core, eth, ethdb, trie: simplify range proofs

core/rawdb/table.go

@@ -176,11 +176,6 @@ func (b *tableBatch) Delete(key []byte) error {
 	return b.batch.Delete(append([]byte(b.prefix), key...))
 }
 
-// KeyCount retrieves the number of keys queued up for writing.
-func (b *tableBatch) KeyCount() int {
-	return b.batch.KeyCount()
-}
-
 // ValueSize retrieves the amount of data queued up for writing.
 func (b *tableBatch) ValueSize() int {
 	return b.batch.ValueSize()

core/state/snapshot/generate.go

@@ -368,7 +368,7 @@ func (dl *diskLayer) proveRange(stats *generatorStats, root common.Hash, prefix
 	}
 	// Verify the snapshot segment with range prover, ensure that all flat states
 	// in this range correspond to merkle trie.
-	_, cont, err := trie.VerifyRangeProof(root, origin, last, keys, vals, proof)
+	cont, err := trie.VerifyRangeProof(root, origin, last, keys, vals, proof)
 	return &proofResult{
 			keys:     keys,
 			vals:     vals,

ethdb/batch.go

@@ -25,9 +25,6 @@ const IdealBatchSize = 100 * 1024
 type Batch interface {
 	KeyValueWriter
 
-	// KeyCount retrieves the number of keys queued up for writing.
-	KeyCount() int
-
 	// ValueSize retrieves the amount of data queued up for writing.
 	ValueSize() int
 
@@ -47,3 +44,28 @@ type Batcher interface {
 	// until a final write is called.
 	NewBatch() Batch
 }
+
+// HookedBatch wraps an arbitrary batch where each operation may be hooked into
+// to monitor from black box code.
+type HookedBatch struct {
+	Batch
+
+	OnPut    func(key []byte, value []byte) // Callback if a key is inserted
+	OnDelete func(key []byte)               // Callback if a key is deleted
+}
+
+// Put inserts the given value into the key-value data store.
+func (b HookedBatch) Put(key []byte, value []byte) error {
+	if b.OnPut != nil {
+		b.OnPut(key, value)
+	}
+	return b.Batch.Put(key, value)
+}
+
+// Delete removes the key from the key-value data store.
+func (b HookedBatch) Delete(key []byte) error {
+	if b.OnDelete != nil {
+		b.OnDelete(key)
+	}
+	return b.Batch.Delete(key)
+}

ethdb/leveldb/leveldb.go

@@ -448,7 +448,6 @@ func (db *Database) meter(refresh time.Duration) {
 type batch struct {
 	db   *leveldb.DB
 	b    *leveldb.Batch
-	keys int
 	size int
 }
 
@@ -462,16 +461,10 @@ func (b *batch) Put(key, value []byte) error {
 // Delete inserts the a key removal into the batch for later committing.
 func (b *batch) Delete(key []byte) error {
 	b.b.Delete(key)
-	b.keys++
 	b.size += len(key)
 	return nil
 }
 
-// KeyCount retrieves the number of keys queued up for writing.
-func (b *batch) KeyCount() int {
-	return b.keys
-}
-
 // ValueSize retrieves the amount of data queued up for writing.
 func (b *batch) ValueSize() int {
 	return b.size
@@ -485,7 +478,7 @@ func (b *batch) Write() error {
 // Reset resets the batch for reuse.
 func (b *batch) Reset() {
 	b.b.Reset()
-	b.keys, b.size = 0, 0
+	b.size = 0
 }
 
 // Replay replays the batch contents.

ethdb/memorydb/memorydb.go

@@ -198,7 +198,6 @@ type keyvalue struct {
 type batch struct {
 	db     *Database
 	writes []keyvalue
-	keys   int
 	size   int
 }
 
@@ -212,16 +211,10 @@ func (b *batch) Put(key, value []byte) error {
 // Delete inserts the a key removal into the batch for later committing.
 func (b *batch) Delete(key []byte) error {
 	b.writes = append(b.writes, keyvalue{common.CopyBytes(key), nil, true})
-	b.keys++
 	b.size += len(key)
 	return nil
 }
 
-// KeyCount retrieves the number of keys queued up for writing.
-func (b *batch) KeyCount() int {
-	return b.keys
-}
-
 // ValueSize retrieves the amount of data queued up for writing.
 func (b *batch) ValueSize() int {
 	return b.size
@@ -245,7 +238,7 @@ func (b *batch) Write() error {
 // Reset resets the batch for reuse.
 func (b *batch) Reset() {
 	b.writes = b.writes[:0]
-	b.keys, b.size = 0, 0
+	b.size = 0
 }
 
 // Replay replays the batch contents.

tests/fuzzers/rangeproof/rangeproof-fuzzer.go

@@ -170,18 +170,11 @@ func (f *fuzzer) fuzz() int {
 		}
 		ok = 1
 		//nodes, subtrie
-		nodes, hasMore, err := trie.VerifyRangeProof(tr.Hash(), first, last, keys, vals, proof)
+		hasMore, err := trie.VerifyRangeProof(tr.Hash(), first, last, keys, vals, proof)
 		if err != nil {
-			if nodes != nil {
-				panic("err != nil && nodes != nil")
-			}
 			if hasMore {
 				panic("err != nil && hasMore == true")
 			}
-		} else {
-			if nodes == nil {
-				panic("err == nil && nodes == nil")
-			}
 		}
 	}
 	return ok

tests/fuzzers/stacktrie/trie_fuzzer.go

@@ -90,7 +90,6 @@ func (b *spongeBatch) Put(key, value []byte) error {
 	return nil
 }
 func (b *spongeBatch) Delete(key []byte) error             { panic("implement me") }
-func (b *spongeBatch) KeyCount() int                       { panic("not implemented") }
 func (b *spongeBatch) ValueSize() int                      { return 100 }
 func (b *spongeBatch) Write() error                        { return nil }
 func (b *spongeBatch) Reset()                              {}

trie/notary.go

-// Copyright 2020 The go-ethereum Authors
-// This file is part of the go-ethereum library.
-//
-// The go-ethereum library is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Lesser General Public License as published by
-// the Free Software Foundation, either version 3 of the License, or
-// (at your option) any later version.
-//
-// The go-ethereum library is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-// GNU Lesser General Public License for more details.
-//
-// You should have received a copy of the GNU Lesser General Public License
-// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
-
-package trie
-
-import (
-	"github.com/ethereum/go-ethereum/ethdb"
-	"github.com/ethereum/go-ethereum/ethdb/memorydb"
-)
-
-// keyValueNotary tracks which keys have been accessed through a key-value reader
-// with te scope of verifying if certain proof datasets are maliciously bloated.
-type keyValueNotary struct {
-	ethdb.KeyValueReader
-	reads map[string]struct{}
-}
-
-// newKeyValueNotary wraps a key-value database with an access notary to track
-// which items have bene accessed.
-func newKeyValueNotary(db ethdb.KeyValueReader) *keyValueNotary {
-	return &keyValueNotary{
-		KeyValueReader: db,
-		reads:          make(map[string]struct{}),
-	}
-}
-
-// Get retrieves an item from the underlying database, but also tracks it as an
-// accessed slot for bloat checks.
-func (k *keyValueNotary) Get(key []byte) ([]byte, error) {
-	k.reads[string(key)] = struct{}{}
-	return k.KeyValueReader.Get(key)
-}
-
-// Accessed returns s snapshot of the original key-value store containing only the
-// data accessed through the notary.
-func (k *keyValueNotary) Accessed() ethdb.KeyValueStore {
-	db := memorydb.New()
-	for keystr := range k.reads {
-		key := []byte(keystr)
-		val, _ := k.KeyValueReader.Get(key)
-		db.Put(key, val)
-	}
-	return db
-}

trie/proof.go

@@ -464,108 +464,91 @@ func hasRightElement(node node, key []byte) bool {
 //
 // Except returning the error to indicate the proof is valid or not, the function will
 // also return a flag to indicate whether there exists more accounts/slots in the trie.
-func VerifyRangeProof(rootHash common.Hash, firstKey []byte, lastKey []byte, keys [][]byte, values [][]byte, proof ethdb.KeyValueReader) (ethdb.KeyValueStore, bool, error) {
+//
+// Note: This method does not verify that the proof is of minimal form. If the input
+// proofs are 'bloated' with neighbour leaves or random data, aside from the 'useful'
+// data, then the proof will still be accepted.
+func VerifyRangeProof(rootHash common.Hash, firstKey []byte, lastKey []byte, keys [][]byte, values [][]byte, proof ethdb.KeyValueReader) (bool, error) {
 	if len(keys) != len(values) {
-		return nil, false, fmt.Errorf("inconsistent proof data, keys: %d, values: %d", len(keys), len(values))
+		return false, fmt.Errorf("inconsistent proof data, keys: %d, values: %d", len(keys), len(values))
 	}
 	// Ensure the received batch is monotonic increasing.
 	for i := 0; i < len(keys)-1; i++ {
 		if bytes.Compare(keys[i], keys[i+1]) >= 0 {
-			return nil, false, errors.New("range is not monotonically increasing")
+			return false, errors.New("range is not monotonically increasing")
 		}
 	}
-	// Create a key-value notary to track which items from the given proof the
-	// range prover actually needed to verify the data
-	notary := newKeyValueNotary(proof)
-
 	// Special case, there is no edge proof at all. The given range is expected
 	// to be the whole leaf-set in the trie.
 	if proof == nil {
-		var (
-			diskdb = memorydb.New()
-			tr     = NewStackTrie(diskdb)
-		)
+		tr := NewStackTrie(nil)
 		for index, key := range keys {
 			tr.TryUpdate(key, values[index])
 		}
 		if have, want := tr.Hash(), rootHash; have != want {
-			return nil, false, fmt.Errorf("invalid proof, want hash %x, got %x", want, have)
-		}
-		// Proof seems valid, serialize remaining nodes into the database
-		if _, err := tr.Commit(); err != nil {
-			return nil, false, err
+			return false, fmt.Errorf("invalid proof, want hash %x, got %x", want, have)
 		}
-		return diskdb, false, nil // No more elements
+		return false, nil // No more elements
 	}
 	// Special case, there is a provided edge proof but zero key/value
 	// pairs, ensure there are no more accounts / slots in the trie.
 	if len(keys) == 0 {
-		root, val, err := proofToPath(rootHash, nil, firstKey, notary, true)
+		root, val, err := proofToPath(rootHash, nil, firstKey, proof, true)
 		if err != nil {
-			return nil, false, err
+			return false, err
 		}
 		if val != nil || hasRightElement(root, firstKey) {
-			return nil, false, errors.New("more entries available")
+			return false, errors.New("more entries available")
 		}
-		// Since the entire proof is a single path, we can construct a trie and a
-		// node database directly out of the inputs, no need to generate them
-		diskdb := notary.Accessed()
-		return diskdb, hasRightElement(root, firstKey), nil
+		return hasRightElement(root, firstKey), nil
 	}
 	// Special case, there is only one element and two edge keys are same.
 	// In this case, we can't construct two edge paths. So handle it here.
 	if len(keys) == 1 && bytes.Equal(firstKey, lastKey) {
-		root, val, err := proofToPath(rootHash, nil, firstKey, notary, false)
+		root, val, err := proofToPath(rootHash, nil, firstKey, proof, false)
 		if err != nil {
-			return nil, false, err
+			return false, err
 		}
 		if !bytes.Equal(firstKey, keys[0]) {
-			return nil, false, errors.New("correct proof but invalid key")
+			return false, errors.New("correct proof but invalid key")
 		}
 		if !bytes.Equal(val, values[0]) {
-			return nil, false, errors.New("correct proof but invalid data")
+			return false, errors.New("correct proof but invalid data")
 		}
-		// Since the entire proof is a single path, we can construct a trie and a
-		// node database directly out of the inputs, no need to generate them
-		diskdb := notary.Accessed()
-		return diskdb, hasRightElement(root, firstKey), nil
+		return hasRightElement(root, firstKey), nil
 	}
 	// Ok, in all other cases, we require two edge paths available.
 	// First check the validity of edge keys.
 	if bytes.Compare(firstKey, lastKey) >= 0 {
-		return nil, false, errors.New("invalid edge keys")
+		return false, errors.New("invalid edge keys")
 	}
 	// todo(rjl493456442) different length edge keys should be supported
 	if len(firstKey) != len(lastKey) {
-		return nil, false, errors.New("inconsistent edge keys")
+		return false, errors.New("inconsistent edge keys")
 	}
 	// Convert the edge proofs to edge trie paths. Then we can
 	// have the same tree architecture with the original one.
 	// For the first edge proof, non-existent proof is allowed.
-	root, _, err := proofToPath(rootHash, nil, firstKey, notary, true)
+	root, _, err := proofToPath(rootHash, nil, firstKey, proof, true)
 	if err != nil {
-		return nil, false, err
+		return false, err
 	}
 	// Pass the root node here, the second path will be merged
 	// with the first one. For the last edge proof, non-existent
 	// proof is also allowed.
-	root, _, err = proofToPath(rootHash, root, lastKey, notary, true)
+	root, _, err = proofToPath(rootHash, root, lastKey, proof, true)
 	if err != nil {
-		return nil, false, err
+		return false, err
 	}
 	// Remove all internal references. All the removed parts should
 	// be re-filled(or re-constructed) by the given leaves range.
 	empty, err := unsetInternal(root, firstKey, lastKey)
 	if err != nil {
-		return nil, false, err
+		return false, err
 	}
 	// Rebuild the trie with the leaf stream, the shape of trie
 	// should be same with the original one.
-	var (
-		diskdb = memorydb.New()
-		triedb = NewDatabase(diskdb)
-	)
-	tr := &Trie{root: root, db: triedb}
+	tr := &Trie{root: root, db: NewDatabase(memorydb.New())}
 	if empty {
 		tr.root = nil
 	}
@@ -573,16 +556,9 @@ func VerifyRangeProof(rootHash common.Hash, firstKey []byte, lastKey []byte, key
 		tr.TryUpdate(key, values[index])
 	}
 	if tr.Hash() != rootHash {
-		return nil, false, fmt.Errorf("invalid proof, want hash %x, got %x", rootHash, tr.Hash())
-	}
-	// Proof seems valid, serialize all the nodes into the database
-	if _, err := tr.Commit(nil); err != nil {
-		return nil, false, err
-	}
-	if err := triedb.Commit(rootHash, false, nil); err != nil {
-		return nil, false, err
+		return false, fmt.Errorf("invalid proof, want hash %x, got %x", rootHash, tr.Hash())
 	}
-	return diskdb, hasRightElement(root, keys[len(keys)-1]), nil
+	return hasRightElement(root, keys[len(keys)-1]), nil
 }
 
 // get returns the child of the given node. Return nil if the

trie/trie_test.go

@@ -706,7 +706,6 @@ func (b *spongeBatch) Put(key, value []byte) error {
 	return nil
 }
 func (b *spongeBatch) Delete(key []byte) error             { panic("implement me") }
-func (b *spongeBatch) KeyCount() int                       { return 100 }
 func (b *spongeBatch) ValueSize() int                      { return 100 }
 func (b *spongeBatch) Write() error                        { return nil }
 func (b *spongeBatch) Reset()                              {}