Merge pull request #2081 from Roasbeef/dont-serialize-tx-for-txhash

btcutil: reuse serialized tx during TxHash

btcutil/bench_test.go

@@ -0,0 +1,80 @@
+package btcutil_test
+
+import (
+	"testing"
+
+	"github.com/btcsuite/btcd/btcutil"
+	"github.com/btcsuite/btcd/chaincfg/chainhash"
+)
+
+var (
+	bencHash *chainhash.Hash
+)
+
+// BenchmarkTxHash benchmarks the performance of calculating the hash of a
+// transaction.
+func BenchmarkTxHash(b *testing.B) {
+	// Make a new block from the test block, we'll then call the Bytes
+	// function to cache the serialized block. Afterwards we all
+	// Transactions to populate the serialization cache.
+	testBlock := btcutil.NewBlock(&Block100000)
+	_, _ = testBlock.Bytes()
+
+	// The second transaction in the block has no witness data. The first
+	// does however.
+	testTx := testBlock.Transactions()[1]
+	testTx2 := testBlock.Transactions()[0]
+
+	// Run a benchmark for the portion that needs to strip the non-witness
+	// data from the transaction.
+	b.Run("tx_hash_has_witness", func(b *testing.B) {
+		b.ResetTimer()
+		b.ReportAllocs()
+
+		var txHash *chainhash.Hash
+		for i := 0; i < b.N; i++ {
+			txHash = testTx2.Hash()
+		}
+
+		bencHash = txHash
+	})
+
+	// Next, run it for the portion that can just hash the bytes directly.
+	b.Run("tx_hash_no_witness", func(b *testing.B) {
+		b.ResetTimer()
+		b.ReportAllocs()
+
+		var txHash *chainhash.Hash
+		for i := 0; i < b.N; i++ {
+			txHash = testTx.Hash()
+		}
+
+		bencHash = txHash
+	})
+
+}
+
+// BenchmarkTxWitnessHash benchmarks the performance of calculating the hash of
+// a transaction.
+func BenchmarkTxWitnessHash(b *testing.B) {
+	// Make a new block from the test block, we'll then call the Bytes
+	// function to cache the serialized block. Afterwards we all
+	// Transactions to populate the serialization cache.
+	testBlock := btcutil.NewBlock(&Block100000)
+	_, _ = testBlock.Bytes()
+
+	// The first transaction in the block has been modified to have witness
+	// data.
+	testTx := testBlock.Transactions()[0]
+
+	b.ResetTimer()
+	b.ReportAllocs()
+
+	var txHash *chainhash.Hash
+	for i := 0; i < b.N; i++ {
+		txHash = testTx.WitnessHash()
+	}
+
+	bencHash = txHash
+
+}

btcutil/block.go

@@ -154,12 +154,32 @@ func (b *Block) Transactions() []*Tx {
 		b.transactions = make([]*Tx, len(b.msgBlock.Transactions))
 	}
 
+	// Offset of each tx.  80 accounts for the block header size.
+	offset := 80 + wire.VarIntSerializeSize(
+		uint64(len(b.msgBlock.Transactions)),
+	)
+
 	// Generate and cache the wrapped transactions for all that haven't
 	// already been done.
 	for i, tx := range b.transactions {
 		if tx == nil {
 			newTx := NewTx(b.msgBlock.Transactions[i])
 			newTx.SetIndex(i)
+
+			size := b.msgBlock.Transactions[i].SerializeSize()
+
+			// The block may not always have the serializedBlock.
+			if len(b.serializedBlock) > 0 {
+				// This allows for the reuse of the already
+				// serialized tx.
+				newTx.setBytes(
+					b.serializedBlock[offset : offset+size],
+				)
+
+				// Increment offset for this block.
+				offset += size
+			}
+
 			b.transactions[i] = newTx
 		}
 	}
@@ -234,6 +254,12 @@ func NewBlockFromBytes(serializedBlock []byte) (*Block, error) {
 		return nil, err
 	}
 	b.serializedBlock = serializedBlock
+
+	// This initializes []btcutil.Tx to have the serialized raw
+	// transactions cached.  Helps speed up things like generating the
+	// txhash.
+	b.Transactions()
+
 	return b, nil
 }
 
@@ -256,10 +282,19 @@ func NewBlockFromReader(r io.Reader) (*Block, error) {
 
 // NewBlockFromBlockAndBytes returns a new instance of a bitcoin block given
 // an underlying wire.MsgBlock and the serialized bytes for it.  See Block.
-func NewBlockFromBlockAndBytes(msgBlock *wire.MsgBlock, serializedBlock []byte) *Block {
-	return &Block{
+func NewBlockFromBlockAndBytes(msgBlock *wire.MsgBlock,
+	serializedBlock []byte) *Block {
+
+	b := &Block{
 		msgBlock:        msgBlock,
 		serializedBlock: serializedBlock,
 		blockHeight:     BlockHeightUnknown,
 	}
+
+	// This initializes []btcutil.Tx to have the serialized raw
+	// transactions cached.  Helps speed up things like generating the
+	// txhash.
+	b.Transactions()
+
+	return b
 }

btcutil/tx.go

@@ -27,6 +27,7 @@ type Tx struct {
 	txHashWitness *chainhash.Hash // Cached transaction witness hash
 	txHasWitness  *bool           // If the transaction has witness data
 	txIndex       int             // Position within a block or TxIndexUnknown
+	rawBytes      []byte          // Raw bytes for the tx in the raw block.
 }
 
 // MsgTx returns the underlying wire.MsgTx for the transaction.
@@ -35,32 +36,96 @@ func (t *Tx) MsgTx() *wire.MsgTx {
 	return t.msgTx
 }
 
-// Hash returns the hash of the transaction.  This is equivalent to
-// calling TxHash on the underlying wire.MsgTx, however it caches the
-// result so subsequent calls are more efficient.
+// Hash returns the hash of the transaction.  This is equivalent to calling
+// TxHash on the underlying wire.MsgTx, however it caches the result so
+// subsequent calls are more efficient.  If the Tx has the raw bytes of the tx
+// cached, it will use that and skip serialization.
 func (t *Tx) Hash() *chainhash.Hash {
 	// Return the cached hash if it has already been generated.
 	if t.txHash != nil {
 		return t.txHash
 	}
 
-	// Cache the hash and return it.
-	hash := t.msgTx.TxHash()
+	// If the rawBytes aren't available, call msgtx.TxHash.
+	if t.rawBytes == nil {
+		hash := t.msgTx.TxHash()
+		t.txHash = &hash
+		return &hash
+	}
+
+	// If we have the raw bytes, then don't call msgTx.TxHash as that has
+	// the overhead of serialization. Instead, we can take the existing
+	// serialized bytes and hash them to speed things up.
+	var hash chainhash.Hash
+	if t.HasWitness() {
+		// If the raw bytes contain the witness, we must strip it out
+		// before calculating the hash.
+		baseSize := t.msgTx.SerializeSizeStripped()
+		nonWitnessBytes := make([]byte, 0, baseSize)
+
+		// Append the version bytes.
+		offset := 4
+		nonWitnessBytes = append(
+			nonWitnessBytes, t.rawBytes[:offset]...,
+		)
+
+		// Append the input and output bytes.  -8 to account for the
+		// version bytes and the locktime bytes.
+		//
+		// Skip the 2 bytes for the witness encoding.
+		offset += 2
+		nonWitnessBytes = append(
+			nonWitnessBytes,
+			t.rawBytes[offset:offset+baseSize-8]...,
+		)
+
+		// Append the last 4 bytes which are the locktime bytes.
+		nonWitnessBytes = append(
+			nonWitnessBytes, t.rawBytes[len(t.rawBytes)-4:]...,
+		)
+
+		// We purposely call doublehashh here instead of doublehashraw
+		// as we don't have the serialization overhead and avoiding the
+		// 1 alloc is better in this case.
+		hash = chainhash.DoubleHashRaw(func(w io.Writer) error {
+			_, err := w.Write(nonWitnessBytes)
+			return err
+		})
+	} else {
+		// If the raw bytes don't have the witness, we can use it
+		// directly.
+		//
+		// We purposely call doublehashh here instead of doublehashraw
+		// as we don't have the serialization overhead and avoiding the
+		// 1 alloc is better in this case.
+		hash = chainhash.DoubleHashRaw(func(w io.Writer) error {
+			_, err := w.Write(t.rawBytes)
+			return err
+		})
+	}
+
 	t.txHash = &hash
 	return &hash
 }
 
 // WitnessHash returns the witness hash (wtxid) of the transaction.  This is
 // equivalent to calling WitnessHash on the underlying wire.MsgTx, however it
-// caches the result so subsequent calls are more efficient.
+// caches the result so subsequent calls are more efficient.  If the Tx has the
+// raw bytes of the tx cached, it will use that and skip serialization.
 func (t *Tx) WitnessHash() *chainhash.Hash {
 	// Return the cached hash if it has already been generated.
 	if t.txHashWitness != nil {
 		return t.txHashWitness
 	}
 
 	// Cache the hash and return it.
-	hash := t.msgTx.WitnessHash()
+	var hash chainhash.Hash
+	if len(t.rawBytes) > 0 {
+		hash = chainhash.DoubleHashH(t.rawBytes)
+	} else {
+		hash = t.msgTx.WitnessHash()
+	}
+
 	t.txHashWitness = &hash
 	return &hash
 }
@@ -99,6 +164,11 @@ func NewTx(msgTx *wire.MsgTx) *Tx {
 	}
 }
 
+// setBytes sets the raw bytes of the tx.
+func (t *Tx) setBytes(bytes []byte) {
+	t.rawBytes = bytes
+}
+
 // NewTxFromBytes returns a new instance of a bitcoin transaction given the
 // serialized bytes.  See Tx.
 func NewTxFromBytes(serializedTx []byte) (*Tx, error) {