diff --git a/accounts/abi/abigen/bind.go b/accounts/abi/abigen/bind.go index 56e5e214de..08624b04ba 100644 --- a/accounts/abi/abigen/bind.go +++ b/accounts/abi/abigen/bind.go @@ -33,6 +33,10 @@ import ( "github.com/ethereum/go-ethereum/log" ) +var ( + intRegex = regexp.MustCompile(`(u)?int([0-9]*)`) +) + func isKeyWord(arg string) bool { switch arg { case "break": @@ -299,7 +303,7 @@ func bindBasicType(kind abi.Type) string { case abi.AddressTy: return "common.Address" case abi.IntTy, abi.UintTy: - parts := regexp.MustCompile(`(u)?int([0-9]*)`).FindStringSubmatch(kind.String()) + parts := intRegex.FindStringSubmatch(kind.String()) switch parts[2] { case "8", "16", "32", "64": return fmt.Sprintf("%sint%s", parts[1], parts[2]) diff --git a/cmd/devp2p/internal/v4test/discv4tests.go b/cmd/devp2p/internal/v4test/discv4tests.go index 963df6cdbc..de97d7a276 100644 --- a/cmd/devp2p/internal/v4test/discv4tests.go +++ b/cmd/devp2p/internal/v4test/discv4tests.go @@ -27,6 +27,7 @@ import ( "github.com/ethereum/go-ethereum/crypto" "github.com/ethereum/go-ethereum/internal/utesting" "github.com/ethereum/go-ethereum/p2p/discover/v4wire" + "github.com/ethereum/go-ethereum/p2p/enode" ) const ( @@ -501,6 +502,36 @@ func FindnodeAmplificationWrongIP(t *utesting.T) { } } +func ENRRequest(t *utesting.T) { + t.Log(`This test sends an ENRRequest packet and expects a response containing a valid ENR.`) + + te := newTestEnv(Remote, Listen1, Listen2) + defer te.close() + bond(t, te) + + req := &v4wire.ENRRequest{Expiration: futureExpiration()} + hash := te.send(te.l1, req) + + response, _, err := te.read(te.l1) + if err != nil { + t.Fatal("read error:", err) + } + enrResp, ok := response.(*v4wire.ENRResponse) + if !ok { + t.Fatalf("expected ENRResponse packet, got %T", response) + } + if !bytes.Equal(enrResp.ReplyTok, hash) { + t.Errorf("wrong hash in response packet: got %x, want %x", enrResp.ReplyTok, hash) + } + node, err := enode.New(enode.ValidSchemes, &enrResp.Record) + if err != nil { + t.Errorf("invalid record in response: %v", err) + } + if node.ID() != te.remote.ID() { + t.Errorf("wrong node ID in response: got %v, want %v", node.ID(), te.remote.ID()) + } +} + var AllTests = []utesting.Test{ {Name: "Ping/Basic", Fn: BasicPing}, {Name: "Ping/WrongTo", Fn: PingWrongTo}, @@ -510,6 +541,7 @@ var AllTests = []utesting.Test{ {Name: "Ping/PastExpiration", Fn: PingPastExpiration}, {Name: "Ping/WrongPacketType", Fn: WrongPacketType}, {Name: "Ping/BondThenPingWithWrongFrom", Fn: BondThenPingWithWrongFrom}, + {Name: "ENRRequest", Fn: ENRRequest}, {Name: "Findnode/WithoutEndpointProof", Fn: FindnodeWithoutEndpointProof}, {Name: "Findnode/BasicFindnode", Fn: BasicFindnode}, {Name: "Findnode/UnsolicitedNeighbors", Fn: UnsolicitedNeighbors}, diff --git a/common/hexutil/hexutil.go b/common/hexutil/hexutil.go index d3201850a8..d6b6b867f2 100644 --- a/common/hexutil/hexutil.go +++ b/common/hexutil/hexutil.go @@ -34,11 +34,10 @@ import ( "encoding/hex" "fmt" "math/big" + "math/bits" "strconv" ) -const uintBits = 32 << (uint64(^uint(0)) >> 63) - // Errors var ( ErrEmptyString = &decError{"empty hex string"} @@ -48,7 +47,7 @@ var ( ErrEmptyNumber = &decError{"hex string \"0x\""} ErrLeadingZero = &decError{"hex number with leading zero digits"} ErrUint64Range = &decError{"hex number > 64 bits"} - ErrUintRange = &decError{fmt.Sprintf("hex number > %d bits", uintBits)} + ErrUintRange = &decError{fmt.Sprintf("hex number > %d bits", bits.UintSize)} ErrBig256Range = &decError{"hex number > 256 bits"} ) diff --git a/common/hexutil/json_test.go b/common/hexutil/json_test.go index 7cca300951..a014438458 100644 --- a/common/hexutil/json_test.go +++ b/common/hexutil/json_test.go @@ -22,6 +22,7 @@ import ( "encoding/json" "errors" "math/big" + "math/bits" "testing" "github.com/holiman/uint256" @@ -384,7 +385,7 @@ func TestUnmarshalUint(t *testing.T) { for _, test := range unmarshalUintTests { var v Uint err := json.Unmarshal([]byte(test.input), &v) - if uintBits == 32 && test.wantErr32bit != nil { + if bits.UintSize == 32 && test.wantErr32bit != nil { checkError(t, test.input, err, test.wantErr32bit) continue } diff --git a/core/types/tx_setcode.go b/core/types/tx_setcode.go index b8e38ef1f7..f2281d4ae7 100644 --- a/core/types/tx_setcode.go +++ b/core/types/tx_setcode.go @@ -89,7 +89,7 @@ type authorizationMarshaling struct { // SignSetCode creates a signed the SetCode authorization. func SignSetCode(prv *ecdsa.PrivateKey, auth SetCodeAuthorization) (SetCodeAuthorization, error) { - sighash := auth.sigHash() + sighash := auth.SigHash() sig, err := crypto.Sign(sighash[:], prv) if err != nil { return SetCodeAuthorization{}, err @@ -105,7 +105,8 @@ func SignSetCode(prv *ecdsa.PrivateKey, auth SetCodeAuthorization) (SetCodeAutho }, nil } -func (a *SetCodeAuthorization) sigHash() common.Hash { +// SigHash returns the hash of SetCodeAuthorization for signing. +func (a *SetCodeAuthorization) SigHash() common.Hash { return prefixedRlpHash(0x05, []any{ a.ChainID, a.Address, @@ -115,7 +116,7 @@ func (a *SetCodeAuthorization) sigHash() common.Hash { // Authority recovers the the authorizing account of an authorization. func (a *SetCodeAuthorization) Authority() (common.Address, error) { - sighash := a.sigHash() + sighash := a.SigHash() if !crypto.ValidateSignatureValues(a.V, a.R.ToBig(), a.S.ToBig(), true) { return common.Address{}, ErrInvalidSig } diff --git a/core/vm/analysis_legacy.go b/core/vm/analysis_legacy.go index 38af9084ac..a445e2048e 100644 --- a/core/vm/analysis_legacy.go +++ b/core/vm/analysis_legacy.go @@ -25,16 +25,16 @@ const ( set7BitsMask = uint16(0b111_1111) ) -// bitvec is a bit vector which maps bytes in a program. +// BitVec is a bit vector which maps bytes in a program. // An unset bit means the byte is an opcode, a set bit means // it's data (i.e. argument of PUSHxx). -type bitvec []byte +type BitVec []byte -func (bits bitvec) set1(pos uint64) { +func (bits BitVec) set1(pos uint64) { bits[pos/8] |= 1 << (pos % 8) } -func (bits bitvec) setN(flag uint16, pos uint64) { +func (bits BitVec) setN(flag uint16, pos uint64) { a := flag << (pos % 8) bits[pos/8] |= byte(a) if b := byte(a >> 8); b != 0 { @@ -42,13 +42,13 @@ func (bits bitvec) setN(flag uint16, pos uint64) { } } -func (bits bitvec) set8(pos uint64) { +func (bits BitVec) set8(pos uint64) { a := byte(0xFF << (pos % 8)) bits[pos/8] |= a bits[pos/8+1] = ^a } -func (bits bitvec) set16(pos uint64) { +func (bits BitVec) set16(pos uint64) { a := byte(0xFF << (pos % 8)) bits[pos/8] |= a bits[pos/8+1] = 0xFF @@ -56,23 +56,23 @@ func (bits bitvec) set16(pos uint64) { } // codeSegment checks if the position is in a code segment. -func (bits *bitvec) codeSegment(pos uint64) bool { +func (bits *BitVec) codeSegment(pos uint64) bool { return (((*bits)[pos/8] >> (pos % 8)) & 1) == 0 } // codeBitmap collects data locations in code. -func codeBitmap(code []byte) bitvec { +func codeBitmap(code []byte) BitVec { // The bitmap is 4 bytes longer than necessary, in case the code // ends with a PUSH32, the algorithm will set bits on the // bitvector outside the bounds of the actual code. - bits := make(bitvec, len(code)/8+1+4) + bits := make(BitVec, len(code)/8+1+4) return codeBitmapInternal(code, bits) } // codeBitmapInternal is the internal implementation of codeBitmap. // It exists for the purpose of being able to run benchmark tests // without dynamic allocations affecting the results. -func codeBitmapInternal(code, bits bitvec) bitvec { +func codeBitmapInternal(code, bits BitVec) BitVec { for pc := uint64(0); pc < uint64(len(code)); { op := OpCode(code[pc]) pc++ diff --git a/core/vm/analysis_legacy_test.go b/core/vm/analysis_legacy_test.go index 471d2b4ffb..f84a4abc92 100644 --- a/core/vm/analysis_legacy_test.go +++ b/core/vm/analysis_legacy_test.go @@ -90,7 +90,7 @@ func BenchmarkJumpdestOpAnalysis(bench *testing.B) { for i := range code { code[i] = byte(op) } - bits := make(bitvec, len(code)/8+1+4) + bits := make(BitVec, len(code)/8+1+4) b.ResetTimer() for i := 0; i < b.N; i++ { clear(bits) diff --git a/core/vm/contract.go b/core/vm/contract.go index 0eaa91d959..165ca833f8 100644 --- a/core/vm/contract.go +++ b/core/vm/contract.go @@ -31,8 +31,8 @@ type Contract struct { caller common.Address address common.Address - jumpdests map[common.Hash]bitvec // Aggregated result of JUMPDEST analysis. - analysis bitvec // Locally cached result of JUMPDEST analysis + jumpDests JumpDestCache // Aggregated result of JUMPDEST analysis. + analysis BitVec // Locally cached result of JUMPDEST analysis Code []byte CodeHash common.Hash @@ -47,15 +47,15 @@ type Contract struct { } // NewContract returns a new contract environment for the execution of EVM. -func NewContract(caller common.Address, address common.Address, value *uint256.Int, gas uint64, jumpDests map[common.Hash]bitvec) *Contract { - // Initialize the jump analysis map if it's nil, mostly for tests +func NewContract(caller common.Address, address common.Address, value *uint256.Int, gas uint64, jumpDests JumpDestCache) *Contract { + // Initialize the jump analysis cache if it's nil, mostly for tests if jumpDests == nil { - jumpDests = make(map[common.Hash]bitvec) + jumpDests = newMapJumpDests() } return &Contract{ caller: caller, address: address, - jumpdests: jumpDests, + jumpDests: jumpDests, Gas: gas, value: value, } @@ -87,12 +87,12 @@ func (c *Contract) isCode(udest uint64) bool { // contracts ( not temporary initcode), we store the analysis in a map if c.CodeHash != (common.Hash{}) { // Does parent context have the analysis? - analysis, exist := c.jumpdests[c.CodeHash] + analysis, exist := c.jumpDests.Load(c.CodeHash) if !exist { // Do the analysis and save in parent context // We do not need to store it in c.analysis analysis = codeBitmap(c.Code) - c.jumpdests[c.CodeHash] = analysis + c.jumpDests.Store(c.CodeHash, analysis) } // Also stash it in current contract for faster access c.analysis = analysis diff --git a/core/vm/evm.go b/core/vm/evm.go index b45a434545..143b7e08a2 100644 --- a/core/vm/evm.go +++ b/core/vm/evm.go @@ -122,9 +122,8 @@ type EVM struct { // precompiles holds the precompiled contracts for the current epoch precompiles map[common.Address]PrecompiledContract - // jumpDests is the aggregated result of JUMPDEST analysis made through - // the life cycle of EVM. - jumpDests map[common.Hash]bitvec + // jumpDests stores results of JUMPDEST analysis. + jumpDests JumpDestCache } // NewEVM constructs an EVM instance with the supplied block context, state @@ -138,7 +137,7 @@ func NewEVM(blockCtx BlockContext, statedb StateDB, chainConfig *params.ChainCon Config: config, chainConfig: chainConfig, chainRules: chainConfig.Rules(blockCtx.BlockNumber, blockCtx.Random != nil, blockCtx.Time), - jumpDests: make(map[common.Hash]bitvec), + jumpDests: newMapJumpDests(), } evm.precompiles = activePrecompiledContracts(evm.chainRules) evm.interpreter = NewEVMInterpreter(evm) @@ -152,6 +151,11 @@ func (evm *EVM) SetPrecompiles(precompiles PrecompiledContracts) { evm.precompiles = precompiles } +// SetJumpDestCache configures the analysis cache. +func (evm *EVM) SetJumpDestCache(jumpDests JumpDestCache) { + evm.jumpDests = jumpDests +} + // SetTxContext resets the EVM with a new transaction context. // This is not threadsafe and should only be done very cautiously. func (evm *EVM) SetTxContext(txCtx TxContext) { diff --git a/core/vm/jumpdests.go b/core/vm/jumpdests.go new file mode 100644 index 0000000000..1a30c1943f --- /dev/null +++ b/core/vm/jumpdests.go @@ -0,0 +1,47 @@ +// Copyright 2024 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 . + +package vm + +import "github.com/ethereum/go-ethereum/common" + +// JumpDestCache represents the cache of jumpdest analysis results. +type JumpDestCache interface { + // Load retrieves the cached jumpdest analysis for the given code hash. + // Returns the BitVec and true if found, or nil and false if not cached. + Load(codeHash common.Hash) (BitVec, bool) + + // Store saves the jumpdest analysis for the given code hash. + Store(codeHash common.Hash, vec BitVec) +} + +// mapJumpDests is the default implementation of JumpDests using a map. +// This implementation is not thread-safe and is meant to be used per EVM instance. +type mapJumpDests map[common.Hash]BitVec + +// newMapJumpDests creates a new map-based JumpDests implementation. +func newMapJumpDests() JumpDestCache { + return make(mapJumpDests) +} + +func (j mapJumpDests) Load(codeHash common.Hash) (BitVec, bool) { + vec, ok := j[codeHash] + return vec, ok +} + +func (j mapJumpDests) Store(codeHash common.Hash, vec BitVec) { + j[codeHash] = vec +} diff --git a/eth/ethconfig/config.go b/eth/ethconfig/config.go index 99aae48882..0579286861 100644 --- a/eth/ethconfig/config.go +++ b/eth/ethconfig/config.go @@ -154,7 +154,7 @@ type Config struct { // RPCEVMTimeout is the global timeout for eth-call. RPCEVMTimeout time.Duration - // RPCTxFeeCap is the global transaction fee(price * gaslimit) cap for + // RPCTxFeeCap is the global transaction fee (price * gas limit) cap for // send-transaction variants. The unit is ether. RPCTxFeeCap float64 diff --git a/eth/filters/api.go b/eth/filters/api.go index 232af23957..c929810a12 100644 --- a/eth/filters/api.go +++ b/eth/filters/api.go @@ -38,6 +38,8 @@ var ( errInvalidTopic = errors.New("invalid topic(s)") errFilterNotFound = errors.New("filter not found") errInvalidBlockRange = errors.New("invalid block range params") + errUnknownBlock = errors.New("unknown block") + errBlockHashWithRange = errors.New("can't specify fromBlock/toBlock with blockHash") errPendingLogsUnsupported = errors.New("pending logs are not supported") errExceedMaxTopics = errors.New("exceed max topics") errExceedMaxAddresses = errors.New("exceed max addresses") @@ -348,8 +350,13 @@ func (api *FilterAPI) GetLogs(ctx context.Context, crit FilterCriteria) ([]*type if len(crit.Addresses) > maxAddresses { return nil, errExceedMaxAddresses } + var filter *Filter if crit.BlockHash != nil { + if crit.FromBlock != nil || crit.ToBlock != nil { + return nil, errBlockHashWithRange + } + // Block filter requested, construct a single-shot filter filter = api.sys.NewBlockFilter(*crit.BlockHash, crit.Addresses, crit.Topics) } else { @@ -372,6 +379,7 @@ func (api *FilterAPI) GetLogs(ctx context.Context, crit FilterCriteria) ([]*type // Construct the range filter filter = api.sys.NewRangeFilter(begin, end, crit.Addresses, crit.Topics) } + // Run the filter and return all the logs logs, err := filter.Logs(ctx) if err != nil { diff --git a/eth/filters/filter.go b/eth/filters/filter.go index ada478ae1d..1a9918d0ee 100644 --- a/eth/filters/filter.go +++ b/eth/filters/filter.go @@ -85,7 +85,7 @@ func (f *Filter) Logs(ctx context.Context) ([]*types.Log, error) { return nil, err } if header == nil { - return nil, errors.New("unknown block") + return nil, errUnknownBlock } if header.Number.Uint64() < f.sys.backend.HistoryPruningCutoff() { return nil, &history.PrunedHistoryError{} @@ -456,7 +456,6 @@ func (f *Filter) blockLogs(ctx context.Context, header *types.Header) ([]*types. // checkMatches checks if the receipts belonging to the given header contain any log events that // match the filter criteria. This function is called when the bloom filter signals a potential match. -// skipFilter signals all logs of the given block are requested. func (f *Filter) checkMatches(ctx context.Context, header *types.Header) ([]*types.Log, error) { hash := header.Hash() // Logs in cache are partially filled with context data diff --git a/eth/filters/filter_system.go b/eth/filters/filter_system.go index 82c91266c0..751cd417e8 100644 --- a/eth/filters/filter_system.go +++ b/eth/filters/filter_system.go @@ -207,7 +207,7 @@ type EventSystem struct { } // NewEventSystem creates a new manager that listens for event on the given mux, -// parses and filters them. It uses the all map to retrieve filter changes. The +// parses and filters them. It uses an internal map to retrieve filter changes. The // work loop holds its own index that is used to forward events to filters. // // The returned manager has a loop that needs to be stopped with the Stop function diff --git a/eth/filters/filter_system_test.go b/eth/filters/filter_system_test.go index 3e686ca2eb..013c1ae527 100644 --- a/eth/filters/filter_system_test.go +++ b/eth/filters/filter_system_test.go @@ -450,24 +450,65 @@ func TestInvalidGetLogsRequest(t *testing.T) { t.Parallel() var ( - db = rawdb.NewMemoryDatabase() - _, sys = newTestFilterSystem(db, Config{}) - api = NewFilterAPI(sys) - blockHash = common.HexToHash("0x1111111111111111111111111111111111111111111111111111111111111111") + genesis = &core.Genesis{ + Config: params.TestChainConfig, + BaseFee: big.NewInt(params.InitialBaseFee), + } + db, blocks, _ = core.GenerateChainWithGenesis(genesis, ethash.NewFaker(), 10, func(i int, gen *core.BlockGen) {}) + _, sys = newTestFilterSystem(db, Config{}) + api = NewFilterAPI(sys) + blockHash = blocks[0].Hash() + unknownBlockHash = common.HexToHash("0x1111111111111111111111111111111111111111111111111111111111111111") ) - // Reason: Cannot specify both BlockHash and FromBlock/ToBlock) - testCases := []FilterCriteria{ - 0: {BlockHash: &blockHash, FromBlock: big.NewInt(100)}, - 1: {BlockHash: &blockHash, ToBlock: big.NewInt(500)}, - 2: {BlockHash: &blockHash, FromBlock: big.NewInt(rpc.LatestBlockNumber.Int64())}, - 3: {BlockHash: &blockHash, Topics: [][]common.Hash{{}, {}, {}, {}, {}}}, - 4: {BlockHash: &blockHash, Addresses: make([]common.Address, maxAddresses+1)}, + // Insert the blocks into the chain so filter can look them up + blockchain, err := core.NewBlockChain(db, genesis, ethash.NewFaker(), nil) + if err != nil { + t.Fatalf("failed to create tester chain: %v", err) + } + if n, err := blockchain.InsertChain(blocks); err != nil { + t.Fatalf("block %d: failed to insert into chain: %v", n, err) + } + + type testcase struct { + f FilterCriteria + err error + } + testCases := []testcase{ + { + f: FilterCriteria{BlockHash: &blockHash, FromBlock: big.NewInt(100)}, + err: errBlockHashWithRange, + }, + { + f: FilterCriteria{BlockHash: &blockHash, ToBlock: big.NewInt(500)}, + err: errBlockHashWithRange, + }, + { + f: FilterCriteria{BlockHash: &blockHash, FromBlock: big.NewInt(rpc.LatestBlockNumber.Int64())}, + err: errBlockHashWithRange, + }, + { + f: FilterCriteria{BlockHash: &unknownBlockHash}, + err: errUnknownBlock, + }, + { + f: FilterCriteria{BlockHash: &blockHash, Topics: [][]common.Hash{{}, {}, {}, {}, {}}}, + err: errExceedMaxTopics, + }, + { + f: FilterCriteria{BlockHash: &blockHash, Topics: [][]common.Hash{{}, {}, {}, {}, {}}}, + err: errExceedMaxTopics, + }, + { + f: FilterCriteria{BlockHash: &blockHash, Addresses: make([]common.Address, maxAddresses+1)}, + err: errExceedMaxAddresses, + }, } for i, test := range testCases { - if _, err := api.GetLogs(context.Background(), test); err == nil { - t.Errorf("Expected Logs for case #%d to fail", i) + _, err := api.GetLogs(context.Background(), test.f) + if !errors.Is(err, test.err) { + t.Errorf("case %d: wrong error: %q\nwant: %q", i, err, test.err) } } } diff --git a/trie/hasher.go b/trie/hasher.go index 16606808c9..a2a1f5b662 100644 --- a/trie/hasher.go +++ b/trie/hasher.go @@ -17,6 +17,8 @@ package trie import ( + "bytes" + "fmt" "sync" "github.com/ethereum/go-ethereum/crypto" @@ -54,7 +56,7 @@ func returnHasherToPool(h *hasher) { } // hash collapses a node down into a hash node. -func (h *hasher) hash(n node, force bool) node { +func (h *hasher) hash(n node, force bool) []byte { // Return the cached hash if it's available if hash, _ := n.cache(); hash != nil { return hash @@ -62,101 +64,110 @@ func (h *hasher) hash(n node, force bool) node { // Trie not processed yet, walk the children switch n := n.(type) { case *shortNode: - collapsed := h.hashShortNodeChildren(n) - hashed := h.shortnodeToHash(collapsed, force) - if hn, ok := hashed.(hashNode); ok { - n.flags.hash = hn - } else { - n.flags.hash = nil + enc := h.encodeShortNode(n) + if len(enc) < 32 && !force { + // Nodes smaller than 32 bytes are embedded directly in their parent. + // In such cases, return the raw encoded blob instead of the node hash. + // It's essential to deep-copy the node blob, as the underlying buffer + // of enc will be reused later. + buf := make([]byte, len(enc)) + copy(buf, enc) + return buf } - return hashed + hash := h.hashData(enc) + n.flags.hash = hash + return hash + case *fullNode: - collapsed := h.hashFullNodeChildren(n) - hashed := h.fullnodeToHash(collapsed, force) - if hn, ok := hashed.(hashNode); ok { - n.flags.hash = hn - } else { - n.flags.hash = nil + enc := h.encodeFullNode(n) + if len(enc) < 32 && !force { + // Nodes smaller than 32 bytes are embedded directly in their parent. + // In such cases, return the raw encoded blob instead of the node hash. + // It's essential to deep-copy the node blob, as the underlying buffer + // of enc will be reused later. + buf := make([]byte, len(enc)) + copy(buf, enc) + return buf } - return hashed - default: - // Value and hash nodes don't have children, so they're left as were + hash := h.hashData(enc) + n.flags.hash = hash + return hash + + case hashNode: + // hash nodes don't have children, so they're left as were return n - } -} -// hashShortNodeChildren returns a copy of the supplied shortNode, with its child -// being replaced by either the hash or an embedded node if the child is small. -func (h *hasher) hashShortNodeChildren(n *shortNode) *shortNode { - var collapsed shortNode - collapsed.Key = hexToCompact(n.Key) - switch n.Val.(type) { - case *fullNode, *shortNode: - collapsed.Val = h.hash(n.Val, false) default: - collapsed.Val = n.Val + panic(fmt.Errorf("unexpected node type, %T", n)) } - return &collapsed } -// hashFullNodeChildren returns a copy of the supplied fullNode, with its child -// being replaced by either the hash or an embedded node if the child is small. -func (h *hasher) hashFullNodeChildren(n *fullNode) *fullNode { - var children [17]node +// encodeShortNode encodes the provided shortNode into the bytes. Notably, the +// return slice must be deep-copied explicitly, otherwise the underlying slice +// will be reused later. +func (h *hasher) encodeShortNode(n *shortNode) []byte { + // Encode leaf node + if hasTerm(n.Key) { + var ln leafNodeEncoder + ln.Key = hexToCompact(n.Key) + ln.Val = n.Val.(valueNode) + ln.encode(h.encbuf) + return h.encodedBytes() + } + // Encode extension node + var en extNodeEncoder + en.Key = hexToCompact(n.Key) + en.Val = h.hash(n.Val, false) + en.encode(h.encbuf) + return h.encodedBytes() +} + +// fnEncoderPool is the pool for storing shared fullNode encoder to mitigate +// the significant memory allocation overhead. +var fnEncoderPool = sync.Pool{ + New: func() interface{} { + var enc fullnodeEncoder + return &enc + }, +} + +// encodeFullNode encodes the provided fullNode into the bytes. Notably, the +// return slice must be deep-copied explicitly, otherwise the underlying slice +// will be reused later. +func (h *hasher) encodeFullNode(n *fullNode) []byte { + fn := fnEncoderPool.Get().(*fullnodeEncoder) + fn.reset() + if h.parallel { var wg sync.WaitGroup for i := 0; i < 16; i++ { - if child := n.Children[i]; child != nil { - wg.Add(1) - go func(i int) { - hasher := newHasher(false) - children[i] = hasher.hash(child, false) - returnHasherToPool(hasher) - wg.Done() - }(i) - } else { - children[i] = nilValueNode + if n.Children[i] == nil { + continue } + wg.Add(1) + go func(i int) { + defer wg.Done() + + h := newHasher(false) + fn.Children[i] = h.hash(n.Children[i], false) + returnHasherToPool(h) + }(i) } wg.Wait() } else { for i := 0; i < 16; i++ { if child := n.Children[i]; child != nil { - children[i] = h.hash(child, false) - } else { - children[i] = nilValueNode + fn.Children[i] = h.hash(child, false) } } } if n.Children[16] != nil { - children[16] = n.Children[16] + fn.Children[16] = n.Children[16].(valueNode) } - return &fullNode{flags: nodeFlag{}, Children: children} -} + fn.encode(h.encbuf) + fnEncoderPool.Put(fn) -// shortNodeToHash computes the hash of the given shortNode. The shortNode must -// first be collapsed, with its key converted to compact form. If the RLP-encoded -// node data is smaller than 32 bytes, the node itself is returned. -func (h *hasher) shortnodeToHash(n *shortNode, force bool) node { - n.encode(h.encbuf) - enc := h.encodedBytes() - - if len(enc) < 32 && !force { - return n // Nodes smaller than 32 bytes are stored inside their parent - } - return h.hashData(enc) -} - -// fullnodeToHash computes the hash of the given fullNode. If the RLP-encoded -// node data is smaller than 32 bytes, the node itself is returned. -func (h *hasher) fullnodeToHash(n *fullNode, force bool) node { - n.encode(h.encbuf) - enc := h.encodedBytes() - - if len(enc) < 32 && !force { - return n // Nodes smaller than 32 bytes are stored inside their parent - } - return h.hashData(enc) + return h.encodedBytes() } // encodedBytes returns the result of the last encoding operation on h.encbuf. @@ -175,9 +186,10 @@ func (h *hasher) encodedBytes() []byte { return h.tmp } -// hashData hashes the provided data -func (h *hasher) hashData(data []byte) hashNode { - n := make(hashNode, 32) +// hashData hashes the provided data. It is safe to modify the returned slice after +// the function returns. +func (h *hasher) hashData(data []byte) []byte { + n := make([]byte, 32) h.sha.Reset() h.sha.Write(data) h.sha.Read(n) @@ -192,20 +204,17 @@ func (h *hasher) hashDataTo(dst, data []byte) { h.sha.Read(dst) } -// proofHash is used to construct trie proofs, and returns the 'collapsed' -// node (for later RLP encoding) as well as the hashed node -- unless the -// node is smaller than 32 bytes, in which case it will be returned as is. -// This method does not do anything on value- or hash-nodes. -func (h *hasher) proofHash(original node) (collapsed, hashed node) { +// proofHash is used to construct trie proofs, returning the rlp-encoded node blobs. +// Note, only resolved node (shortNode or fullNode) is expected for proofing. +// +// It is safe to modify the returned slice after the function returns. +func (h *hasher) proofHash(original node) []byte { switch n := original.(type) { case *shortNode: - sn := h.hashShortNodeChildren(n) - return sn, h.shortnodeToHash(sn, false) + return bytes.Clone(h.encodeShortNode(n)) case *fullNode: - fn := h.hashFullNodeChildren(n) - return fn, h.fullnodeToHash(fn, false) + return bytes.Clone(h.encodeFullNode(n)) default: - // Value and hash nodes don't have children, so they're left as were - return n, n + panic(fmt.Errorf("unexpected node type, %T", original)) } } diff --git a/trie/iterator.go b/trie/iterator.go index fa01611063..e6fedf2430 100644 --- a/trie/iterator.go +++ b/trie/iterator.go @@ -240,9 +240,9 @@ func (it *nodeIterator) LeafProof() [][]byte { for i, item := range it.stack[:len(it.stack)-1] { // Gather nodes that end up as hash nodes (or the root) - node, hashed := hasher.proofHash(item.node) - if _, ok := hashed.(hashNode); ok || i == 0 { - proofs = append(proofs, nodeToBytes(node)) + enc := hasher.proofHash(item.node) + if len(enc) >= 32 || i == 0 { + proofs = append(proofs, enc) } } return proofs diff --git a/trie/node.go b/trie/node.go index 96f077ebbb..74fac4fd4e 100644 --- a/trie/node.go +++ b/trie/node.go @@ -68,10 +68,6 @@ type ( } ) -// nilValueNode is used when collapsing internal trie nodes for hashing, since -// unset children need to serialize correctly. -var nilValueNode = valueNode(nil) - // EncodeRLP encodes a full node into the consensus RLP format. func (n *fullNode) EncodeRLP(w io.Writer) error { eb := rlp.NewEncoderBuffer(w) diff --git a/trie/node_enc.go b/trie/node_enc.go index c95587eeab..02b93ee6f3 100644 --- a/trie/node_enc.go +++ b/trie/node_enc.go @@ -42,18 +42,29 @@ func (n *fullNode) encode(w rlp.EncoderBuffer) { func (n *fullnodeEncoder) encode(w rlp.EncoderBuffer) { offset := w.List() - for _, c := range n.Children { - if c == nil { + for i, c := range n.Children { + if len(c) == 0 { w.Write(rlp.EmptyString) - } else if len(c) < 32 { - w.Write(c) // rawNode } else { - w.WriteBytes(c) // hashNode + // valueNode or hashNode + if i == 16 || len(c) >= 32 { + w.WriteBytes(c) + } else { + w.Write(c) // rawNode + } } } w.ListEnd(offset) } +func (n *fullnodeEncoder) reset() { + for i, c := range n.Children { + if len(c) != 0 { + n.Children[i] = n.Children[i][:0] + } + } +} + func (n *shortNode) encode(w rlp.EncoderBuffer) { offset := w.List() w.WriteBytes(n.Key) @@ -70,7 +81,7 @@ func (n *extNodeEncoder) encode(w rlp.EncoderBuffer) { w.WriteBytes(n.Key) if n.Val == nil { - w.Write(rlp.EmptyString) + w.Write(rlp.EmptyString) // theoretically impossible to happen } else if len(n.Val) < 32 { w.Write(n.Val) // rawNode } else { diff --git a/trie/proof.go b/trie/proof.go index 751d6f620f..53b7acc30c 100644 --- a/trie/proof.go +++ b/trie/proof.go @@ -22,6 +22,7 @@ import ( "fmt" "github.com/ethereum/go-ethereum/common" + "github.com/ethereum/go-ethereum/crypto" "github.com/ethereum/go-ethereum/ethdb" "github.com/ethereum/go-ethereum/log" ) @@ -85,16 +86,9 @@ func (t *Trie) Prove(key []byte, proofDb ethdb.KeyValueWriter) error { defer returnHasherToPool(hasher) for i, n := range nodes { - var hn node - n, hn = hasher.proofHash(n) - if hash, ok := hn.(hashNode); ok || i == 0 { - // If the node's database encoding is a hash (or is the - // root node), it becomes a proof element. - enc := nodeToBytes(n) - if !ok { - hash = hasher.hashData(enc) - } - proofDb.Put(hash, enc) + enc := hasher.proofHash(n) + if len(enc) >= 32 || i == 0 { + proofDb.Put(crypto.Keccak256(enc), enc) } } return nil diff --git a/trie/trie.go b/trie/trie.go index fdb4da9be4..222bf8b1f0 100644 --- a/trie/trie.go +++ b/trie/trie.go @@ -626,7 +626,7 @@ func (t *Trie) resolveAndTrack(n hashNode, prefix []byte) (node, error) { // Hash returns the root hash of the trie. It does not write to the // database and can be used even if the trie doesn't have one. func (t *Trie) Hash() common.Hash { - return common.BytesToHash(t.hashRoot().(hashNode)) + return common.BytesToHash(t.hashRoot()) } // Commit collects all dirty nodes in the trie and replaces them with the @@ -677,9 +677,9 @@ func (t *Trie) Commit(collectLeaf bool) (common.Hash, *trienode.NodeSet) { } // hashRoot calculates the root hash of the given trie -func (t *Trie) hashRoot() node { +func (t *Trie) hashRoot() []byte { if t.root == nil { - return hashNode(types.EmptyRootHash.Bytes()) + return types.EmptyRootHash.Bytes() } // If the number of changes is below 100, we let one thread handle it h := newHasher(t.unhashed >= 100) diff --git a/trie/trie_test.go b/trie/trie_test.go index b806ae6b0c..edd85677fe 100644 --- a/trie/trie_test.go +++ b/trie/trie_test.go @@ -863,7 +863,6 @@ func (s *spongeDb) Flush() { s.sponge.Write([]byte(key)) s.sponge.Write([]byte(s.values[key])) } - fmt.Println(len(s.keys)) } // spongeBatch is a dummy batch which immediately writes to the underlying spongedb