cmd, core: resolve scheme from a read-write database (#28313)

* cmd, core: resolve scheme from a read-write database

* cmd, core, eth: move the scheme check in the ethereum constructor

* cmd/geth: dump should in ro mode

* cmd: reverts

cmd/geth/chaincmd.go

@@ -655,7 +655,7 @@ func dump(ctx *cli.Context) error {
 	if err != nil {
 		return err
 	}
-	triedb := utils.MakeTrieDatabase(ctx, db, true, false) // always enable preimage lookup
+	triedb := utils.MakeTrieDatabase(ctx, db, true, true) // always enable preimage lookup
 	defer triedb.Close()
 
 	state, err := state.New(root, state.NewDatabaseWithNodeDB(db, triedb), nil)

cmd/utils/flags.go

@@ -1884,15 +1884,9 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *ethconfig.Config) {
 	if ctx.IsSet(StateHistoryFlag.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)
+	if ctx.IsSet(StateSchemeFlag.Name) {
+		cfg.StateScheme = ctx.String(StateSchemeFlag.Name)
 	}
-	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 {
@@ -2355,7 +2349,7 @@ 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)
+	scheme, err := rawdb.ParseStateScheme(ctx.String(StateSchemeFlag.Name), chainDb)
 	if err != nil {
 		Fatalf("%v", err)
 	}
@@ -2418,52 +2412,12 @@ func MakeConsolePreloads(ctx *cli.Context) []string {
 	return preloads
 }
 
-// ParseStateScheme checks if the specified state scheme is compatible with
-// the stored state.
-//
-//   - If the provided scheme is none, use the scheme consistent with persistent
-//     state, or fallback to hash-based scheme if state is empty.
-//
-//   - If the provided scheme is hash, use hash-based scheme or error out if not
-//     compatible with persistent state scheme.
-//
-//   - If the provided scheme is 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.
-	provided, err := compareCLIWithConfig(ctx)
-	if err != nil {
-		log.Error("failed to compare CLI with config", "error", err)
-		return "", err
-	}
-
-	stored := rawdb.ReadStateScheme(disk)
-	if provided == "" {
-		if stored == "" {
-			// use default scheme for empty database, flip it when
-			// path mode is chosen as default
-			log.Info("State scheme set to default", "scheme", rawdb.HashScheme)
-			return rawdb.HashScheme, nil
-		}
-		log.Info("State scheme set to already existing disk db", "scheme", stored)
-		return stored, nil // reuse scheme of persistent scheme
-	}
-	// If state scheme is specified, ensure it's compatible with persistent state.
-	if stored == "" || provided == stored {
-		log.Info("State scheme set by user", "scheme", provided)
-		return provided, nil
-	}
-	return "", fmt.Errorf("incompatible state scheme, db stored: %s, user provided: %s", stored, provided)
-}
-
 // 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)
+	scheme, err := rawdb.ParseStateScheme(ctx.String(StateSchemeFlag.Name), disk)
 	if err != nil {
 		Fatalf("%v", err)
 	}

core/genesis.go

@@ -120,8 +120,8 @@ func (ga *GenesisAlloc) UnmarshalJSON(data []byte) error {
 	return nil
 }
 
-// deriveHash computes the state root according to the genesis specification.
-func (ga *GenesisAlloc) deriveHash() (common.Hash, error) {
+// hash computes the state root according to the genesis specification.
+func (ga *GenesisAlloc) hash() (common.Hash, error) {
 	// Create an ephemeral in-memory database for computing hash,
 	// all the derived states will be discarded to not pollute disk.
 	db := state.NewDatabase(rawdb.NewMemoryDatabase())
@@ -142,9 +142,9 @@ func (ga *GenesisAlloc) deriveHash() (common.Hash, error) {
 	return root, err
 }
 
-// flush is very similar with deriveHash, but the main difference is
-// all the generated states will be persisted into the given database.
-// Also, the genesis state specification will be flushed as well.
+// flush is very similar with hash, but the main difference is all the generated
+// states will be persisted into the given database. Also, the genesis state
+// specification will be flushed as well.
 func (ga *GenesisAlloc) flush(db ethdb.Database, triedb *trie.Database, blockhash common.Hash) error {
 	trieConfig := triedb.Config()
 	if trieConfig != nil {
@@ -455,7 +455,7 @@ func (g *Genesis) configOrDefault(ghash common.Hash) *params.ChainConfig {
 
 // ToBlock returns the genesis block according to genesis specification.
 func (g *Genesis) ToBlock() *types.Block {
-	root, err := g.Alloc.deriveHash()
+	root, err := g.Alloc.hash()
 	if err != nil {
 		panic(err)
 	}

core/genesis_test.go

@@ -209,7 +209,7 @@ func TestReadWriteGenesisAlloc(t *testing.T) {
 			{1}: {Balance: big.NewInt(1), Storage: map[common.Hash]common.Hash{{1}: {1}}},
 			{2}: {Balance: big.NewInt(2), Storage: map[common.Hash]common.Hash{{2}: {2}}},
 		}
-		hash, _ = alloc.deriveHash()
+		hash, _ = alloc.hash()
 	)
 	blob, _ := json.Marshal(alloc)
 	rawdb.WriteGenesisStateSpec(db, hash, blob)

core/rawdb/accessors_trie.go

@@ -314,3 +314,38 @@ func ValidateStateScheme(stateScheme string) bool {
 	}
 	return false
 }
+
+// ParseStateScheme checks if the specified state scheme is compatible with
+// the stored state.
+//
+//   - If the provided scheme is none, use the scheme consistent with persistent
+//     state, or fallback to hash-based scheme if state is empty.
+//
+//   - If the provided scheme is hash, use hash-based scheme or error out if not
+//     compatible with persistent state scheme.
+//
+//   - If the provided scheme is path: use path-based scheme or error out if not
+//     compatible with persistent state scheme.
+func ParseStateScheme(provided string, 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 := ReadStateScheme(disk)
+	if provided == "" {
+		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 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.
+	if stored == "" || provided == stored {
+		log.Info("State scheme set by user", "scheme", provided)
+		return provided, nil
+	}
+	return "", fmt.Errorf("incompatible state scheme, stored: %s, provided: %s", stored, provided)
+}

eth/backend.go

@@ -156,7 +156,12 @@ func New(stack *node.Node, config *ethconfig.Config) (*Ethereum, error) {
 	if err != nil {
 		return nil, err
 	}
-	if config.StateScheme == rawdb.HashScheme {
+	scheme, err := rawdb.ParseStateScheme(config.StateScheme, chainDb)
+	if err != nil {
+		return nil, err
+	}
+	// Try to recover offline state pruning only in hash-based.
+	if scheme == rawdb.HashScheme {
 		if err := pruner.RecoverPruning(stack.ResolvePath(""), chainDb, config.TriesInMemory); err != nil {
 			log.Error("Failed to recover state", "error", err)
 		}
@@ -242,8 +247,8 @@ func New(stack *node.Node, config *ethconfig.Config) (*Ethereum, error) {
 			TriesInMemory:       config.TriesInMemory,
 			Preimages:           config.Preimages,
 			StateHistory:        config.StateHistory,
-			StateScheme:         config.StateScheme,
 			PathSyncFlush:       config.PathSyncFlush,
+			StateScheme:         scheme,
 		}
 	)
 	bcOps := make([]core.BlockChainOption, 0)

eth/ethconfig/config.go

@@ -112,9 +112,13 @@ type Config struct {
 	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
 	PathSyncFlush      bool   `toml:",omitempty"` // State scheme used to store ethereum state and merkle trie nodes on top
 
+	// State scheme represents the scheme used to store ethereum states and trie
+	// nodes on top. It can be 'hash', 'path', or none which means use the scheme
+	// consistent with persistent state.
+	StateScheme string `toml:",omitempty"`
+
 	// 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
 	// presence of these blocks for every new peer connection.