core/rawdb: check hash before return data from ancient db (#20195)

* core/rawdb: check hash before return data from ancient db

* core/rawdb: fix lint

* core/rawdb: calculate the hash in the fly

Author: rjl493456442

core/rawdb/accessors_chain.go

@@ -23,6 +23,7 @@ import (
 
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/core/types"
+	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/log"
 	"github.com/ethereum/go-ethereum/params"
@@ -173,18 +174,27 @@ func WriteFastTrieProgress(db ethdb.KeyValueWriter, count uint64) {
 
 // ReadHeaderRLP retrieves a block header in its raw RLP database encoding.
 func ReadHeaderRLP(db ethdb.Reader, hash common.Hash, number uint64) rlp.RawValue {
+	// First try to look up the data in ancient database. Extra hash
+	// comparison is necessary since ancient database only maintains
+	// the canonical data.
 	data, _ := db.Ancient(freezerHeaderTable, number)
-	if len(data) == 0 {
-		data, _ = db.Get(headerKey(number, hash))
-		// In the background freezer is moving data from leveldb to flatten files.
-		// So during the first check for ancient db, the data is not yet in there,
-		// but when we reach into leveldb, the data was already moved. That would
-		// result in a not found error.
-		if len(data) == 0 {
-			data, _ = db.Ancient(freezerHeaderTable, number)
-		}
+	if len(data) > 0 && crypto.Keccak256Hash(data) == hash {
+		return data
+	}
+	// Then try to look up the data in leveldb.
+	data, _ = db.Get(headerKey(number, hash))
+	if len(data) > 0 {
+		return data
+	}
+	// In the background freezer is moving data from leveldb to flatten files.
+	// So during the first check for ancient db, the data is not yet in there,
+	// but when we reach into leveldb, the data was already moved. That would
+	// result in a not found error.
+	data, _ = db.Ancient(freezerHeaderTable, number)
+	if len(data) > 0 && crypto.Keccak256Hash(data) == hash {
+		return data
 	}
-	return data
+	return nil // Can't find the data anywhere.
 }
 
 // HasHeader verifies the existence of a block header corresponding to the hash.
@@ -251,18 +261,33 @@ func deleteHeaderWithoutNumber(db ethdb.KeyValueWriter, hash common.Hash, number
 
 // ReadBodyRLP retrieves the block body (transactions and uncles) in RLP encoding.
 func ReadBodyRLP(db ethdb.Reader, hash common.Hash, number uint64) rlp.RawValue {
+	// First try to look up the data in ancient database. Extra hash
+	// comparison is necessary since ancient database only maintains
+	// the canonical data.
 	data, _ := db.Ancient(freezerBodiesTable, number)
-	if len(data) == 0 {
-		data, _ = db.Get(blockBodyKey(number, hash))
-		// In the background freezer is moving data from leveldb to flatten files.
-		// So during the first check for ancient db, the data is not yet in there,
-		// but when we reach into leveldb, the data was already moved. That would
-		// result in a not found error.
-		if len(data) == 0 {
-			data, _ = db.Ancient(freezerBodiesTable, number)
+	if len(data) > 0 {
+		h, _ := db.Ancient(freezerHashTable, number)
+		if common.BytesToHash(h) == hash {
+			return data
+		}
+	}
+	// Then try to look up the data in leveldb.
+	data, _ = db.Get(blockBodyKey(number, hash))
+	if len(data) > 0 {
+		return data
+	}
+	// In the background freezer is moving data from leveldb to flatten files.
+	// So during the first check for ancient db, the data is not yet in there,
+	// but when we reach into leveldb, the data was already moved. That would
+	// result in a not found error.
+	data, _ = db.Ancient(freezerBodiesTable, number)
+	if len(data) > 0 {
+		h, _ := db.Ancient(freezerHashTable, number)
+		if common.BytesToHash(h) == hash {
+			return data
 		}
 	}
-	return data
+	return nil // Can't find the data anywhere.
 }
 
 // WriteBodyRLP stores an RLP encoded block body into the database.
@@ -315,18 +340,33 @@ func DeleteBody(db ethdb.KeyValueWriter, hash common.Hash, number uint64) {
 
 // ReadTdRLP retrieves a block's total difficulty corresponding to the hash in RLP encoding.
 func ReadTdRLP(db ethdb.Reader, hash common.Hash, number uint64) rlp.RawValue {
+	// First try to look up the data in ancient database. Extra hash
+	// comparison is necessary since ancient database only maintains
+	// the canonical data.
 	data, _ := db.Ancient(freezerDifficultyTable, number)
-	if len(data) == 0 {
-		data, _ = db.Get(headerTDKey(number, hash))
-		// In the background freezer is moving data from leveldb to flatten files.
-		// So during the first check for ancient db, the data is not yet in there,
-		// but when we reach into leveldb, the data was already moved. That would
-		// result in a not found error.
-		if len(data) == 0 {
-			data, _ = db.Ancient(freezerDifficultyTable, number)
+	if len(data) > 0 {
+		h, _ := db.Ancient(freezerHashTable, number)
+		if common.BytesToHash(h) == hash {
+			return data
+		}
+	}
+	// Then try to look up the data in leveldb.
+	data, _ = db.Get(headerTDKey(number, hash))
+	if len(data) > 0 {
+		return data
+	}
+	// In the background freezer is moving data from leveldb to flatten files.
+	// So during the first check for ancient db, the data is not yet in there,
+	// but when we reach into leveldb, the data was already moved. That would
+	// result in a not found error.
+	data, _ = db.Ancient(freezerDifficultyTable, number)
+	if len(data) > 0 {
+		h, _ := db.Ancient(freezerHashTable, number)
+		if common.BytesToHash(h) == hash {
+			return data
 		}
 	}
-	return data
+	return nil // Can't find the data anywhere.
 }
 
 // ReadTd retrieves a block's total difficulty corresponding to the hash.
@@ -375,18 +415,33 @@ func HasReceipts(db ethdb.Reader, hash common.Hash, number uint64) bool {
 
 // ReadReceiptsRLP retrieves all the transaction receipts belonging to a block in RLP encoding.
 func ReadReceiptsRLP(db ethdb.Reader, hash common.Hash, number uint64) rlp.RawValue {
+	// First try to look up the data in ancient database. Extra hash
+	// comparison is necessary since ancient database only maintains
+	// the canonical data.
 	data, _ := db.Ancient(freezerReceiptTable, number)
-	if len(data) == 0 {
-		data, _ = db.Get(blockReceiptsKey(number, hash))
-		// In the background freezer is moving data from leveldb to flatten files.
-		// So during the first check for ancient db, the data is not yet in there,
-		// but when we reach into leveldb, the data was already moved. That would
-		// result in a not found error.
-		if len(data) == 0 {
-			data, _ = db.Ancient(freezerReceiptTable, number)
+	if len(data) > 0 {
+		h, _ := db.Ancient(freezerHashTable, number)
+		if common.BytesToHash(h) == hash {
+			return data
+		}
+	}
+	// Then try to look up the data in leveldb.
+	data, _ = db.Get(blockReceiptsKey(number, hash))
+	if len(data) > 0 {
+		return data
+	}
+	// In the background freezer is moving data from leveldb to flatten files.
+	// So during the first check for ancient db, the data is not yet in there,
+	// but when we reach into leveldb, the data was already moved. That would
+	// result in a not found error.
+	data, _ = db.Ancient(freezerReceiptTable, number)
+	if len(data) > 0 {
+		h, _ := db.Ancient(freezerHashTable, number)
+		if common.BytesToHash(h) == hash {
+			return data
 		}
 	}
-	return data
+	return nil // Can't find the data anywhere.
 }
 
 // ReadRawReceipts retrieves all the transaction receipts belonging to a block.

core/rawdb/accessors_chain_test.go

@@ -20,7 +20,9 @@ import (
 	"bytes"
 	"encoding/hex"
 	"fmt"
+	"io/ioutil"
 	"math/big"
+	"os"
 	"testing"
 
 	"github.com/ethereum/go-ethereum/common"
@@ -358,3 +360,67 @@ func checkReceiptsRLP(have, want types.Receipts) error {
 	}
 	return nil
 }
+
+func TestAncientStorage(t *testing.T) {
+	// Freezer style fast import the chain.
+	frdir, err := ioutil.TempDir("", "")
+	if err != nil {
+		t.Fatalf("failed to create temp freezer dir: %v", err)
+	}
+	defer os.Remove(frdir)
+
+	db, err := NewDatabaseWithFreezer(NewMemoryDatabase(), frdir, "")
+	if err != nil {
+		t.Fatalf("failed to create database with ancient backend")
+	}
+	// Create a test block
+	block := types.NewBlockWithHeader(&types.Header{
+		Number:      big.NewInt(0),
+		Extra:       []byte("test block"),
+		UncleHash:   types.EmptyUncleHash,
+		TxHash:      types.EmptyRootHash,
+		ReceiptHash: types.EmptyRootHash,
+	})
+	// Ensure nothing non-existent will be read
+	hash, number := block.Hash(), block.NumberU64()
+	if blob := ReadHeaderRLP(db, hash, number); len(blob) > 0 {
+		t.Fatalf("non existent header returned")
+	}
+	if blob := ReadBodyRLP(db, hash, number); len(blob) > 0 {
+		t.Fatalf("non existent body returned")
+	}
+	if blob := ReadReceiptsRLP(db, hash, number); len(blob) > 0 {
+		t.Fatalf("non existent receipts returned")
+	}
+	if blob := ReadTdRLP(db, hash, number); len(blob) > 0 {
+		t.Fatalf("non existent td returned")
+	}
+	// Write and verify the header in the database
+	WriteAncientBlock(db, block, nil, big.NewInt(100))
+	if blob := ReadHeaderRLP(db, hash, number); len(blob) == 0 {
+		t.Fatalf("no header returned")
+	}
+	if blob := ReadBodyRLP(db, hash, number); len(blob) == 0 {
+		t.Fatalf("no body returned")
+	}
+	if blob := ReadReceiptsRLP(db, hash, number); len(blob) == 0 {
+		t.Fatalf("no receipts returned")
+	}
+	if blob := ReadTdRLP(db, hash, number); len(blob) == 0 {
+		t.Fatalf("no td returned")
+	}
+	// Use a fake hash for data retrieval, nothing should be returned.
+	fakeHash := common.BytesToHash([]byte{0x01, 0x02, 0x03})
+	if blob := ReadHeaderRLP(db, fakeHash, number); len(blob) != 0 {
+		t.Fatalf("invalid header returned")
+	}
+	if blob := ReadBodyRLP(db, fakeHash, number); len(blob) != 0 {
+		t.Fatalf("invalid body returned")
+	}
+	if blob := ReadReceiptsRLP(db, fakeHash, number); len(blob) != 0 {
+		t.Fatalf("invalid receipts returned")
+	}
+	if blob := ReadTdRLP(db, fakeHash, number); len(blob) != 0 {
+		t.Fatalf("invalid td returned")
+	}
+}