core/state, light, les: make signature of ContractCode hash-independe…

…nt (ethereum#27209)

* core/state, light, les: make signature of ContractCode hash-independent

* push current state for feedback

* les: fix unit test

* core, les, light: fix les unittests

* core/state, trie, les, light: fix state iterator

* core, les: address comments

* les: fix lint

---------

Co-authored-by: Gary Rong <garyrong0905@gmail.com>

core/blockchain_reader.go

@@ -306,9 +306,11 @@ func (bc *BlockChain) TrieNode(hash common.Hash) ([]byte, error) {
 // new code scheme.
 func (bc *BlockChain) ContractCodeWithPrefix(hash common.Hash) ([]byte, error) {
 	type codeReader interface {
-		ContractCodeWithPrefix(addrHash, codeHash common.Hash) ([]byte, error)
+		ContractCodeWithPrefix(address common.Address, codeHash common.Hash) ([]byte, error)
 	}
-	return bc.stateCache.(codeReader).ContractCodeWithPrefix(common.Hash{}, hash)
+	// TODO(rjl493456442) The associated account address is also required
+	// in Verkle scheme. Fix it once snap-sync is supported for Verkle.
+	return bc.stateCache.(codeReader).ContractCodeWithPrefix(common.Address{}, hash)
 }
 
 // State returns a new mutable state based on the current HEAD block.

core/state/database.go

@@ -24,6 +24,7 @@ import (
 	"github.com/ethereum/go-ethereum/common/lru"
 	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/types"
+	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/trie"
 	"github.com/ethereum/go-ethereum/trie/trienode"
@@ -43,16 +44,16 @@ type Database interface {
 	OpenTrie(root common.Hash) (Trie, error)
 
 	// OpenStorageTrie opens the storage trie of an account.
-	OpenStorageTrie(stateRoot common.Hash, addrHash, root common.Hash) (Trie, error)
+	OpenStorageTrie(stateRoot common.Hash, address common.Address, root common.Hash) (Trie, error)
 
 	// CopyTrie returns an independent copy of the given trie.
 	CopyTrie(Trie) Trie
 
 	// ContractCode retrieves a particular contract's code.
-	ContractCode(addrHash, codeHash common.Hash) ([]byte, error)
+	ContractCode(addr common.Address, codeHash common.Hash) ([]byte, error)
 
 	// ContractCodeSize retrieves a particular contracts code's size.
-	ContractCodeSize(addrHash, codeHash common.Hash) (int, error)
+	ContractCodeSize(addr common.Address, codeHash common.Hash) (int, error)
 
 	// DiskDB returns the underlying key-value disk database.
 	DiskDB() ethdb.KeyValueStore
@@ -177,8 +178,8 @@ func (db *cachingDB) OpenTrie(root common.Hash) (Trie, error) {
 }
 
 // OpenStorageTrie opens the storage trie of an account.
-func (db *cachingDB) OpenStorageTrie(stateRoot common.Hash, addrHash, root common.Hash) (Trie, error) {
-	tr, err := trie.NewStateTrie(trie.StorageTrieID(stateRoot, addrHash, root), db.triedb)
+func (db *cachingDB) OpenStorageTrie(stateRoot common.Hash, address common.Address, root common.Hash) (Trie, error) {
+	tr, err := trie.NewStateTrie(trie.StorageTrieID(stateRoot, crypto.Keccak256Hash(address.Bytes()), root), db.triedb)
 	if err != nil {
 		return nil, err
 	}
@@ -196,7 +197,7 @@ func (db *cachingDB) CopyTrie(t Trie) Trie {
 }
 
 // ContractCode retrieves a particular contract's code.
-func (db *cachingDB) ContractCode(addrHash, codeHash common.Hash) ([]byte, error) {
+func (db *cachingDB) ContractCode(address common.Address, codeHash common.Hash) ([]byte, error) {
 	code, _ := db.codeCache.Get(codeHash)
 	if len(code) > 0 {
 		return code, nil
@@ -213,7 +214,7 @@ func (db *cachingDB) ContractCode(addrHash, codeHash common.Hash) ([]byte, error
 // ContractCodeWithPrefix retrieves a particular contract's code. If the
 // code can't be found in the cache, then check the existence with **new**
 // db scheme.
-func (db *cachingDB) ContractCodeWithPrefix(addrHash, codeHash common.Hash) ([]byte, error) {
+func (db *cachingDB) ContractCodeWithPrefix(address common.Address, codeHash common.Hash) ([]byte, error) {
 	code, _ := db.codeCache.Get(codeHash)
 	if len(code) > 0 {
 		return code, nil
@@ -228,11 +229,11 @@ func (db *cachingDB) ContractCodeWithPrefix(addrHash, codeHash common.Hash) ([]b
 }
 
 // ContractCodeSize retrieves a particular contracts code's size.
-func (db *cachingDB) ContractCodeSize(addrHash, codeHash common.Hash) (int, error) {
+func (db *cachingDB) ContractCodeSize(addr common.Address, codeHash common.Hash) (int, error) {
 	if cached, ok := db.codeSizeCache.Get(codeHash); ok {
 		return cached, nil
 	}
-	code, err := db.ContractCode(addrHash, codeHash)
+	code, err := db.ContractCode(addr, codeHash)
 	return len(code), err
 }
 

core/state/iterator.go

@@ -18,6 +18,7 @@ package state
 
 import (
 	"bytes"
+	"errors"
 	"fmt"
 
 	"github.com/ethereum/go-ethereum/common"
@@ -27,7 +28,8 @@ import (
 )
 
 // nodeIterator is an iterator to traverse the entire state trie post-order,
-// including all of the contract code and contract state tries.
+// including all of the contract code and contract state tries. Preimage is
+// required in order to resolve the contract address.
 type nodeIterator struct {
 	state *StateDB // State being iterated
 
@@ -113,7 +115,15 @@ func (it *nodeIterator) step() error {
 	if err := rlp.Decode(bytes.NewReader(it.stateIt.LeafBlob()), &account); err != nil {
 		return err
 	}
-	dataTrie, err := it.state.db.OpenStorageTrie(it.state.originalRoot, common.BytesToHash(it.stateIt.LeafKey()), account.Root)
+	// Lookup the preimage of account hash
+	preimage := it.state.trie.GetKey(it.stateIt.LeafKey())
+	if preimage == nil {
+		return errors.New("account address is not available")
+	}
+	address := common.BytesToAddress(preimage)
+
+	// Traverse the storage slots belong to the account
+	dataTrie, err := it.state.db.OpenStorageTrie(it.state.originalRoot, address, account.Root)
 	if err != nil {
 		return err
 	}
@@ -126,8 +136,7 @@ func (it *nodeIterator) step() error {
 	}
 	if !bytes.Equal(account.CodeHash, types.EmptyCodeHash.Bytes()) {
 		it.codeHash = common.BytesToHash(account.CodeHash)
-		addrHash := common.BytesToHash(it.stateIt.LeafKey())
-		it.code, err = it.state.db.ContractCode(addrHash, common.BytesToHash(account.CodeHash))
+		it.code, err = it.state.db.ContractCode(address, common.BytesToHash(account.CodeHash))
 		if err != nil {
 			return fmt.Errorf("code %x: %v", account.CodeHash, err)
 		}

core/state/state_object.go

@@ -142,7 +142,7 @@ func (s *stateObject) getTrie(db Database) (Trie, error) {
 			s.trie = s.db.prefetcher.trie(s.addrHash, s.data.Root)
 		}
 		if s.trie == nil {
-			tr, err := db.OpenStorageTrie(s.db.originalRoot, s.addrHash, s.data.Root)
+			tr, err := db.OpenStorageTrie(s.db.originalRoot, s.address, s.data.Root)
 			if err != nil {
 				return nil, err
 			}
@@ -441,7 +441,7 @@ func (s *stateObject) Code(db Database) []byte {
 	if bytes.Equal(s.CodeHash(), types.EmptyCodeHash.Bytes()) {
 		return nil
 	}
-	code, err := db.ContractCode(s.addrHash, common.BytesToHash(s.CodeHash()))
+	code, err := db.ContractCode(s.address, common.BytesToHash(s.CodeHash()))
 	if err != nil {
 		s.db.setError(fmt.Errorf("can't load code hash %x: %v", s.CodeHash(), err))
 	}
@@ -459,7 +459,7 @@ func (s *stateObject) CodeSize(db Database) int {
 	if bytes.Equal(s.CodeHash(), types.EmptyCodeHash.Bytes()) {
 		return 0
 	}
-	size, err := db.ContractCodeSize(s.addrHash, common.BytesToHash(s.CodeHash()))
+	size, err := db.ContractCodeSize(s.address, common.BytesToHash(s.CodeHash()))
 	if err != nil {
 		s.db.setError(fmt.Errorf("can't load code size %x: %v", s.CodeHash(), err))
 	}

core/state/sync_test.go

@@ -42,7 +42,7 @@ type testAccount struct {
 func makeTestState() (ethdb.Database, Database, common.Hash, []*testAccount) {
 	// Create an empty state
 	db := rawdb.NewMemoryDatabase()
-	sdb := NewDatabase(db)
+	sdb := NewDatabaseWithConfig(db, &trie.Config{Preimages: true})
 	state, _ := New(types.EmptyRootHash, sdb, nil)
 
 	// Fill it with some arbitrary data
@@ -100,28 +100,9 @@ func checkStateAccounts(t *testing.T, db ethdb.Database, root common.Hash, accou
 	}
 }
 
-// checkTrieConsistency checks that all nodes in a (sub-)trie are indeed present.
-func checkTrieConsistency(db ethdb.Database, root common.Hash) error {
-	if v, _ := db.Get(root[:]); v == nil {
-		return nil // Consider a non existent state consistent.
-	}
-	trie, err := trie.New(trie.StateTrieID(root), trie.NewDatabase(db))
-	if err != nil {
-		return err
-	}
-	it := trie.MustNodeIterator(nil)
-	for it.Next(true) {
-	}
-	return it.Error()
-}
-
 // checkStateConsistency checks that all data of a state root is present.
 func checkStateConsistency(db ethdb.Database, root common.Hash) error {
-	// Create and iterate a state trie rooted in a sub-node
-	if _, err := db.Get(root.Bytes()); err != nil {
-		return nil // Consider a non existent state consistent.
-	}
-	state, err := New(root, NewDatabase(db), nil)
+	state, err := New(root, NewDatabaseWithConfig(db, &trie.Config{Preimages: true}), nil)
 	if err != nil {
 		return err
 	}
@@ -171,7 +152,7 @@ type stateElement struct {
 
 func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 	// Create a random state to copy
-	_, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
 	if commit {
 		srcDb.TrieDB().Commit(srcRoot, false)
 	}
@@ -204,7 +185,7 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 			codeResults = make([]trie.CodeSyncResult, len(codeElements))
 		)
 		for i, element := range codeElements {
-			data, err := srcDb.ContractCode(common.Hash{}, element.code)
+			data, err := srcDb.ContractCode(common.Address{}, element.code)
 			if err != nil {
 				t.Fatalf("failed to retrieve contract bytecode for hash %x", element.code)
 			}
@@ -274,6 +255,10 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 			})
 		}
 	}
+	// Copy the preimages from source db in order to traverse the state.
+	srcDb.TrieDB().WritePreimages()
+	copyPreimages(srcDisk, dstDb)
+
 	// Cross check that the two states are in sync
 	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
 }
@@ -282,7 +267,7 @@ func testIterativeStateSync(t *testing.T, count int, commit bool, bypath bool) {
 // partial results are returned, and the others sent only later.
 func TestIterativeDelayedStateSync(t *testing.T) {
 	// Create a random state to copy
-	_, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
@@ -312,7 +297,7 @@ func TestIterativeDelayedStateSync(t *testing.T) {
 		if len(codeElements) > 0 {
 			codeResults := make([]trie.CodeSyncResult, len(codeElements)/2+1)
 			for i, element := range codeElements[:len(codeResults)] {
-				data, err := srcDb.ContractCode(common.Hash{}, element.code)
+				data, err := srcDb.ContractCode(common.Address{}, element.code)
 				if err != nil {
 					t.Fatalf("failed to retrieve contract bytecode for %x", element.code)
 				}
@@ -363,6 +348,10 @@ func TestIterativeDelayedStateSync(t *testing.T) {
 			})
 		}
 	}
+	// Copy the preimages from source db in order to traverse the state.
+	srcDb.TrieDB().WritePreimages()
+	copyPreimages(srcDisk, dstDb)
+
 	// Cross check that the two states are in sync
 	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
 }
@@ -375,7 +364,7 @@ func TestIterativeRandomStateSyncBatched(t *testing.T)    { testIterativeRandomS
 
 func testIterativeRandomStateSync(t *testing.T, count int) {
 	// Create a random state to copy
-	_, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
@@ -399,7 +388,7 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 		if len(codeQueue) > 0 {
 			results := make([]trie.CodeSyncResult, 0, len(codeQueue))
 			for hash := range codeQueue {
-				data, err := srcDb.ContractCode(common.Hash{}, hash)
+				data, err := srcDb.ContractCode(common.Address{}, hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for %x", hash)
 				}
@@ -447,6 +436,10 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 			codeQueue[hash] = struct{}{}
 		}
 	}
+	// Copy the preimages from source db in order to traverse the state.
+	srcDb.TrieDB().WritePreimages()
+	copyPreimages(srcDisk, dstDb)
+
 	// Cross check that the two states are in sync
 	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
 }
@@ -455,7 +448,7 @@ func testIterativeRandomStateSync(t *testing.T, count int) {
 // partial results are returned (Even those randomly), others sent only later.
 func TestIterativeRandomDelayedStateSync(t *testing.T) {
 	// Create a random state to copy
-	_, srcDb, srcRoot, srcAccounts := makeTestState()
+	srcDisk, srcDb, srcRoot, srcAccounts := makeTestState()
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
@@ -481,7 +474,7 @@ func TestIterativeRandomDelayedStateSync(t *testing.T) {
 			for hash := range codeQueue {
 				delete(codeQueue, hash)
 
-				data, err := srcDb.ContractCode(common.Hash{}, hash)
+				data, err := srcDb.ContractCode(common.Address{}, hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for %x", hash)
 				}
@@ -537,6 +530,10 @@ func TestIterativeRandomDelayedStateSync(t *testing.T) {
 			codeQueue[hash] = struct{}{}
 		}
 	}
+	// Copy the preimages from source db in order to traverse the state.
+	srcDb.TrieDB().WritePreimages()
+	copyPreimages(srcDisk, dstDb)
+
 	// Cross check that the two states are in sync
 	checkStateAccounts(t, dstDb, srcRoot, srcAccounts)
 }
@@ -555,7 +552,6 @@ func TestIncompleteStateSync(t *testing.T) {
 		}
 	}
 	isCode[types.EmptyCodeHash] = struct{}{}
-	checkTrieConsistency(db, srcRoot)
 
 	// Create a destination state and sync with the scheduler
 	dstDb := rawdb.NewMemoryDatabase()
@@ -588,7 +584,7 @@ func TestIncompleteStateSync(t *testing.T) {
 		if len(codeQueue) > 0 {
 			results := make([]trie.CodeSyncResult, 0, len(codeQueue))
 			for hash := range codeQueue {
-				data, err := srcDb.ContractCode(common.Hash{}, hash)
+				data, err := srcDb.ContractCode(common.Address{}, hash)
 				if err != nil {
 					t.Fatalf("failed to retrieve node data for %x", hash)
 				}
@@ -602,7 +598,6 @@ func TestIncompleteStateSync(t *testing.T) {
 				}
 			}
 		}
-		var nodehashes []common.Hash
 		if len(nodeQueue) > 0 {
 			results := make([]trie.NodeSyncResult, 0, len(nodeQueue))
 			for path, element := range nodeQueue {
@@ -617,7 +612,6 @@ func TestIncompleteStateSync(t *testing.T) {
 					addedPaths = append(addedPaths, element.path)
 					addedHashes = append(addedHashes, element.hash)
 				}
-				nodehashes = append(nodehashes, element.hash)
 			}
 			// Process each of the state nodes
 			for _, result := range results {
@@ -632,13 +626,6 @@ func TestIncompleteStateSync(t *testing.T) {
 		}
 		batch.Write()
 
-		for _, root := range nodehashes {
-			// Can't use checkStateConsistency here because subtrie keys may have odd
-			// length and crash in LeafKey.
-			if err := checkTrieConsistency(dstDb, root); err != nil {
-				t.Fatalf("state inconsistent: %v", err)
-			}
-		}
 		// Fetch the next batch to retrieve
 		nodeQueue = make(map[string]stateElement)
 		codeQueue = make(map[common.Hash]struct{})
@@ -654,6 +641,10 @@ func TestIncompleteStateSync(t *testing.T) {
 			codeQueue[hash] = struct{}{}
 		}
 	}
+	// Copy the preimages from source db in order to traverse the state.
+	srcDb.TrieDB().WritePreimages()
+	copyPreimages(db, dstDb)
+
 	// Sanity check that removing any node from the database is detected
 	for _, node := range addedCodes {
 		val := rawdb.ReadCode(dstDb, node)
@@ -678,3 +669,15 @@ func TestIncompleteStateSync(t *testing.T) {
 		rawdb.WriteTrieNode(dstDb, owner, inner, hash, val, scheme)
 	}
 }
+
+func copyPreimages(srcDb, dstDb ethdb.Database) {
+	it := srcDb.NewIterator(rawdb.PreimagePrefix, nil)
+	defer it.Release()
+
+	preimages := make(map[common.Hash][]byte)
+	for it.Next() {
+		hash := it.Key()[len(rawdb.PreimagePrefix):]
+		preimages[common.BytesToHash(hash)] = common.CopyBytes(it.Value())
+	}
+	rawdb.WritePreimages(dstDb, preimages)
+}

core/state/trie_prefetcher.go

@@ -302,7 +302,7 @@ func (sf *subfetcher) loop() {
 		}
 		sf.trie = trie
 	} else {
-		trie, err := sf.db.OpenStorageTrie(sf.state, sf.owner, sf.root)
+		trie, err := sf.db.OpenStorageTrie(sf.state, sf.addr, sf.root)
 		if err != nil {
 			log.Warn("Trie prefetcher failed opening trie", "root", sf.root, "err", err)
 			return