	mkSource := func(name string) *testPeer {
 		source := newTestPeer(name, t, term)
@@ -1727,7 +1830,7 @@ func TestSyncAccountPerformance(t *testing.T) {
 	if err := syncer.Sync(sourceAccountTrie.Hash(), cancel); err != nil {
 		t.Fatalf("sync failed: %v", err)
 	}
-	verifyTrie(syncer.db, sourceAccountTrie.Hash(), t)
+	verifyTrie(scheme, syncer.db, sourceAccountTrie.Hash(), t)
 	// The trie root will always be requested, since it is added when the snap
 	// sync cycle starts. When popping the queue, we do not look it up again.
 	// Doing so would bring this number down to zero in this artificial testcase,
@@ -1787,3 +1890,10 @@ func TestSlotEstimation(t *testing.T) {
 		}
 	}
 }
+
+func newDbConfig(scheme string) *trie.Config {
+	if scheme == rawdb.HashScheme {
+		return &trie.Config{}
+	}
+	return &trie.Config{PathDB: pathdb.Defaults}
+}

eth/state_accessor.go

@@ -24,6 +24,7 @@ import (
 
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/core"
+	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/state"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/core/vm"
@@ -36,31 +37,11 @@ import (
 // for releasing state.
 var noopReleaser = tracers.StateReleaseFunc(func() {})
 
-// StateAtBlock retrieves the state database associated with a certain block.
+func (eth *Ethereum) hashState(ctx context.Context, block *types.Block, reexec uint64, base *state.StateDB, readOnly bool, preferDisk bool) (statedb *state.StateDB, release tracers.StateReleaseFunc, err error) {
-// If no state is locally available for the given block, a number of blocks
-// are attempted to be reexecuted to generate the desired state. The optional
-// base layer statedb can be provided which is regarded as the statedb of the
-// parent block.
-//
-// An additional release function will be returned if the requested state is
-// available. Release is expected to be invoked when the returned state is no longer needed.
-// Its purpose is to prevent resource leaking. Though it can be noop in some cases.
-//
-// Parameters:
-//   - block:      The block for which we want the state(state = block.Root)
-//   - reexec:     The maximum number of blocks to reprocess trying to obtain the desired state
-//   - base:       If the caller is tracing multiple blocks, the caller can provide the parent
-//     state continuously from the callsite.
-//   - readOnly:   If true, then the live 'blockchain' state database is used. No mutation should
-//     be made from caller, e.g. perform Commit or other 'save-to-disk' changes.
-//     Otherwise, the trash generated by caller may be persisted permanently.
-//   - preferDisk: this arg can be used by the caller to signal that even though the 'base' is
-//     provided, it would be preferable to start from a fresh state, if we have it
-//     on disk.
-func (eth *Ethereum) StateAtBlock(ctx context.Context, block *types.Block, reexec uint64, base *state.StateDB, readOnly bool, preferDisk bool) (statedb *state.StateDB, release tracers.StateReleaseFunc, err error) {
 	var (
 		current  *types.Block
 		database state.Database
+		triedb   *trie.Database
 		report   = true
 		origin   = block.NumberU64()
 	)
@@ -71,9 +52,9 @@ func (eth *Ethereum) StateAtBlock(ctx context.Context, block *types.Block, reexe
 		// on top to prevent garbage collection and return a release
 		// function to deref it.
 		if statedb, err = eth.blockchain.StateAt(block.Root()); err == nil {
-			statedb.Database().TrieDB().Reference(block.Root(), common.Hash{})
+			eth.blockchain.TrieDB().Reference(block.Root(), common.Hash{})
 			return statedb, func() {
-				statedb.Database().TrieDB().Dereference(block.Root())
+				eth.blockchain.TrieDB().Dereference(block.Root())
 			}, nil
 		}
 	}
@@ -84,14 +65,16 @@ func (eth *Ethereum) StateAtBlock(ctx context.Context, block *types.Block, reexe
 		if preferDisk {
 			// Create an ephemeral trie.Database for isolating the live one. Otherwise
 			// the internal junks created by tracing will be persisted into the disk.
-			database = state.NewDatabaseWithConfig(eth.chainDb, &trie.Config{Cache: 16})
+			// TODO(rjl493456442), clean cache is disabled to prevent memory leak,
+			// please re-enable it for better performance.
+			database = state.NewDatabaseWithConfig(eth.chainDb, trie.HashDefaults)
 			if statedb, err = state.New(block.Root(), database, nil); err == nil {
 				log.Info("Found disk backend for state trie", "root", block.Root(), "number", block.Number())
 				return statedb, noopReleaser, nil
 			}
 		}
 		// The optional base statedb is given, mark the start point as parent block
-		statedb, database, report = base, base.Database(), false
+		statedb, database, triedb, report = base, base.Database(), base.Database().TrieDB(), false
 		current = eth.blockchain.GetBlock(block.ParentHash(), block.NumberU64()-1)
 	} else {
 		// Otherwise, try to reexec blocks until we find a state or reach our limit
@@ -99,7 +82,10 @@ func (eth *Ethereum) StateAtBlock(ctx context.Context, block *types.Block, reexe
 
 		// Create an ephemeral trie.Database for isolating the live one. Otherwise
 		// the internal junks created by tracing will be persisted into the disk.
-		database = state.NewDatabaseWithConfig(eth.chainDb, &trie.Config{Cache: 16})
+		// TODO(rjl493456442), clean cache is disabled to prevent memory leak,
+		// please re-enable it for better performance.
+		triedb = trie.NewDatabase(eth.chainDb, trie.HashDefaults)
+		database = state.NewDatabaseWithNodeDB(eth.chainDb, triedb)
 
 		// If we didn't check the live database, do check state over ephemeral database,
 		// otherwise we would rewind past a persisted block (specific corner case is
@@ -175,17 +161,58 @@ func (eth *Ethereum) StateAtBlock(ctx context.Context, block *types.Block, reexe
 		}
 		// Hold the state reference and also drop the parent state
 		// to prevent accumulating too many nodes in memory.
-		database.TrieDB().Reference(root, common.Hash{})
+		triedb.Reference(root, common.Hash{})
 		if parent != (common.Hash{}) {
-			database.TrieDB().Dereference(parent)
+			triedb.Dereference(parent)
 		}
 		parent = root
 	}
 	if report {
-		nodes, imgs := database.TrieDB().Size()
+		nodes, imgs := triedb.Size()
 		log.Info("Historical state regenerated", "block", current.NumberU64(), "elapsed", time.Since(start), "nodes", nodes, "preimages", imgs)
 	}
-	return statedb, func() { database.TrieDB().Dereference(block.Root()) }, nil
+	return statedb, func() { triedb.Dereference(block.Root()) }, nil
+}
+
+func (eth *Ethereum) pathState(block *types.Block) (*state.StateDB, func(), error) {
+	// Check if the requested state is available in the live chain.
+	statedb, err := eth.blockchain.StateAt(block.Root())
+	if err == nil {
+		return statedb, noopReleaser, nil
+	}
+	// TODO historic state is not supported in path-based scheme.
+	// Fully archive node in pbss will be implemented by relying
+	// on state history, but needs more work on top.
+	return nil, nil, errors.New("historical state not available in path scheme yet")
+}
+
+// stateAtBlock retrieves the state database associated with a certain block.
+// If no state is locally available for the given block, a number of blocks
+// are attempted to be reexecuted to generate the desired state. The optional
+// base layer statedb can be provided which is regarded as the statedb of the
+// parent block.
+//
+// An additional release function will be returned if the requested state is
+// available. Release is expected to be invoked when the returned state is no
+// longer needed. Its purpose is to prevent resource leaking. Though it can be
+// noop in some cases.
+//
+// Parameters:
+//   - block:      The block for which we want the state(state = block.Root)
+//   - reexec:     The maximum number of blocks to reprocess trying to obtain the desired state
+//   - base:       If the caller is tracing multiple blocks, the caller can provide the parent
+//     state continuously from the callsite.
+//   - readOnly:   If true, then the live 'blockchain' state database is used. No mutation should
+//     be made from caller, e.g. perform Commit or other 'save-to-disk' changes.
+//     Otherwise, the trash generated by caller may be persisted permanently.
+//   - preferDisk: This arg can be used by the caller to signal that even though the 'base' is
+//     provided, it would be preferable to start from a fresh state, if we have it
+//     on disk.
+func (eth *Ethereum) stateAtBlock(ctx context.Context, block *types.Block, reexec uint64, base *state.StateDB, readOnly bool, preferDisk bool) (statedb *state.StateDB, release tracers.StateReleaseFunc, err error) {
+	if eth.blockchain.TrieDB().Scheme() == rawdb.HashScheme {
+		return eth.hashState(ctx, block, reexec, base, readOnly, preferDisk)
+	}
+	return eth.pathState(block)
 }
 
 // stateAtTransaction returns the execution environment of a certain transaction.
@@ -201,7 +228,7 @@ func (eth *Ethereum) stateAtTransaction(ctx context.Context, block *types.Block,
 	}
 	// Lookup the statedb of parent block from the live database,
 	// otherwise regenerate it on the flight.
-	statedb, release, err := eth.StateAtBlock(ctx, parent, reexec, nil, true, false)
+	statedb, release, err := eth.stateAtBlock(ctx, parent, reexec, nil, true, false)
 	if err != nil {
 		return nil, vm.BlockContext{}, nil, nil, err
 	}

eth/tracers/internal/tracetest/calltrace_test.go

@@ -137,8 +137,10 @@ func testCallTracer(tracerName string, dirPath string, t *testing.T) {
 					GasLimit:    uint64(test.Context.GasLimit),
 					BaseFee:     test.Genesis.BaseFee,
 				}
-				_, statedb = tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false)
+				triedb, _, statedb = tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false, rawdb.HashScheme)
 			)
+			triedb.Close()
+
 			tracer, err := tracers.DefaultDirectory.New(tracerName, new(tracers.Context), test.TracerConfig)
 			if err != nil {
 				t.Fatalf("failed to create call tracer: %v", err)
@@ -237,7 +239,8 @@ func benchTracer(tracerName string, test *callTracerTest, b *testing.B) {
 		Difficulty:  (*big.Int)(test.Context.Difficulty),
 		GasLimit:    uint64(test.Context.GasLimit),
 	}
-	_, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false)
+	triedb, _, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false, rawdb.HashScheme)
+	defer triedb.Close()
 
 	b.ReportAllocs()
 	b.ResetTimer()
@@ -363,7 +366,7 @@ func TestInternals(t *testing.T) {
 		},
 	} {
 		t.Run(tc.name, func(t *testing.T) {
-			_, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(),
+			triedb, _, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(),
 				core.GenesisAlloc{
 					to: core.GenesisAccount{
 						Code: tc.code,
@@ -371,7 +374,9 @@ func TestInternals(t *testing.T) {
 					origin: core.GenesisAccount{
 						Balance: big.NewInt(500000000000000),
 					},
-				}, false)
+				}, false, rawdb.HashScheme)
+			defer triedb.Close()
+
 			evm := vm.NewEVM(context, txContext, statedb, params.MainnetChainConfig, vm.Config{Tracer: tc.tracer})
 			msg := &core.Message{
 				To:                &to,

eth/tracers/internal/tracetest/flat_calltrace_test.go

@@ -100,7 +100,8 @@ func flatCallTracerTestRunner(tracerName string, filename string, dirPath string
 		Difficulty:  (*big.Int)(test.Context.Difficulty),
 		GasLimit:    uint64(test.Context.GasLimit),
 	}
-	_, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false)
+	triedb, _, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false, rawdb.HashScheme)
+	defer triedb.Close()
 
 	// Create the tracer, the EVM environment and run it
 	tracer, err := tracers.DefaultDirectory.New(tracerName, new(tracers.Context), test.TracerConfig)

eth/tracers/internal/tracetest/prestate_test.go

@@ -108,8 +108,10 @@ func testPrestateDiffTracer(tracerName string, dirPath string, t *testing.T) {
 					GasLimit:    uint64(test.Context.GasLimit),
 					BaseFee:     test.Genesis.BaseFee,
 				}
-				_, statedb = tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false)
+				triedb, _, statedb = tests.MakePreState(rawdb.NewMemoryDatabase(), test.Genesis.Alloc, false, rawdb.HashScheme)
 			)
+			defer triedb.Close()
+
 			tracer, err := tracers.DefaultDirectory.New(tracerName, new(tracers.Context), test.TracerConfig)
 			if err != nil {
 				t.Fatalf("failed to create call tracer: %v", err)

eth/tracers/tracers_test.go

@@ -79,7 +79,9 @@ func BenchmarkTransactionTrace(b *testing.B) {
 		Code:    []byte{},
 		Balance: big.NewInt(500000000000000),
 	}
-	_, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(), alloc, false)
+	triedb, _, statedb := tests.MakePreState(rawdb.NewMemoryDatabase(), alloc, false, rawdb.HashScheme)
+	defer triedb.Close()
+
 	// Create the tracer, the EVM environment and run it
 	tracer := logger.NewStructLogger(&logger.Config{
 		Debug: false,

internal/flags/categories.go

@@ -22,7 +22,7 @@ const (
 	EthCategory        = "ETHEREUM"
 	LightCategory      = "LIGHT CLIENT"
 	DevCategory        = "DEVELOPER CHAIN"
-	EthashCategory     = "ETHASH"
+	StateCategory      = "STATE HISTORY MANAGEMENT"
 	TxPoolCategory     = "TRANSACTION POOL (EVM)"
 	BlobPoolCategory   = "TRANSACTION POOL (BLOB)"
 	PerfCategory       = "PERFORMANCE TUNING"

les/client.go

@@ -98,7 +98,8 @@ func New(stack *node.Node, config *ethconfig.Config) (*LightEthereum, error) {
 	if config.OverrideVerkle != nil {
 		overrides.OverrideVerkle = config.OverrideVerkle
 	}
-	chainConfig, genesisHash, genesisErr := core.SetupGenesisBlockWithOverride(chainDb, trie.NewDatabase(chainDb), config.Genesis, &overrides)
+	triedb := trie.NewDatabase(chainDb, trie.HashDefaults)
+	chainConfig, genesisHash, genesisErr := core.SetupGenesisBlockWithOverride(chainDb, triedb, config.Genesis, &overrides)
 	if _, isCompat := genesisErr.(*params.ConfigCompatError); genesisErr != nil && !isCompat {
 		return nil, genesisErr
 	}

les/handler_test.go

@@ -406,7 +406,7 @@ func testGetProofs(t *testing.T, protocol int) {
 	accounts := []common.Address{bankAddr, userAddr1, userAddr2, signerAddr, {}}
 	for i := uint64(0); i <= bc.CurrentBlock().Number.Uint64(); i++ {
 		header := bc.GetHeaderByNumber(i)
-		trie, _ := trie.New(trie.StateTrieID(header.Root), trie.NewDatabase(server.db))
+		trie, _ := trie.New(trie.StateTrieID(header.Root), server.backend.Blockchain().TrieDB())
 
 		for _, acc := range accounts {
 			req := ProofReq{
@@ -457,7 +457,7 @@ func testGetStaleProof(t *testing.T, protocol int) {
 		var expected []rlp.RawValue
 		if wantOK {
 			proofsV2 := light.NewNodeSet()
-			t, _ := trie.New(trie.StateTrieID(header.Root), trie.NewDatabase(server.db))
+			t, _ := trie.New(trie.StateTrieID(header.Root), server.backend.Blockchain().TrieDB())
 			t.Prove(account, proofsV2)
 			expected = proofsV2.NodeList()
 		}
@@ -513,7 +513,7 @@ func testGetCHTProofs(t *testing.T, protocol int) {
 		AuxData: [][]byte{rlp},
 	}
 	root := light.GetChtRoot(server.db, 0, bc.GetHeaderByNumber(config.ChtSize-1).Hash())
-	trie, _ := trie.New(trie.TrieID(root), trie.NewDatabase(rawdb.NewTable(server.db, string(rawdb.ChtTablePrefix))))
+	trie, _ := trie.New(trie.TrieID(root), trie.NewDatabase(rawdb.NewTable(server.db, string(rawdb.ChtTablePrefix)), trie.HashDefaults))
 	trie.Prove(key, &proofsV2.Proofs)
 	// Assemble the requests for the different protocols
 	requestsV2 := []HelperTrieReq{{
@@ -578,7 +578,7 @@ func testGetBloombitsProofs(t *testing.T, protocol int) {
 		var proofs HelperTrieResps
 
 		root := light.GetBloomTrieRoot(server.db, 0, bc.GetHeaderByNumber(config.BloomTrieSize-1).Hash())
-		trie, _ := trie.New(trie.TrieID(root), trie.NewDatabase(rawdb.NewTable(server.db, string(rawdb.BloomTrieTablePrefix))))
+		trie, _ := trie.New(trie.TrieID(root), trie.NewDatabase(rawdb.NewTable(server.db, string(rawdb.BloomTrieTablePrefix)), trie.HashDefaults))
 		trie.Prove(key, &proofs.Proofs)
 
 		// Send the proof request and verify the response

les/odr_test.go

@@ -104,7 +104,7 @@ func odrAccounts(ctx context.Context, db ethdb.Database, config *params.ChainCon
 	for _, addr := range acc {
 		if bc != nil {
 			header := bc.GetHeaderByHash(bhash)
-			st, err = state.New(header.Root, state.NewDatabase(db), nil)
+			st, err = state.New(header.Root, bc.StateCache(), nil)
 		} else {
 			header := lc.GetHeaderByHash(bhash)
 			st = light.NewState(ctx, header, lc.Odr())

les/server_handler.go

@@ -390,7 +390,8 @@ func (h *serverHandler) GetHelperTrie(typ uint, index uint64) *trie.Trie {
 	if root == (common.Hash{}) {
 		return nil
 	}
-	trie, _ := trie.New(trie.TrieID(root), trie.NewDatabase(rawdb.NewTable(h.chainDb, prefix)))
+	triedb := trie.NewDatabase(rawdb.NewTable(h.chainDb, prefix), trie.HashDefaults)
+	trie, _ := trie.New(trie.TrieID(root), triedb)
 	return trie
 }
 

les/server_requests.go

@@ -303,9 +303,8 @@ func handleGetCode(msg Decoder) (serveRequestFn, uint64, uint64, error) {
 				p.bumpInvalid()
 				continue
 			}
-			triedb := bc.StateCache().TrieDB()
 			address := common.BytesToAddress(request.AccountAddress)
-			account, err := getAccount(triedb, header.Root, address)
+			account, err := getAccount(bc.TrieDB(), header.Root, address)
 			if err != nil {
 				p.Log().Warn("Failed to retrieve account for code", "block", header.Number, "hash", header.Hash(), "account", address, "err", err)
 				p.bumpInvalid()
@@ -424,7 +423,7 @@ func handleGetProofs(msg Decoder) (serveRequestFn, uint64, uint64, error) {
 			default:
 				// Account key specified, open a storage trie
 				address := common.BytesToAddress(request.AccountAddress)
-				account, err := getAccount(statedb.TrieDB(), root, address)
+				account, err := getAccount(bc.TrieDB(), root, address)
 				if err != nil {
 					p.Log().Warn("Failed to retrieve account for proof", "block", header.Number, "hash", header.Hash(), "account", address, "err", err)
 					p.bumpInvalid()

les/test_helper.go

@@ -49,6 +49,7 @@ import (
 	"github.com/ethereum/go-ethereum/p2p"
 	"github.com/ethereum/go-ethereum/p2p/enode"
 	"github.com/ethereum/go-ethereum/params"
+	"github.com/ethereum/go-ethereum/trie"
 )
 
 var (
@@ -188,7 +189,7 @@ func newTestClientHandler(backend *backends.SimulatedBackend, odr *LesOdr, index
 			BaseFee:  big.NewInt(params.InitialBaseFee),
 		}
 	)
-	genesis := gspec.MustCommit(db)
+	genesis := gspec.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults))
 	chain, _ := light.NewLightChain(odr, gspec.Config, engine)
 
 	client := &LightEthereum{
@@ -226,7 +227,7 @@ func newTestServerHandler(blocks int, indexers []*core.ChainIndexer, db ethdb.Da
 			BaseFee:  big.NewInt(params.InitialBaseFee),
 		}
 	)
-	genesis := gspec.MustCommit(db)
+	genesis := gspec.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults))
 
 	// create a simulation backend and pre-commit several customized block to the database.
 	simulation := backends.NewSimulatedBackendWithDatabase(db, gspec.Alloc, 100000000)

light/lightchain_test.go

@@ -29,6 +29,7 @@ import (
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/params"
+	"github.com/ethereum/go-ethereum/trie"
 )
 
 // So we can deterministically seed different blockchains
@@ -55,7 +56,7 @@ func makeHeaderChain(parent *types.Header, n int, db ethdb.Database, seed int) [
 func newCanonical(n int) (ethdb.Database, *LightChain, error) {
 	db := rawdb.NewMemoryDatabase()
 	gspec := core.Genesis{Config: params.TestChainConfig}
-	genesis := gspec.MustCommit(db)
+	genesis := gspec.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults))
 	blockchain, _ := NewLightChain(&dummyOdr{db: db, indexerConfig: TestClientIndexerConfig}, gspec.Config, ethash.NewFaker())
 
 	// Create and inject the requested chain
@@ -75,7 +76,7 @@ func newTestLightChain() *LightChain {
 		Difficulty: big.NewInt(1),
 		Config:     params.TestChainConfig,
 	}
-	gspec.MustCommit(db)
+	gspec.MustCommit(db, trie.NewDatabase(db, trie.HashDefaults))
 	lc, err := NewLightChain(&dummyOdr{db: db}, gspec.Config, ethash.NewFullFaker())
 	if err != nil {
 		panic(err)

light/odr_test.go

@@ -36,6 +36,7 @@ import (
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/rlp"
+	"github.com/ethereum/go-ethereum/trie"
 )
 
 var (
@@ -282,7 +283,7 @@ func testChainOdr(t *testing.T, protocol int, fn odrTestFn) {
 		t.Fatal(err)
 	}
 
-	gspec.MustCommit(ldb)
+	gspec.MustCommit(ldb, trie.NewDatabase(ldb, trie.HashDefaults))
 	odr := &testOdr{sdb: sdb, ldb: ldb, serverState: blockchain.StateCache(), indexerConfig: TestClientIndexerConfig}
 	lightchain, err := NewLightChain(odr, gspec.Config, ethash.NewFullFaker())
 	if err != nil {

light/postprocess.go

@@ -145,7 +145,7 @@ func NewChtIndexer(db ethdb.Database, odr OdrBackend, size, confirms uint64, dis
 		diskdb:         db,
 		odr:            odr,
 		trieTable:      trieTable,
-		triedb:         trie.NewDatabaseWithConfig(trieTable, &trie.Config{Cache: 1}), // Use a tiny cache only to keep memory down
+		triedb:         trie.NewDatabase(trieTable, trie.HashDefaults),
 		sectionSize:    size,
 		disablePruning: disablePruning,
 	}
@@ -348,7 +348,7 @@ func NewBloomTrieIndexer(db ethdb.Database, odr OdrBackend, parentSize, size uin
 		diskdb:         db,
 		odr:            odr,
 		trieTable:      trieTable,
-		triedb:         trie.NewDatabaseWithConfig(trieTable, &trie.Config{Cache: 1}), // Use a tiny cache only to keep memory down
+		triedb:         trie.NewDatabase(trieTable, trie.HashDefaults),
 		parentSize:     parentSize,
 		size:           size,
 		disablePruning: disablePruning,

light/trie.go

@@ -215,7 +215,8 @@ func (t *odrTrie) do(key []byte, fn func() error) error {
 			} else {
 				id = trie.StateTrieID(t.id.StateRoot)
 			}
-			t.trie, err = trie.New(id, trie.NewDatabase(t.db.backend.Database()))
+			triedb := trie.NewDatabase(t.db.backend.Database(), trie.HashDefaults)
+			t.trie, err = trie.New(id, triedb)
 		}
 		if err == nil {
 			err = fn()
@@ -247,7 +248,8 @@ func newNodeIterator(t *odrTrie, startkey []byte) trie.NodeIterator {
 			} else {
 				id = trie.StateTrieID(t.id.StateRoot)
 			}
-			t, err := trie.New(id, trie.NewDatabase(t.db.backend.Database()))
+			triedb := trie.NewDatabase(t.db.backend.Database(), trie.HashDefaults)
+			t, err := trie.New(id, triedb)
 			if err == nil {
 				it.t.trie = t
 			}

light/trie_test.go

@@ -50,7 +50,7 @@ func TestNodeIterator(t *testing.T) {
 		panic(err)
 	}
 
-	gspec.MustCommit(lightdb)
+	gspec.MustCommit(lightdb, trie.NewDatabase(lightdb, trie.HashDefaults))
 	ctx := context.Background()
 	odr := &testOdr{sdb: fulldb, ldb: lightdb, serverState: blockchain.StateCache(), indexerConfig: TestClientIndexerConfig}
 	head := blockchain.CurrentHeader()

light/txpool_test.go

@@ -30,6 +30,7 @@ import (
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/params"
+	"github.com/ethereum/go-ethereum/trie"
 )
 
 type testTxRelay struct {
@@ -96,7 +97,7 @@ func TestTxPool(t *testing.T) {
 		panic(err)
 	}
 
-	gspec.MustCommit(ldb)
+	gspec.MustCommit(ldb, trie.NewDatabase(ldb, trie.HashDefaults))
 	odr := &testOdr{sdb: sdb, ldb: ldb, serverState: blockchain.StateCache(), indexerConfig: TestClientIndexerConfig}
 	relay := &testTxRelay{
 		send:    make(chan int, 1),

miner/miner_test.go

@@ -288,8 +288,9 @@ func createMiner(t *testing.T) (*Miner, *event.TypeMux, func(skipMiner bool)) {
 	}
 	// Create chainConfig
 	chainDB := rawdb.NewMemoryDatabase()
+	triedb := trie.NewDatabase(chainDB, nil)
 	genesis := minerTestGenesisBlock(15, 11_500_000, common.HexToAddress("12345"))
-	chainConfig, _, err := core.SetupGenesisBlock(chainDB, trie.NewDatabase(chainDB), genesis)
+	chainConfig, _, err := core.SetupGenesisBlock(chainDB, triedb, genesis)
 	if err != nil {
 		t.Fatalf("can't create new chain config: %v", err)
 	}
@@ -300,7 +301,7 @@ func createMiner(t *testing.T) (*Miner, *event.TypeMux, func(skipMiner bool)) {
 	if err != nil {
 		t.Fatalf("can't create new chain %v", err)
 	}
-	statedb, _ := state.New(types.EmptyRootHash, state.NewDatabase(chainDB), nil)
+	statedb, _ := state.New(bc.Genesis().Root(), bc.StateCache(), nil)
 	blockchain := &testBlockChain{chainConfig, statedb, 10000000, new(event.Feed)}
 
 	pool := legacypool.New(testTxPoolConfig, blockchain)

tests/block_test.go

@@ -18,6 +18,8 @@ package tests
 
 import (
 	"testing"
+
+	"github.com/ethereum/go-ethereum/core/rawdb"
 )
 
 func TestBlockchain(t *testing.T) {
@@ -48,11 +50,17 @@ func TestBlockchain(t *testing.T) {
 	bt.skipLoad(`.*randomStatetest94.json.*`)
 
 	bt.walk(t, blockTestDir, func(t *testing.T, name string, test *BlockTest) {
-		if err := bt.checkFailure(t, test.Run(false, nil)); err != nil {
+		if err := bt.checkFailure(t, test.Run(false, rawdb.HashScheme, nil)); err != nil {
-			t.Errorf("test without snapshotter failed: %v", err)
+			t.Errorf("test in hash mode without snapshotter failed: %v", err)
 		}
-		if err := bt.checkFailure(t, test.Run(true, nil)); err != nil {
+		if err := bt.checkFailure(t, test.Run(true, rawdb.HashScheme, nil)); err != nil {
-			t.Errorf("test with snapshotter failed: %v", err)
+			t.Errorf("test in hash mode with snapshotter failed: %v", err)
+		}
+		if err := bt.checkFailure(t, test.Run(false, rawdb.PathScheme, nil)); err != nil {
+			t.Errorf("test in path mode without snapshotter failed: %v", err)
+		}
+		if err := bt.checkFailure(t, test.Run(true, rawdb.PathScheme, nil)); err != nil {
+			t.Errorf("test in path mode with snapshotter failed: %v", err)
 		}
 	})
 	// There is also a LegacyTests folder, containing blockchain tests generated

tests/block_test_util.go

@@ -38,6 +38,9 @@ import (
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/rlp"
+	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 )
 
 // A BlockTest checks handling of entire blocks.
@@ -100,16 +103,30 @@ type btHeaderMarshaling struct {
 	BaseFeePerGas *math.HexOrDecimal256
 }
 
-func (t *BlockTest) Run(snapshotter bool, tracer vm.EVMLogger) error {
+func (t *BlockTest) Run(snapshotter bool, scheme string, tracer vm.EVMLogger) error {
 	config, ok := Forks[t.json.Network]
 	if !ok {
 		return UnsupportedForkError{t.json.Network}
 	}
-
 	// import pre accounts & construct test genesis block & state root
-	db := rawdb.NewMemoryDatabase()
+	var (
+		db    = rawdb.NewMemoryDatabase()
+		tconf = &trie.Config{}
+	)
+	if scheme == rawdb.PathScheme {
+		tconf.PathDB = pathdb.Defaults
+	} else {
+		tconf.HashDB = hashdb.Defaults
+	}
+	// Commit genesis state
 	gspec := t.genesis(config)
-	gblock := gspec.MustCommit(db)
+	triedb := trie.NewDatabase(db, tconf)
+	gblock, err := gspec.Commit(db, triedb)
+	if err != nil {
+		return err
+	}
+	triedb.Close() // close the db to prevent memory leak
+
 	if gblock.Hash() != t.json.Genesis.Hash {
 		return fmt.Errorf("genesis block hash doesn't match test: computed=%x, test=%x", gblock.Hash().Bytes()[:6], t.json.Genesis.Hash[:6])
 	}
@@ -119,7 +136,7 @@ func (t *BlockTest) Run(snapshotter bool, tracer vm.EVMLogger) error {
 	// Wrap the original engine within the beacon-engine
 	engine := beacon.New(ethash.NewFaker())
 
-	cache := &core.CacheConfig{TrieCleanLimit: 0}
+	cache := &core.CacheConfig{TrieCleanLimit: 0, StateScheme: scheme}
 	if snapshotter {
 		cache.SnapshotLimit = 1
 		cache.SnapshotWait = true

tests/fuzzers/les/les-fuzzer.go

@@ -88,8 +88,8 @@ func makechain() (bc *core.BlockChain, addresses []common.Address, txHashes []co
 }
 
 func makeTries() (chtTrie *trie.Trie, bloomTrie *trie.Trie, chtKeys, bloomKeys [][]byte) {
-	chtTrie = trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase()))
+	chtTrie = trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), trie.HashDefaults))
-	bloomTrie = trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase()))
+	bloomTrie = trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), trie.HashDefaults))
 	for i := 0; i < testChainLen; i++ {
 		// The element in CHT is <big-endian block number> -> <block hash>
 		key := make([]byte, 8)

tests/fuzzers/rangeproof/rangeproof-fuzzer.go

@@ -56,7 +56,7 @@ func (f *fuzzer) readInt() uint64 {
 }
 
 func (f *fuzzer) randomTrie(n int) (*trie.Trie, map[string]*kv) {
-	trie := trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := trie.NewEmpty(trie.NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	vals := make(map[string]*kv)
 	size := f.readInt()
 	// Fill it with some fluff

tests/fuzzers/stacktrie/trie_fuzzer.go

@@ -136,10 +136,10 @@ func (f *fuzzer) fuzz() int {
 	// This spongeDb is used to check the sequence of disk-db-writes
 	var (
 		spongeA = &spongeDb{sponge: sha3.NewLegacyKeccak256()}
-		dbA     = trie.NewDatabase(rawdb.NewDatabase(spongeA))
+		dbA     = trie.NewDatabase(rawdb.NewDatabase(spongeA), nil)
 		trieA   = trie.NewEmpty(dbA)
 		spongeB = &spongeDb{sponge: sha3.NewLegacyKeccak256()}
-		dbB     = trie.NewDatabase(rawdb.NewDatabase(spongeB))
+		dbB     = trie.NewDatabase(rawdb.NewDatabase(spongeB), nil)
 		trieB   = trie.NewStackTrie(func(owner common.Hash, path []byte, hash common.Hash, blob []byte) {
 			rawdb.WriteTrieNode(spongeB, owner, path, hash, blob, dbB.Scheme())
 		})

tests/fuzzers/trie/trie-fuzzer.go

@@ -143,7 +143,7 @@ func Fuzz(input []byte) int {
 
 func runRandTest(rt randTest) error {
 	var (
-		triedb = trie.NewDatabase(rawdb.NewMemoryDatabase())
+		triedb = trie.NewDatabase(rawdb.NewMemoryDatabase(), nil)
 		tr     = trie.NewEmpty(triedb)
 		origin = types.EmptyRootHash
 		values = make(map[string]string) // tracks content of the trie

tests/state_test.go

@@ -30,6 +30,8 @@ import (
 
 	"github.com/ethereum/go-ethereum/core"
 	"github.com/ethereum/go-ethereum/core/rawdb"
+	"github.com/ethereum/go-ethereum/core/state"
+	"github.com/ethereum/go-ethereum/core/state/snapshot"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/core/vm"
 	"github.com/ethereum/go-ethereum/eth/tracers/logger"
@@ -78,21 +80,52 @@ func TestState(t *testing.T) {
 				subtest := subtest
 				key := fmt.Sprintf("%s/%d", subtest.Fork, subtest.Index)
 
-				t.Run(key+"/trie", func(t *testing.T) {
+				t.Run(key+"/hash/trie", func(t *testing.T) {
 					withTrace(t, test.gasLimit(subtest), func(vmconfig vm.Config) error {
-						_, _, err := test.Run(subtest, vmconfig, false)
+						var result error
-						return st.checkFailure(t, err)
+						test.Run(subtest, vmconfig, false, rawdb.HashScheme, func(err error, snaps *snapshot.Tree, state *state.StateDB) {
+							result = st.checkFailure(t, err)
+						})
+						return result
 					})
 				})
-				t.Run(key+"/snap", func(t *testing.T) {
+				t.Run(key+"/hash/snap", func(t *testing.T) {
 					withTrace(t, test.gasLimit(subtest), func(vmconfig vm.Config) error {
-						snaps, statedb, err := test.Run(subtest, vmconfig, true)
+						var result error
-						if snaps != nil && statedb != nil {
+						test.Run(subtest, vmconfig, true, rawdb.HashScheme, func(err error, snaps *snapshot.Tree, state *state.StateDB) {
-							if _, err := snaps.Journal(statedb.IntermediateRoot(false)); err != nil {
+							if snaps != nil && state != nil {
-								return err
+								if _, err := snaps.Journal(state.IntermediateRoot(false)); err != nil {
+									result = err
+									return
 								}
 							}
-						return st.checkFailure(t, err)
+							result = st.checkFailure(t, err)
+						})
+						return result
+					})
+				})
+				t.Run(key+"/path/trie", func(t *testing.T) {
+					withTrace(t, test.gasLimit(subtest), func(vmconfig vm.Config) error {
+						var result error
+						test.Run(subtest, vmconfig, false, rawdb.PathScheme, func(err error, snaps *snapshot.Tree, state *state.StateDB) {
+							result = st.checkFailure(t, err)
+						})
+						return result
+					})
+				})
+				t.Run(key+"/path/snap", func(t *testing.T) {
+					withTrace(t, test.gasLimit(subtest), func(vmconfig vm.Config) error {
+						var result error
+						test.Run(subtest, vmconfig, true, rawdb.PathScheme, func(err error, snaps *snapshot.Tree, state *state.StateDB) {
+							if snaps != nil && state != nil {
+								if _, err := snaps.Journal(state.IntermediateRoot(false)); err != nil {
+									result = err
+									return
+								}
+							}
+							result = st.checkFailure(t, err)
+						})
+						return result
 					})
 				})
 			}
@@ -190,7 +223,8 @@ func runBenchmark(b *testing.B, t *StateTest) {
 
 			vmconfig.ExtraEips = eips
 			block := t.genesis(config).ToBlock()
-			_, statedb := MakePreState(rawdb.NewMemoryDatabase(), t.json.Pre, false)
+			triedb, _, statedb := MakePreState(rawdb.NewMemoryDatabase(), t.json.Pre, false, rawdb.HashScheme)
+			defer triedb.Close()
 
 			var baseFee *big.Int
 			if rules.IsLondon {

tests/state_test_util.go

@@ -39,6 +39,8 @@ import (
 	"github.com/ethereum/go-ethereum/params"
 	"github.com/ethereum/go-ethereum/rlp"
 	"github.com/ethereum/go-ethereum/trie"
+	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
+	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 	"golang.org/x/crypto/sha3"
 )
 
@@ -187,43 +189,50 @@ func (t *StateTest) checkError(subtest StateSubtest, err error) error {
 }
 
 // Run executes a specific subtest and verifies the post-state and logs
-func (t *StateTest) Run(subtest StateSubtest, vmconfig vm.Config, snapshotter bool) (*snapshot.Tree, *state.StateDB, error) {
+func (t *StateTest) Run(subtest StateSubtest, vmconfig vm.Config, snapshotter bool, scheme string, postCheck func(err error, snaps *snapshot.Tree, state *state.StateDB)) (result error) {
-	snaps, statedb, root, err := t.RunNoVerify(subtest, vmconfig, snapshotter)
+	triedb, snaps, statedb, root, err := t.RunNoVerify(subtest, vmconfig, snapshotter, scheme)
-	if checkedErr := t.checkError(subtest, err); checkedErr != nil {
+
-		return snaps, statedb, checkedErr
+	// Invoke the callback at the end of function for further analysis.
+	defer func() {
+		postCheck(result, snaps, statedb)
+
+		if triedb != nil {
+			triedb.Close()
+		}
+	}()
+	checkedErr := t.checkError(subtest, err)
+	if checkedErr != nil {
+		return checkedErr
 	}
 	// The error has been checked; if it was unexpected, it's already returned.
 	if err != nil {
 		// Here, an error exists but it was expected.
 		// We do not check the post state or logs.
-		return snaps, statedb, nil
+		return nil
 	}
 	post := t.json.Post[subtest.Fork][subtest.Index]
 	// N.B: We need to do this in a two-step process, because the first Commit takes care
 	// of self-destructs, and we need to touch the coinbase _after_ it has potentially self-destructed.
 	if root != common.Hash(post.Root) {
-		return snaps, statedb, fmt.Errorf("post state root mismatch: got %x, want %x", root, post.Root)
+		return fmt.Errorf("post state root mismatch: got %x, want %x", root, post.Root)
 	}
 	if logs := rlpHash(statedb.Logs()); logs != common.Hash(post.Logs) {
-		return snaps, statedb, fmt.Errorf("post state logs hash mismatch: got %x, want %x", logs, post.Logs)
+		return fmt.Errorf("post state logs hash mismatch: got %x, want %x", logs, post.Logs)
 	}
-	// Re-init the post-state instance for further operation
+	statedb, _ = state.New(root, statedb.Database(), snaps)
-	statedb, err = state.New(root, statedb.Database(), snaps)
+	return nil
-	if err != nil {
-		return nil, nil, err
-	}
-	return snaps, statedb, nil
 }
 
 // RunNoVerify runs a specific subtest and returns the statedb and post-state root
-func (t *StateTest) RunNoVerify(subtest StateSubtest, vmconfig vm.Config, snapshotter bool) (*snapshot.Tree, *state.StateDB, common.Hash, error) {
+func (t *StateTest) RunNoVerify(subtest StateSubtest, vmconfig vm.Config, snapshotter bool, scheme string) (*trie.Database, *snapshot.Tree, *state.StateDB, common.Hash, error) {
 	config, eips, err := GetChainConfig(subtest.Fork)
 	if err != nil {
-		return nil, nil, common.Hash{}, UnsupportedForkError{subtest.Fork}
+		return nil, nil, nil, common.Hash{}, UnsupportedForkError{subtest.Fork}
 	}
 	vmconfig.ExtraEips = eips
+
 	block := t.genesis(config).ToBlock()
-	snaps, statedb := MakePreState(rawdb.NewMemoryDatabase(), t.json.Pre, snapshotter)
+	triedb, snaps, statedb := MakePreState(rawdb.NewMemoryDatabase(), t.json.Pre, snapshotter, scheme)
 
 	var baseFee *big.Int
 	if config.IsLondon(new(big.Int)) {
@@ -237,7 +246,8 @@ func (t *StateTest) RunNoVerify(subtest StateSubtest, vmconfig vm.Config, snapsh
 	post := t.json.Post[subtest.Fork][subtest.Index]
 	msg, err := t.json.Tx.toMessage(post, baseFee)
 	if err != nil {
-		return nil, nil, common.Hash{}, err
+		triedb.Close()
+		return nil, nil, nil, common.Hash{}, err
 	}
 
 	// Try to recover tx with current signer
@@ -245,11 +255,13 @@ func (t *StateTest) RunNoVerify(subtest StateSubtest, vmconfig vm.Config, snapsh
 		var ttx types.Transaction
 		err := ttx.UnmarshalBinary(post.TxBytes)
 		if err != nil {
-			return nil, nil, common.Hash{}, err
+			triedb.Close()
+			return nil, nil, nil, common.Hash{}, err
 		}
 
 		if _, err := types.Sender(types.LatestSigner(config), &ttx); err != nil {
-			return nil, nil, common.Hash{}, err
+			triedb.Close()
+			return nil, nil, nil, common.Hash{}, err
 		}
 	}
 
@@ -268,6 +280,7 @@ func (t *StateTest) RunNoVerify(subtest StateSubtest, vmconfig vm.Config, snapsh
 		context.Difficulty = big.NewInt(0)
 	}
 	evm := vm.NewEVM(context, txContext, statedb, config, vmconfig)
+
 	// Execute the message.
 	snapshot := statedb.Snapshot()
 	gaspool := new(core.GasPool)
@@ -282,17 +295,25 @@ func (t *StateTest) RunNoVerify(subtest StateSubtest, vmconfig vm.Config, snapsh
 	// - there are only 'bad' transactions, which aren't executed. In those cases,
 	//   the coinbase gets no txfee, so isn't created, and thus needs to be touched
 	statedb.AddBalance(block.Coinbase(), new(big.Int))
-	// Commit block
+
+	// Commit state mutations into database.
 	root, _ := statedb.Commit(block.NumberU64(), config.IsEIP158(block.Number()))
-	return snaps, statedb, root, err
+	return triedb, snaps, statedb, root, err
 }
 
 func (t *StateTest) gasLimit(subtest StateSubtest) uint64 {
 	return t.json.Tx.GasLimit[t.json.Post[subtest.Fork][subtest.Index].Indexes.Gas]
 }
 
-func MakePreState(db ethdb.Database, accounts core.GenesisAlloc, snapshotter bool) (*snapshot.Tree, *state.StateDB) {
+func MakePreState(db ethdb.Database, accounts core.GenesisAlloc, snapshotter bool, scheme string) (*trie.Database, *snapshot.Tree, *state.StateDB) {
-	sdb := state.NewDatabaseWithConfig(db, &trie.Config{Preimages: true})
+	tconf := &trie.Config{Preimages: true}
+	if scheme == rawdb.HashScheme {
+		tconf.HashDB = hashdb.Defaults
+	} else {
+		tconf.PathDB = pathdb.Defaults
+	}
+	triedb := trie.NewDatabase(db, tconf)
+	sdb := state.NewDatabaseWithNodeDB(db, triedb)
 	statedb, _ := state.New(types.EmptyRootHash, sdb, nil)
 	for addr, a := range accounts {
 		statedb.SetCode(addr, a.Code)
@@ -313,10 +334,10 @@ func MakePreState(db ethdb.Database, accounts core.GenesisAlloc, snapshotter boo
 			NoBuild:    false,
 			AsyncBuild: false,
 		}
-		snaps, _ = snapshot.New(snapconfig, db, sdb.TrieDB(), root)
+		snaps, _ = snapshot.New(snapconfig, db, triedb, root)
 	}
 	statedb, _ = state.New(root, sdb, snaps)
-	return snaps, statedb
+	return triedb, snaps, statedb
 }
 
 func (t *StateTest) genesis(config *params.ChainConfig) *core.Genesis {

trie/database.go

@@ -21,6 +21,7 @@ import (
 
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/ethdb"
+	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
 	"github.com/ethereum/go-ethereum/trie/triedb/pathdb"
 	"github.com/ethereum/go-ethereum/trie/trienode"
@@ -29,14 +30,21 @@ import (
 
 // Config defines all necessary options for database.
 type Config struct {
-	Cache     int            // Memory allowance (MB) to use for caching trie nodes in memory
+	Preimages bool           // Flag whether the preimage of node key is recorded
-	Preimages bool           // Flag whether the preimage of trie key is recorded
+	HashDB    *hashdb.Config // Configs for hash-based scheme
-	PathDB    *pathdb.Config // Configs for experimental path-based scheme, not used yet.
+	PathDB    *pathdb.Config // Configs for experimental path-based scheme
 
 	// Testing hooks
 	OnCommit func(states *triestate.Set) // Hook invoked when commit is performed
 }
 
+// HashDefaults represents a config for using hash-based scheme with
+// default settings.
+var HashDefaults = &Config{
+	Preimages: false,
+	HashDB:    hashdb.Defaults,
+}
+
 // backend defines the methods needed to access/update trie nodes in different
 // state scheme.
 type backend interface {
@@ -91,22 +99,30 @@ func prepare(diskdb ethdb.Database, config *Config) *Database {
 	}
 }
 
-// NewDatabase initializes the trie database with default settings, namely
+// NewDatabase initializes the trie database with default settings, note
 // the legacy hash-based scheme is used by default.
-func NewDatabase(diskdb ethdb.Database) *Database {
+func NewDatabase(diskdb ethdb.Database, config *Config) *Database {
-	return NewDatabaseWithConfig(diskdb, nil)
+	// Sanitize the config and use the default one if it's not specified.
-}
+	if config == nil {
-
+		config = HashDefaults
-// NewDatabaseWithConfig initializes the trie database with provided configs.
+	}
-// The path-based scheme is not activated yet, always initialized with legacy
+	var preimages *preimageStore
-// hash-based scheme by default.
+	if config.Preimages {
-func NewDatabaseWithConfig(diskdb ethdb.Database, config *Config) *Database {
+		preimages = newPreimageStore(diskdb)
-	var cleans int
+	}
-	if config != nil && config.Cache != 0 {
+	db := &Database{
-		cleans = config.Cache * 1024 * 1024
+		config:    config,
+		diskdb:    diskdb,
+		preimages: preimages,
+	}
+	if config.HashDB != nil && config.PathDB != nil {
+		log.Crit("Both 'hash' and 'path' mode are configured")
+	}
+	if config.PathDB != nil {
+		db.backend = pathdb.New(diskdb, config.PathDB)
+	} else {
+		db.backend = hashdb.New(diskdb, config.HashDB, mptResolver{})
 	}
-	db := prepare(diskdb, config)
-	db.backend = hashdb.New(diskdb, cleans, mptResolver{})
 	return db
 }
 
@@ -240,3 +256,60 @@ func (db *Database) Node(hash common.Hash) ([]byte, error) {
 	}
 	return hdb.Node(hash)
 }
+
+// Recover rollbacks the database to a specified historical point. The state is
+// supported as the rollback destination only if it's canonical state and the
+// corresponding trie histories are existent. It's only supported by path-based
+// database and will return an error for others.
+func (db *Database) Recover(target common.Hash) error {
+	pdb, ok := db.backend.(*pathdb.Database)
+	if !ok {
+		return errors.New("not supported")
+	}
+	return pdb.Recover(target, &trieLoader{db: db})
+}
+
+// Recoverable returns the indicator if the specified state is enabled to be
+// recovered. It's only supported by path-based database and will return an
+// error for others.
+func (db *Database) Recoverable(root common.Hash) (bool, error) {
+	pdb, ok := db.backend.(*pathdb.Database)
+	if !ok {
+		return false, errors.New("not supported")
+	}
+	return pdb.Recoverable(root), nil
+}
+
+// Reset wipes all available journal from the persistent database and discard
+// all caches and diff layers. Using the given root to create a new disk layer.
+// It's only supported by path-based database and will return an error for others.
+func (db *Database) Reset(root common.Hash) error {
+	pdb, ok := db.backend.(*pathdb.Database)
+	if !ok {
+		return errors.New("not supported")
+	}
+	return pdb.Reset(root)
+}
+
+// Journal commits an entire diff hierarchy to disk into a single journal entry.
+// This is meant to be used during shutdown to persist the snapshot without
+// flattening everything down (bad for reorgs). It's only supported by path-based
+// database and will return an error for others.
+func (db *Database) Journal(root common.Hash) error {
+	pdb, ok := db.backend.(*pathdb.Database)
+	if !ok {
+		return errors.New("not supported")
+	}
+	return pdb.Journal(root)
+}
+
+// SetBufferSize sets the node buffer size to the provided value(in bytes).
+// It's only supported by path-based database and will return an error for
+// others.
+func (db *Database) SetBufferSize(size int) error {
+	pdb, ok := db.backend.(*pathdb.Database)
+	if !ok {
+		return errors.New("not supported")
+	}
+	return pdb.SetBufferSize(size)
+}

trie/database_test.go

@@ -27,7 +27,7 @@ import (
 func newTestDatabase(diskdb ethdb.Database, scheme string) *Database {
 	db := prepare(diskdb, nil)
 	if scheme == rawdb.HashScheme {
-		db.backend = hashdb.New(diskdb, 0, mptResolver{})
+		db.backend = hashdb.New(diskdb, &hashdb.Config{}, mptResolver{})
 	} else {
 		db.backend = pathdb.New(diskdb, &pathdb.Config{}) // disable clean/dirty cache
 	}

trie/iterator_test.go

@@ -18,7 +18,6 @@ package trie
 
 import (
 	"bytes"
-	"encoding/binary"
 	"fmt"
 	"math/rand"
 	"testing"
@@ -27,13 +26,11 @@ import (
 	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/crypto"
-	"github.com/ethereum/go-ethereum/ethdb"
-	"github.com/ethereum/go-ethereum/ethdb/memorydb"
 	"github.com/ethereum/go-ethereum/trie/trienode"
 )
 
 func TestEmptyIterator(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	iter := trie.MustNodeIterator(nil)
 
 	seen := make(map[string]struct{})
@@ -46,7 +43,7 @@ func TestEmptyIterator(t *testing.T) {
 }
 
 func TestIterator(t *testing.T) {
-	db := NewDatabase(rawdb.NewMemoryDatabase())
+	db := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie := NewEmpty(db)
 	vals := []struct{ k, v string }{
 		{"do", "verb"},
@@ -89,7 +86,7 @@ func (k *kv) less(other *kv) bool {
 }
 
 func TestIteratorLargeData(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	vals := make(map[string]*kv)
 
 	for i := byte(0); i < 255; i++ {
@@ -208,7 +205,7 @@ var testdata2 = []kvs{
 }
 
 func TestIteratorSeek(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for _, val := range testdata1 {
 		trie.MustUpdate([]byte(val.k), []byte(val.v))
 	}
@@ -249,7 +246,7 @@ func checkIteratorOrder(want []kvs, it *Iterator) error {
 }
 
 func TestDifferenceIterator(t *testing.T) {
-	dba := NewDatabase(rawdb.NewMemoryDatabase())
+	dba := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	triea := NewEmpty(dba)
 	for _, val := range testdata1 {
 		triea.MustUpdate([]byte(val.k), []byte(val.v))
@@ -258,7 +255,7 @@ func TestDifferenceIterator(t *testing.T) {
 	dba.Update(rootA, types.EmptyRootHash, 0, trienode.NewWithNodeSet(nodesA), nil)
 	triea, _ = New(TrieID(rootA), dba)
 
-	dbb := NewDatabase(rawdb.NewMemoryDatabase())
+	dbb := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trieb := NewEmpty(dbb)
 	for _, val := range testdata2 {
 		trieb.MustUpdate([]byte(val.k), []byte(val.v))
@@ -291,7 +288,7 @@ func TestDifferenceIterator(t *testing.T) {
 }
 
 func TestUnionIterator(t *testing.T) {
-	dba := NewDatabase(rawdb.NewMemoryDatabase())
+	dba := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	triea := NewEmpty(dba)
 	for _, val := range testdata1 {
 		triea.MustUpdate([]byte(val.k), []byte(val.v))
@@ -300,7 +297,7 @@ func TestUnionIterator(t *testing.T) {
 	dba.Update(rootA, types.EmptyRootHash, 0, trienode.NewWithNodeSet(nodesA), nil)
 	triea, _ = New(TrieID(rootA), dba)
 
-	dbb := NewDatabase(rawdb.NewMemoryDatabase())
+	dbb := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trieb := NewEmpty(dbb)
 	for _, val := range testdata2 {
 		trieb.MustUpdate([]byte(val.k), []byte(val.v))
@@ -344,7 +341,7 @@ func TestUnionIterator(t *testing.T) {
 }
 
 func TestIteratorNoDups(t *testing.T) {
-	tr := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	tr := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for _, val := range testdata1 {
 		tr.MustUpdate([]byte(val.k), []byte(val.v))
 	}
@@ -537,96 +534,6 @@ func TestIteratorNodeBlob(t *testing.T) {
 	testIteratorNodeBlob(t, rawdb.PathScheme)
 }
 
-type loggingDb struct {
-	getCount uint64
-	backend  ethdb.KeyValueStore
-}
-
-func (l *loggingDb) Has(key []byte) (bool, error) {
-	return l.backend.Has(key)
-}
-
-func (l *loggingDb) Get(key []byte) ([]byte, error) {
-	l.getCount++
-	return l.backend.Get(key)
-}
-
-func (l *loggingDb) Put(key []byte, value []byte) error {
-	return l.backend.Put(key, value)
-}
-
-func (l *loggingDb) Delete(key []byte) error {
-	return l.backend.Delete(key)
-}
-
-func (l *loggingDb) NewBatch() ethdb.Batch {
-	return l.backend.NewBatch()
-}
-
-func (l *loggingDb) NewBatchWithSize(size int) ethdb.Batch {
-	return l.backend.NewBatchWithSize(size)
-}
-
-func (l *loggingDb) NewIterator(prefix []byte, start []byte) ethdb.Iterator {
-	return l.backend.NewIterator(prefix, start)
-}
-
-func (l *loggingDb) NewSnapshot() (ethdb.Snapshot, error) {
-	return l.backend.NewSnapshot()
-}
-
-func (l *loggingDb) Stat(property string) (string, error) {
-	return l.backend.Stat(property)
-}
-
-func (l *loggingDb) Compact(start []byte, limit []byte) error {
-	return l.backend.Compact(start, limit)
-}
-
-func (l *loggingDb) Close() error {
-	return l.backend.Close()
-}
-
-// makeLargeTestTrie create a sample test trie
-func makeLargeTestTrie() (*Database, *StateTrie, *loggingDb) {
-	// Create an empty trie
-	logDb := &loggingDb{0, memorydb.New()}
-	triedb := NewDatabase(rawdb.NewDatabase(logDb))
-	trie, _ := NewStateTrie(TrieID(types.EmptyRootHash), triedb)
-
-	// Fill it with some arbitrary data
-	for i := 0; i < 10000; i++ {
-		key := make([]byte, 32)
-		val := make([]byte, 32)
-		binary.BigEndian.PutUint64(key, uint64(i))
-		binary.BigEndian.PutUint64(val, uint64(i))
-		key = crypto.Keccak256(key)
-		val = crypto.Keccak256(val)
-		trie.MustUpdate(key, val)
-	}
-	root, nodes, _ := trie.Commit(false)
-	triedb.Update(root, types.EmptyRootHash, 0, trienode.NewWithNodeSet(nodes), nil)
-	triedb.Commit(root, false)
-
-	// Return the generated trie
-	trie, _ = NewStateTrie(TrieID(root), triedb)
-	return triedb, trie, logDb
-}
-
-// Tests that the node iterator indeed walks over the entire database contents.
-func TestNodeIteratorLargeTrie(t *testing.T) {
-	// Create some arbitrary test trie to iterate
-	db, trie, logDb := makeLargeTestTrie()
-	db.Cap(0) // flush everything
-	// Do a seek operation
-	trie.NodeIterator(common.FromHex("0x77667766776677766778855885885885"))
-	// master: 24 get operations
-	// this pr: 6 get operations
-	if have, want := logDb.getCount, uint64(6); have != want {
-		t.Fatalf("Too many lookups during seek, have %d want %d", have, want)
-	}
-}
-
 func testIteratorNodeBlob(t *testing.T, scheme string) {
 	var (
 		db     = rawdb.NewMemoryDatabase()
@@ -700,7 +607,7 @@ func isTrieNode(scheme string, key, val []byte) (bool, []byte, common.Hash) {
 		}
 		hash = common.BytesToHash(key)
 	} else {
-		ok, remain := rawdb.IsAccountTrieNode(key)
+		ok, remain := rawdb.ResolveAccountTrieNodeKey(key)
 		if !ok {
 			return false, nil, common.Hash{}
 		}

trie/proof_test.go

@@ -94,7 +94,7 @@ func TestProof(t *testing.T) {
 }
 
 func TestOneElementProof(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	updateString(trie, "k", "v")
 	for i, prover := range makeProvers(trie) {
 		proof := prover([]byte("k"))
@@ -145,7 +145,7 @@ func TestBadProof(t *testing.T) {
 // Tests that missing keys can also be proven. The test explicitly uses a single
 // entry trie and checks for missing keys both before and after the single entry.
 func TestMissingKeyProof(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	updateString(trie, "k", "v")
 
 	for i, key := range []string{"a", "j", "l", "z"} {
@@ -395,7 +395,7 @@ func TestOneElementRangeProof(t *testing.T) {
 	}
 
 	// Test the mini trie with only a single element.
-	tinyTrie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	tinyTrie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	entry := &kv{randBytes(32), randBytes(20), false}
 	tinyTrie.MustUpdate(entry.k, entry.v)
 
@@ -467,7 +467,7 @@ func TestAllElementsProof(t *testing.T) {
 // TestSingleSideRangeProof tests the range starts from zero.
 func TestSingleSideRangeProof(t *testing.T) {
 	for i := 0; i < 64; i++ {
-		trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+		trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 		var entries []*kv
 		for i := 0; i < 4096; i++ {
 			value := &kv{randBytes(32), randBytes(20), false}
@@ -502,7 +502,7 @@ func TestSingleSideRangeProof(t *testing.T) {
 // TestReverseSingleSideRangeProof tests the range ends with 0xffff...fff.
 func TestReverseSingleSideRangeProof(t *testing.T) {
 	for i := 0; i < 64; i++ {
-		trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+		trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 		var entries []*kv
 		for i := 0; i < 4096; i++ {
 			value := &kv{randBytes(32), randBytes(20), false}
@@ -609,7 +609,7 @@ func TestBadRangeProof(t *testing.T) {
 // TestGappedRangeProof focuses on the small trie with embedded nodes.
 // If the gapped node is embedded in the trie, it should be detected too.
 func TestGappedRangeProof(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	var entries []*kv // Sorted entries
 	for i := byte(0); i < 10; i++ {
 		value := &kv{common.LeftPadBytes([]byte{i}, 32), []byte{i}, false}
@@ -683,7 +683,7 @@ func TestSameSideProofs(t *testing.T) {
 }
 
 func TestHasRightElement(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	var entries []*kv
 	for i := 0; i < 4096; i++ {
 		value := &kv{randBytes(32), randBytes(20), false}
@@ -1036,7 +1036,7 @@ func benchmarkVerifyRangeNoProof(b *testing.B, size int) {
 }
 
 func randomTrie(n int) (*Trie, map[string]*kv) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	vals := make(map[string]*kv)
 	for i := byte(0); i < 100; i++ {
 		value := &kv{common.LeftPadBytes([]byte{i}, 32), []byte{i}, false}
@@ -1055,7 +1055,7 @@ func randomTrie(n int) (*Trie, map[string]*kv) {
 }
 
 func nonRandomTrie(n int) (*Trie, map[string]*kv) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	vals := make(map[string]*kv)
 	max := uint64(0xffffffffffffffff)
 	for i := uint64(0); i < uint64(n); i++ {
@@ -1080,7 +1080,7 @@ func TestRangeProofKeysWithSharedPrefix(t *testing.T) {
 		common.Hex2Bytes("02"),
 		common.Hex2Bytes("03"),
 	}
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for i, key := range keys {
 		trie.MustUpdate(key, vals[i])
 	}

trie/secure_trie_test.go

@@ -31,14 +31,14 @@ import (
 )
 
 func newEmptySecure() *StateTrie {
-	trie, _ := NewStateTrie(TrieID(types.EmptyRootHash), NewDatabase(rawdb.NewMemoryDatabase()))
+	trie, _ := NewStateTrie(TrieID(types.EmptyRootHash), NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	return trie
 }
 
 // makeTestStateTrie creates a large enough secure trie for testing.
 func makeTestStateTrie() (*Database, *StateTrie, map[string][]byte) {
 	// Create an empty trie
-	triedb := NewDatabase(rawdb.NewMemoryDatabase())
+	triedb := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie, _ := NewStateTrie(TrieID(types.EmptyRootHash), triedb)
 
 	// Fill it with some arbitrary data

trie/stacktrie_test.go

@@ -188,7 +188,7 @@ func TestStackTrieInsertAndHash(t *testing.T) {
 
 func TestSizeBug(t *testing.T) {
 	st := NewStackTrie(nil)
-	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 
 	leaf := common.FromHex("290decd9548b62a8d60345a988386fc84ba6bc95484008f6362f93160ef3e563")
 	value := common.FromHex("94cf40d0d2b44f2b66e07cace1372ca42b73cf21a3")
@@ -203,7 +203,7 @@ func TestSizeBug(t *testing.T) {
 
 func TestEmptyBug(t *testing.T) {
 	st := NewStackTrie(nil)
-	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 
 	//leaf := common.FromHex("290decd9548b62a8d60345a988386fc84ba6bc95484008f6362f93160ef3e563")
 	//value := common.FromHex("94cf40d0d2b44f2b66e07cace1372ca42b73cf21a3")
@@ -229,7 +229,7 @@ func TestEmptyBug(t *testing.T) {
 
 func TestValLength56(t *testing.T) {
 	st := NewStackTrie(nil)
-	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 
 	//leaf := common.FromHex("290decd9548b62a8d60345a988386fc84ba6bc95484008f6362f93160ef3e563")
 	//value := common.FromHex("94cf40d0d2b44f2b66e07cace1372ca42b73cf21a3")
@@ -254,7 +254,7 @@ func TestValLength56(t *testing.T) {
 // which causes a lot of node-within-node. This case was found via fuzzing.
 func TestUpdateSmallNodes(t *testing.T) {
 	st := NewStackTrie(nil)
-	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	kvs := []struct {
 		K string
 		V string
@@ -282,7 +282,7 @@ func TestUpdateSmallNodes(t *testing.T) {
 func TestUpdateVariableKeys(t *testing.T) {
 	t.SkipNow()
 	st := NewStackTrie(nil)
-	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	nt := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	kvs := []struct {
 		K string
 		V string
@@ -351,7 +351,7 @@ func TestStacktrieNotModifyValues(t *testing.T) {
 func TestStacktrieSerialization(t *testing.T) {
 	var (
 		st       = NewStackTrie(nil)
-		nt       = NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+		nt       = NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 		keyB     = big.NewInt(1)
 		keyDelta = big.NewInt(1)
 		vals     [][]byte

trie/sync_test.go

@@ -109,8 +109,8 @@ type trieElement struct {
 
 // Tests that an empty trie is not scheduled for syncing.
 func TestEmptySync(t *testing.T) {
-	dbA := NewDatabase(rawdb.NewMemoryDatabase())
+	dbA := newTestDatabase(rawdb.NewMemoryDatabase(), rawdb.HashScheme)
-	dbB := NewDatabase(rawdb.NewMemoryDatabase())
+	dbB := newTestDatabase(rawdb.NewMemoryDatabase(), rawdb.HashScheme)
 	dbC := newTestDatabase(rawdb.NewMemoryDatabase(), rawdb.PathScheme)
 	dbD := newTestDatabase(rawdb.NewMemoryDatabase(), rawdb.PathScheme)
 

trie/tracer_test.go

@@ -61,7 +61,7 @@ func TestTrieTracer(t *testing.T) {
 // Tests if the trie diffs are tracked correctly. Tracer should capture
 // all non-leaf dirty nodes, no matter the node is embedded or not.
 func testTrieTracer(t *testing.T, vals []struct{ k, v string }) {
-	db := NewDatabase(rawdb.NewMemoryDatabase())
+	db := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie := NewEmpty(db)
 
 	// Determine all new nodes are tracked
@@ -104,7 +104,7 @@ func TestTrieTracerNoop(t *testing.T) {
 }
 
 func testTrieTracerNoop(t *testing.T, vals []struct{ k, v string }) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for _, val := range vals {
 		trie.MustUpdate([]byte(val.k), []byte(val.v))
 	}
@@ -128,7 +128,7 @@ func TestAccessList(t *testing.T) {
 
 func testAccessList(t *testing.T, vals []struct{ k, v string }) {
 	var (
-		db   = NewDatabase(rawdb.NewMemoryDatabase())
+		db   = NewDatabase(rawdb.NewMemoryDatabase(), nil)
 		trie = NewEmpty(db)
 		orig = trie.Copy()
 	)
@@ -211,7 +211,7 @@ func testAccessList(t *testing.T, vals []struct{ k, v string }) {
 // Tests origin values won't be tracked in Iterator or Prover
 func TestAccessListLeak(t *testing.T) {
 	var (
-		db   = NewDatabase(rawdb.NewMemoryDatabase())
+		db   = NewDatabase(rawdb.NewMemoryDatabase(), nil)
 		trie = NewEmpty(db)
 	)
 	// Create trie from scratch
@@ -262,7 +262,7 @@ func TestAccessListLeak(t *testing.T) {
 // in its parent due to the smaller size of the original tree node.
 func TestTinyTree(t *testing.T) {
 	var (
-		db   = NewDatabase(rawdb.NewMemoryDatabase())
+		db   = NewDatabase(rawdb.NewMemoryDatabase(), nil)
 		trie = NewEmpty(db)
 	)
 	for _, val := range tiny {

trie/trie_reader.go

@@ -20,6 +20,7 @@ import (
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/log"
+	"github.com/ethereum/go-ethereum/trie/triestate"
 )
 
 // Reader wraps the Node method of a backing trie store.
@@ -83,3 +84,18 @@ func (r *trieReader) node(path []byte, hash common.Hash) ([]byte, error) {
 	}
 	return blob, nil
 }
+
+// trieLoader implements triestate.TrieLoader for constructing tries.
+type trieLoader struct {
+	db *Database
+}
+
+// OpenTrie opens the main account trie.
+func (l *trieLoader) OpenTrie(root common.Hash) (triestate.Trie, error) {
+	return New(TrieID(root), l.db)
+}
+
+// OpenStorageTrie opens the storage trie of an account.
+func (l *trieLoader) OpenStorageTrie(stateRoot common.Hash, addrHash, root common.Hash) (triestate.Trie, error) {
+	return New(StorageTrieID(stateRoot, addrHash, root), l.db)
+}

trie/trie_test.go

@@ -45,7 +45,7 @@ func init() {
 }
 
 func TestEmptyTrie(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	res := trie.Hash()
 	exp := types.EmptyRootHash
 	if res != exp {
@@ -54,7 +54,7 @@ func TestEmptyTrie(t *testing.T) {
 }
 
 func TestNull(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	key := make([]byte, 32)
 	value := []byte("test")
 	trie.MustUpdate(key, value)
@@ -64,8 +64,13 @@ func TestNull(t *testing.T) {
 }
 
 func TestMissingRoot(t *testing.T) {
+	testMissingRoot(t, rawdb.HashScheme)
+	testMissingRoot(t, rawdb.PathScheme)
+}
+
+func testMissingRoot(t *testing.T, scheme string) {
 	root := common.HexToHash("0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33")
-	trie, err := New(TrieID(root), NewDatabase(rawdb.NewMemoryDatabase()))
+	trie, err := New(TrieID(root), newTestDatabase(rawdb.NewMemoryDatabase(), scheme))
 	if trie != nil {
 		t.Error("New returned non-nil trie for invalid root")
 	}
@@ -161,7 +166,7 @@ func testMissingNode(t *testing.T, memonly bool, scheme string) {
 }
 
 func TestInsert(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 
 	updateString(trie, "doe", "reindeer")
 	updateString(trie, "dog", "puppy")
@@ -173,7 +178,7 @@ func TestInsert(t *testing.T) {
 		t.Errorf("case 1: exp %x got %x", exp, root)
 	}
 
-	trie = NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie = NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	updateString(trie, "A", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa")
 
 	exp = common.HexToHash("d23786fb4a010da3ce639d66d5e904a11dbc02746d1ce25029e53290cabf28ab")
@@ -184,7 +189,7 @@ func TestInsert(t *testing.T) {
 }
 
 func TestGet(t *testing.T) {
-	db := NewDatabase(rawdb.NewMemoryDatabase())
+	db := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie := NewEmpty(db)
 	updateString(trie, "doe", "reindeer")
 	updateString(trie, "dog", "puppy")
@@ -209,7 +214,7 @@ func TestGet(t *testing.T) {
 }
 
 func TestDelete(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	vals := []struct{ k, v string }{
 		{"do", "verb"},
 		{"ether", "wookiedoo"},
@@ -236,7 +241,7 @@ func TestDelete(t *testing.T) {
 }
 
 func TestEmptyValues(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 
 	vals := []struct{ k, v string }{
 		{"do", "verb"},
@@ -260,7 +265,7 @@ func TestEmptyValues(t *testing.T) {
 }
 
 func TestReplication(t *testing.T) {
-	db := NewDatabase(rawdb.NewMemoryDatabase())
+	db := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie := NewEmpty(db)
 	vals := []struct{ k, v string }{
 		{"do", "verb"},
@@ -321,7 +326,7 @@ func TestReplication(t *testing.T) {
 }
 
 func TestLargeValue(t *testing.T) {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	trie.MustUpdate([]byte("key1"), []byte{99, 99, 99, 99})
 	trie.MustUpdate([]byte("key2"), bytes.Repeat([]byte{1}, 32))
 	trie.Hash()
@@ -604,7 +609,7 @@ func BenchmarkUpdateLE(b *testing.B) { benchUpdate(b, binary.LittleEndian) }
 const benchElemCount = 20000
 
 func benchGet(b *testing.B) {
-	triedb := NewDatabase(rawdb.NewMemoryDatabase())
+	triedb := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie := NewEmpty(triedb)
 	k := make([]byte, 32)
 	for i := 0; i < benchElemCount; i++ {
@@ -621,7 +626,7 @@ func benchGet(b *testing.B) {
 }
 
 func benchUpdate(b *testing.B, e binary.ByteOrder) *Trie {
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	k := make([]byte, 32)
 	b.ReportAllocs()
 	for i := 0; i < b.N; i++ {
@@ -651,7 +656,7 @@ func BenchmarkHash(b *testing.B) {
 	// entries, then adding N more.
 	addresses, accounts := makeAccounts(2 * b.N)
 	// Insert the accounts into the trie and hash it
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	i := 0
 	for ; i < len(addresses)/2; i++ {
 		trie.MustUpdate(crypto.Keccak256(addresses[i][:]), accounts[i])
@@ -682,7 +687,7 @@ func BenchmarkCommitAfterHash(b *testing.B) {
 func benchmarkCommitAfterHash(b *testing.B, collectLeaf bool) {
 	// Make the random benchmark deterministic
 	addresses, accounts := makeAccounts(b.N)
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for i := 0; i < len(addresses); i++ {
 		trie.MustUpdate(crypto.Keccak256(addresses[i][:]), accounts[i])
 	}
@@ -696,7 +701,7 @@ func benchmarkCommitAfterHash(b *testing.B, collectLeaf bool) {
 func TestTinyTrie(t *testing.T) {
 	// Create a realistic account trie to hash
 	_, accounts := makeAccounts(5)
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	trie.MustUpdate(common.Hex2Bytes("0000000000000000000000000000000000000000000000000000000000001337"), accounts[3])
 	if exp, root := common.HexToHash("8c6a85a4d9fda98feff88450299e574e5378e32391f75a055d470ac0653f1005"), trie.Hash(); exp != root {
 		t.Errorf("1: got %x, exp %x", root, exp)
@@ -709,7 +714,7 @@ func TestTinyTrie(t *testing.T) {
 	if exp, root := common.HexToHash("0608c1d1dc3905fa22204c7a0e43644831c3b6d3def0f274be623a948197e64a"), trie.Hash(); exp != root {
 		t.Errorf("3: got %x, exp %x", root, exp)
 	}
-	checktr := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	checktr := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	it := NewIterator(trie.MustNodeIterator(nil))
 	for it.Next() {
 		checktr.MustUpdate(it.Key, it.Value)
@@ -722,7 +727,7 @@ func TestTinyTrie(t *testing.T) {
 func TestCommitAfterHash(t *testing.T) {
 	// Create a realistic account trie to hash
 	addresses, accounts := makeAccounts(1000)
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for i := 0; i < len(addresses); i++ {
 		trie.MustUpdate(crypto.Keccak256(addresses[i][:]), accounts[i])
 	}
@@ -788,11 +793,17 @@ func (s *spongeDb) Stat(property string) (string, error)     { panic("implement
 func (s *spongeDb) Compact(start []byte, limit []byte) error { panic("implement me") }
 func (s *spongeDb) Close() error                             { return nil }
 func (s *spongeDb) Put(key []byte, value []byte) error {
-	valbrief := value
+	var (
+		keybrief = key
+		valbrief = value
+	)
+	if len(keybrief) > 8 {
+		keybrief = keybrief[:8]
+	}
 	if len(valbrief) > 8 {
 		valbrief = valbrief[:8]
 	}
-	s.journal = append(s.journal, fmt.Sprintf("%v: PUT([%x...], [%d bytes] %x...)\n", s.id, key[:8], len(value), valbrief))
+	s.journal = append(s.journal, fmt.Sprintf("%v: PUT([%x...], [%d bytes] %x...)\n", s.id, keybrief, len(value), valbrief))
 	s.sponge.Write(key)
 	s.sponge.Write(value)
 	return nil
@@ -830,7 +841,7 @@ func TestCommitSequence(t *testing.T) {
 		addresses, accounts := makeAccounts(tc.count)
 		// This spongeDb is used to check the sequence of disk-db-writes
 		s := &spongeDb{sponge: sha3.NewLegacyKeccak256()}
-		db := NewDatabase(rawdb.NewDatabase(s))
+		db := NewDatabase(rawdb.NewDatabase(s), nil)
 		trie := NewEmpty(db)
 		// Fill the trie with elements
 		for i := 0; i < tc.count; i++ {
@@ -861,7 +872,7 @@ func TestCommitSequenceRandomBlobs(t *testing.T) {
 		prng := rand.New(rand.NewSource(int64(i)))
 		// This spongeDb is used to check the sequence of disk-db-writes
 		s := &spongeDb{sponge: sha3.NewLegacyKeccak256()}
-		db := NewDatabase(rawdb.NewDatabase(s))
+		db := NewDatabase(rawdb.NewDatabase(s), nil)
 		trie := NewEmpty(db)
 		// Fill the trie with elements
 		for i := 0; i < tc.count; i++ {
@@ -893,7 +904,7 @@ func TestCommitSequenceStackTrie(t *testing.T) {
 		prng := rand.New(rand.NewSource(int64(count)))
 		// This spongeDb is used to check the sequence of disk-db-writes
 		s := &spongeDb{sponge: sha3.NewLegacyKeccak256(), id: "a"}
-		db := NewDatabase(rawdb.NewDatabase(s))
+		db := NewDatabase(rawdb.NewDatabase(s), nil)
 		trie := NewEmpty(db)
 		// Another sponge is used for the stacktrie commits
 		stackTrieSponge := &spongeDb{sponge: sha3.NewLegacyKeccak256(), id: "b"}
@@ -952,7 +963,7 @@ func TestCommitSequenceStackTrie(t *testing.T) {
 // not fit into 32 bytes, rlp-encoded. However, it's still the correct thing to do.
 func TestCommitSequenceSmallRoot(t *testing.T) {
 	s := &spongeDb{sponge: sha3.NewLegacyKeccak256(), id: "a"}
-	db := NewDatabase(rawdb.NewDatabase(s))
+	db := NewDatabase(rawdb.NewDatabase(s), nil)
 	trie := NewEmpty(db)
 	// Another sponge is used for the stacktrie commits
 	stackTrieSponge := &spongeDb{sponge: sha3.NewLegacyKeccak256(), id: "b"}
@@ -1029,7 +1040,7 @@ func BenchmarkHashFixedSize(b *testing.B) {
 
 func benchmarkHashFixedSize(b *testing.B, addresses [][20]byte, accounts [][]byte) {
 	b.ReportAllocs()
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for i := 0; i < len(addresses); i++ {
 		trie.MustUpdate(crypto.Keccak256(addresses[i][:]), accounts[i])
 	}
@@ -1080,7 +1091,7 @@ func BenchmarkCommitAfterHashFixedSize(b *testing.B) {
 
 func benchmarkCommitAfterHashFixedSize(b *testing.B, addresses [][20]byte, accounts [][]byte) {
 	b.ReportAllocs()
-	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase()))
+	trie := NewEmpty(NewDatabase(rawdb.NewMemoryDatabase(), nil))
 	for i := 0; i < len(addresses); i++ {
 		trie.MustUpdate(crypto.Keccak256(addresses[i][:]), accounts[i])
 	}
@@ -1132,7 +1143,7 @@ func BenchmarkDerefRootFixedSize(b *testing.B) {
 
 func benchmarkDerefRootFixedSize(b *testing.B, addresses [][20]byte, accounts [][]byte) {
 	b.ReportAllocs()
-	triedb := NewDatabase(rawdb.NewMemoryDatabase())
+	triedb := NewDatabase(rawdb.NewMemoryDatabase(), nil)
 	trie := NewEmpty(triedb)
 	for i := 0; i < len(addresses); i++ {
 		trie.MustUpdate(crypto.Keccak256(addresses[i][:]), accounts[i])

trie/triedb/hashdb/database.go

@@ -65,6 +65,20 @@ type ChildResolver interface {
 	ForEach(node []byte, onChild func(common.Hash))
 }
 
+// Config contains the settings for database.
+type Config struct {
+	CleanCacheSize int // Maximum memory allowance (in bytes) for caching clean nodes
+}
+
+// Defaults is the default setting for database if it's not specified.
+// Notably, clean cache is disabled explicitly,
+var Defaults = &Config{
+	// Explicitly set clean cache size to 0 to avoid creating fastcache,
+	// otherwise database must be closed when it's no longer needed to
+	// prevent memory leak.
+	CleanCacheSize: 0,
+}
+
 // Database is an intermediate write layer between the trie data structures and
 // the disk database. The aim is to accumulate trie writes in-memory and only
 // periodically flush a couple tries to disk, garbage collecting the remainder.
@@ -122,12 +136,13 @@ func (n *cachedNode) forChildren(resolver ChildResolver, onChild func(hash commo
 }
 
 // New initializes the hash-based node database.
-func New(diskdb ethdb.Database, size int, resolver ChildResolver) *Database {
+func New(diskdb ethdb.Database, config *Config, resolver ChildResolver) *Database {
-	// Initialize the clean cache if the specified cache allowance
+	if config == nil {
-	// is non-zero. Note, the size is in bytes.
+		config = Defaults
+	}
 	var cleans *fastcache.Cache
-	if size > 0 {
+	if config.CleanCacheSize > 0 {
-		cleans = fastcache.New(size)
+		cleans = fastcache.New(config.CleanCacheSize)
 	}
 	return &Database{
 		diskdb:   diskdb,
@@ -621,7 +636,13 @@ func (db *Database) Size() common.StorageSize {
 }
 
 // Close closes the trie database and releases all held resources.
-func (db *Database) Close() error { return nil }
+func (db *Database) Close() error {
+	if db.cleans != nil {
+		db.cleans.Reset()
+		db.cleans = nil
+	}
+	return nil
+}
 
 // Scheme returns the node scheme used in the database.
 func (db *Database) Scheme() string {

trie/triedb/pathdb/database.go

@@ -33,8 +33,26 @@ import (
 	"github.com/ethereum/go-ethereum/trie/triestate"
 )
 
-// maxDiffLayers is the maximum diff layers allowed in the layer tree.
+const (
-const maxDiffLayers = 128
+	// maxDiffLayers is the maximum diff layers allowed in the layer tree.
+	maxDiffLayers = 128
+
+	// defaultCleanSize is the default memory allowance of clean cache.
+	defaultCleanSize = 16 * 1024 * 1024
+
+	// maxBufferSize is the maximum memory allowance of node buffer.
+	// Too large nodebuffer will cause the system to pause for a long
+	// time when write happens. Also, the largest batch that pebble can
+	// support is 4GB, node will panic if batch size exceeds this limit.
+	maxBufferSize = 256 * 1024 * 1024
+
+	// DefaultBufferSize is the default memory allowance of node buffer
+	// that aggregates the writes from above until it's flushed into the
+	// disk. It's meant to be used once the initial sync is finished.
+	// Do not increase the buffer size arbitrarily, otherwise the system
+	// pause time will increase when the database writes happen.
+	DefaultBufferSize = 64 * 1024 * 1024
+)
 
 // layer is the interface implemented by all state layers which includes some
 // public methods and some additional methods for internal usage.
@@ -68,30 +86,33 @@ type layer interface {
 
 // Config contains the settings for database.
 type Config struct {
-	StateLimit uint64 // Number of recent blocks to maintain state history for
+	StateHistory   uint64 // Number of recent blocks to maintain state history for
-	CleanSize  int    // Maximum memory allowance (in bytes) for caching clean nodes
+	CleanCacheSize int    // Maximum memory allowance (in bytes) for caching clean nodes
-	DirtySize  int    // Maximum memory allowance (in bytes) for caching dirty nodes
+	DirtyCacheSize int    // Maximum memory allowance (in bytes) for caching dirty nodes
 	ReadOnly       bool   // Flag whether the database is opened in read only mode.
 }
 
-var (
+// sanitize checks the provided user configurations and changes anything that's
-	// defaultCleanSize is the default memory allowance of clean cache.
+// unreasonable or unworkable.
-	defaultCleanSize = 16 * 1024 * 1024
+func (c *Config) sanitize() *Config {
-
+	conf := *c
-	// defaultBufferSize is the default memory allowance of node buffer
+	if conf.DirtyCacheSize > maxBufferSize {
-	// that aggregates the writes from above until it's flushed into the
+		log.Warn("Sanitizing invalid node buffer size", "provided", common.StorageSize(conf.DirtyCacheSize), "updated", common.StorageSize(maxBufferSize))
-	// disk. Do not increase the buffer size arbitrarily, otherwise the
+		conf.DirtyCacheSize = maxBufferSize
-	// system pause time will increase when the database writes happen.
+	}
-	defaultBufferSize = 128 * 1024 * 1024
+	return &conf
-)
+}
 
 // Defaults contains default settings for Ethereum mainnet.
 var Defaults = &Config{
-	StateLimit: params.FullImmutabilityThreshold,
+	StateHistory:   params.FullImmutabilityThreshold,
-	CleanSize:  defaultCleanSize,
+	CleanCacheSize: defaultCleanSize,
-	DirtySize:  defaultBufferSize,
+	DirtyCacheSize: DefaultBufferSize,
 }
 
+// ReadOnly is the config in order to open database in read only mode.
+var ReadOnly = &Config{ReadOnly: true}
+
 // Database is a multiple-layered structure for maintaining in-memory trie nodes.
 // It consists of one persistent base layer backed by a key-value store, on top
 // of which arbitrarily many in-memory diff layers are stacked. The memory diffs
@@ -123,9 +144,11 @@ func New(diskdb ethdb.Database, config *Config) *Database {
 	if config == nil {
 		config = Defaults
 	}
+	config = config.sanitize()
+
 	db := &Database{
 		readOnly:   config.ReadOnly,
-		bufferSize: config.DirtySize,
+		bufferSize: config.DirtyCacheSize,
 		config:     config,
 		diskdb:     diskdb,
 	}
@@ -140,7 +163,7 @@ func New(diskdb ethdb.Database, config *Config) *Database {
 	// mechanism also ensures that at most one **non-readOnly** database
 	// is opened at the same time to prevent accidental mutation.
 	if ancient, err := diskdb.AncientDatadir(); err == nil && ancient != "" && !db.readOnly {
-		freezer, err := rawdb.NewStateHistoryFreezer(ancient, false)
+		freezer, err := rawdb.NewStateFreezer(ancient, false)
 		if err != nil {
 			log.Crit("Failed to open state history freezer", "err", err)
 		}
@@ -344,7 +367,14 @@ func (db *Database) Close() error {
 	db.lock.Lock()
 	defer db.lock.Unlock()
 
+	// Set the database to read-only mode to prevent all
+	// following mutations.
 	db.readOnly = true
+
+	// Release the memory held by clean cache.
+	db.tree.bottom().resetCache()
+
+	// Close the attached state history freezer.
 	if db.freezer == nil {
 		return nil
 	}
@@ -382,6 +412,10 @@ func (db *Database) SetBufferSize(size int) error {
 	db.lock.Lock()
 	defer db.lock.Unlock()
 
+	if size > maxBufferSize {
+		log.Info("Capped node buffer size", "provided", common.StorageSize(size), "adjusted", common.StorageSize(maxBufferSize))
+		size = maxBufferSize
+	}
 	db.bufferSize = size
 	return db.tree.bottom().setBufferSize(db.bufferSize)
 }

trie/triedb/pathdb/database_test.go

@@ -46,7 +46,8 @@ func updateTrie(addrHash common.Hash, root common.Hash, dirties, cleans map[comm
 			h.Update(key.Bytes(), val)
 		}
 	}
-	return h.Commit(false)
+	root, nodes, _ := h.Commit(false)
+	return root, nodes
 }
 
 func generateAccount(storageRoot common.Hash) types.StateAccount {
@@ -98,7 +99,7 @@ type tester struct {
 func newTester(t *testing.T) *tester {
 	var (
 		disk, _ = rawdb.NewDatabaseWithFreezer(rawdb.NewMemoryDatabase(), t.TempDir(), "", false)
-		db      = New(disk, &Config{CleanSize: 256 * 1024, DirtySize: 256 * 1024})
+		db      = New(disk, &Config{CleanCacheSize: 256 * 1024, DirtyCacheSize: 256 * 1024})
 		obj     = &tester{
 			db:           db,
 			preimages:    make(map[common.Hash]common.Address),

trie/triedb/pathdb/difflayer_test.go

@@ -29,7 +29,7 @@ import (
 func emptyLayer() *diskLayer {
 	return &diskLayer{
 		db:     New(rawdb.NewMemoryDatabase(), nil),
-		buffer: newNodeBuffer(defaultBufferSize, nil, 0),
+		buffer: newNodeBuffer(DefaultBufferSize, nil, 0),
 	}
 }
 

trie/triedb/pathdb/disklayer.go

@@ -47,8 +47,8 @@ func newDiskLayer(root common.Hash, id uint64, db *Database, cleans *fastcache.C
 	// Initialize a clean cache if the memory allowance is not zero
 	// or reuse the provided cache if it is not nil (inherited from
 	// the original disk layer).
-	if cleans == nil && db.config.CleanSize != 0 {
+	if cleans == nil && db.config.CleanCacheSize != 0 {
-		cleans = fastcache.New(db.config.CleanSize)
+		cleans = fastcache.New(db.config.CleanCacheSize)
 	}
 	return &diskLayer{
 		root:   root,
@@ -177,7 +177,7 @@ func (dl *diskLayer) commit(bottom *diffLayer, force bool) (*diskLayer, error) {
 	// corresponding states(journal), the stored state history will
 	// be truncated in the next restart.
 	if dl.db.freezer != nil {
-		err := writeHistory(dl.db.diskdb, dl.db.freezer, bottom, dl.db.config.StateLimit)
+		err := writeHistory(dl.db.diskdb, dl.db.freezer, bottom, dl.db.config.StateHistory)
 		if err != nil {
 			return nil, err
 		}
@@ -276,6 +276,20 @@ func (dl *diskLayer) size() common.StorageSize {
 	return common.StorageSize(dl.buffer.size)
 }
 
+// resetCache releases the memory held by clean cache to prevent memory leak.
+func (dl *diskLayer) resetCache() {
+	dl.lock.RLock()
+	defer dl.lock.RUnlock()
+
+	// Stale disk layer loses the ownership of clean cache.
+	if dl.stale {
+		return
+	}
+	if dl.cleans != nil {
+		dl.cleans.Reset()
+	}
+}
+
 // hasher is used to compute the sha256 hash of the provided data.
 type hasher struct{ sha crypto.KeccakState }
 

trie/triedb/pathdb/history_test.go

@@ -226,7 +226,7 @@ func TestTruncateTailHistories(t *testing.T) {
 
 // openFreezer initializes the freezer instance for storing state histories.
 func openFreezer(datadir string, readOnly bool) (*rawdb.ResettableFreezer, error) {
-	return rawdb.NewStateHistoryFreezer(datadir, readOnly)
+	return rawdb.NewStateFreezer(datadir, readOnly)
 }
 
 func compareSet[k comparable](a, b map[k][]byte) bool {

trie/triedb/pathdb/testutils.go

@@ -80,7 +80,7 @@ func (h *testHasher) Delete(key []byte) error {
 
 // Commit computes the new hash of the states and returns the set with all
 // state changes.
-func (h *testHasher) Commit(collectLeaf bool) (common.Hash, *trienode.NodeSet) {
+func (h *testHasher) Commit(collectLeaf bool) (common.Hash, *trienode.NodeSet, error) {
 	var (
 		nodes = make(map[common.Hash][]byte)
 		set   = trienode.NewNodeSet(h.owner)
@@ -108,7 +108,7 @@ func (h *testHasher) Commit(collectLeaf bool) (common.Hash, *trienode.NodeSet) {
 	if root == types.EmptyRootHash && h.root != types.EmptyRootHash {
 		set.AddNode(nil, trienode.NewDeleted())
 	}
-	return root, set
+	return root, set, nil
 }
 
 // hash performs the hash computation upon the provided states.

trie/triestate/state.go

@@ -43,7 +43,7 @@ type Trie interface {
 
 	// Commit the trie and returns a set of dirty nodes generated along with
 	// the new root hash.
-	Commit(collectLeaf bool) (common.Hash, *trienode.NodeSet)
+	Commit(collectLeaf bool) (common.Hash, *trienode.NodeSet, error)
 }
 
 // TrieLoader wraps functions to load tries.
@@ -129,7 +129,10 @@ func Apply(prevRoot common.Hash, postRoot common.Hash, accounts map[common.Addre
 			return nil, fmt.Errorf("failed to revert state, err: %w", err)
 		}
 	}
-	root, result := tr.Commit(false)
+	root, result, err := tr.Commit(false)
+	if err != nil {
+		return nil, err
+	}
 	if root != prevRoot {
 		return nil, fmt.Errorf("failed to revert state, want %#x, got %#x", prevRoot, root)
 	}
@@ -181,7 +184,10 @@ func updateAccount(ctx *context, loader TrieLoader, addr common.Address) error {
 			return err
 		}
 	}
-	root, result := st.Commit(false)
+	root, result, err := st.Commit(false)
+	if err != nil {
+		return err
+	}
 	if root != prev.Root {
 		return errors.New("failed to reset storage trie")
 	}
@@ -232,7 +238,10 @@ func deleteAccount(ctx *context, loader TrieLoader, addr common.Address) error {
 			return err
 		}
 	}
-	root, result := st.Commit(false)
+	root, result, err := st.Commit(false)
+	if err != nil {
+		return err
+	}
 	if root != types.EmptyRootHash {
 		return errors.New("failed to clear storage trie")
 	}