diff --git a/.github/workflows/validate_pr.yml b/.github/workflows/validate_pr.yml index 57e8c12b5e..aa3c74cc67 100644 --- a/.github/workflows/validate_pr.yml +++ b/.github/workflows/validate_pr.yml @@ -8,6 +8,45 @@ jobs: validate-pr: runs-on: ubuntu-latest steps: + - name: Check for Spam PR + uses: actions/github-script@v7 + with: + script: | + const prTitle = context.payload.pull_request.title; + const spamRegex = /^(feat|chore|fix)(\(.*\))?\s*:/i; + + if (spamRegex.test(prTitle)) { + // Leave a comment explaining why + await github.rest.issues.createComment({ + owner: context.repo.owner, + repo: context.repo.repo, + issue_number: context.payload.pull_request.number, + body: `## PR Closed as Spam + + This PR was automatically closed because the title format \`feat:\`, \`fix:\`, or \`chore:\` is commonly associated with spam contributions. + + If this is a legitimate contribution, please: + 1. Review our contribution guidelines + 2. Use the correct PR title format: \`directory, ...: description\` + 3. Open a new PR with the proper title format + + Thank you for your understanding.` + }); + + // Close the PR + await github.rest.pulls.update({ + owner: context.repo.owner, + repo: context.repo.repo, + pull_number: context.payload.pull_request.number, + state: 'closed' + }); + + core.setFailed('PR closed as spam due to suspicious title format'); + return; + } + + console.log('✅ PR passed spam check'); + - name: Checkout repository uses: actions/checkout@v4 diff --git a/beacon/engine/gen_ed.go b/beacon/engine/gen_ed.go index 0ae5a3b8f1..6893d64a16 100644 --- a/beacon/engine/gen_ed.go +++ b/beacon/engine/gen_ed.go @@ -17,24 +17,23 @@ var _ = (*executableDataMarshaling)(nil) // MarshalJSON marshals as JSON. func (e ExecutableData) MarshalJSON() ([]byte, error) { type ExecutableData struct { - ParentHash common.Hash `json:"parentHash" gencodec:"required"` - FeeRecipient common.Address `json:"feeRecipient" gencodec:"required"` - StateRoot common.Hash `json:"stateRoot" gencodec:"required"` - ReceiptsRoot common.Hash `json:"receiptsRoot" gencodec:"required"` - LogsBloom hexutil.Bytes `json:"logsBloom" gencodec:"required"` - Random common.Hash `json:"prevRandao" gencodec:"required"` - Number hexutil.Uint64 `json:"blockNumber" gencodec:"required"` - GasLimit hexutil.Uint64 `json:"gasLimit" gencodec:"required"` - GasUsed hexutil.Uint64 `json:"gasUsed" gencodec:"required"` - Timestamp hexutil.Uint64 `json:"timestamp" gencodec:"required"` - ExtraData hexutil.Bytes `json:"extraData" gencodec:"required"` - BaseFeePerGas *hexutil.Big `json:"baseFeePerGas" gencodec:"required"` - BlockHash common.Hash `json:"blockHash" gencodec:"required"` - Transactions []hexutil.Bytes `json:"transactions" gencodec:"required"` - Withdrawals []*types.Withdrawal `json:"withdrawals"` - BlobGasUsed *hexutil.Uint64 `json:"blobGasUsed"` - ExcessBlobGas *hexutil.Uint64 `json:"excessBlobGas"` - ExecutionWitness *types.ExecutionWitness `json:"executionWitness,omitempty"` + ParentHash common.Hash `json:"parentHash" gencodec:"required"` + FeeRecipient common.Address `json:"feeRecipient" gencodec:"required"` + StateRoot common.Hash `json:"stateRoot" gencodec:"required"` + ReceiptsRoot common.Hash `json:"receiptsRoot" gencodec:"required"` + LogsBloom hexutil.Bytes `json:"logsBloom" gencodec:"required"` + Random common.Hash `json:"prevRandao" gencodec:"required"` + Number hexutil.Uint64 `json:"blockNumber" gencodec:"required"` + GasLimit hexutil.Uint64 `json:"gasLimit" gencodec:"required"` + GasUsed hexutil.Uint64 `json:"gasUsed" gencodec:"required"` + Timestamp hexutil.Uint64 `json:"timestamp" gencodec:"required"` + ExtraData hexutil.Bytes `json:"extraData" gencodec:"required"` + BaseFeePerGas *hexutil.Big `json:"baseFeePerGas" gencodec:"required"` + BlockHash common.Hash `json:"blockHash" gencodec:"required"` + Transactions []hexutil.Bytes `json:"transactions" gencodec:"required"` + Withdrawals []*types.Withdrawal `json:"withdrawals"` + BlobGasUsed *hexutil.Uint64 `json:"blobGasUsed"` + ExcessBlobGas *hexutil.Uint64 `json:"excessBlobGas"` } var enc ExecutableData enc.ParentHash = e.ParentHash @@ -59,31 +58,29 @@ func (e ExecutableData) MarshalJSON() ([]byte, error) { enc.Withdrawals = e.Withdrawals enc.BlobGasUsed = (*hexutil.Uint64)(e.BlobGasUsed) enc.ExcessBlobGas = (*hexutil.Uint64)(e.ExcessBlobGas) - enc.ExecutionWitness = e.ExecutionWitness return json.Marshal(&enc) } // UnmarshalJSON unmarshals from JSON. func (e *ExecutableData) UnmarshalJSON(input []byte) error { type ExecutableData struct { - ParentHash *common.Hash `json:"parentHash" gencodec:"required"` - FeeRecipient *common.Address `json:"feeRecipient" gencodec:"required"` - StateRoot *common.Hash `json:"stateRoot" gencodec:"required"` - ReceiptsRoot *common.Hash `json:"receiptsRoot" gencodec:"required"` - LogsBloom *hexutil.Bytes `json:"logsBloom" gencodec:"required"` - Random *common.Hash `json:"prevRandao" gencodec:"required"` - Number *hexutil.Uint64 `json:"blockNumber" gencodec:"required"` - GasLimit *hexutil.Uint64 `json:"gasLimit" gencodec:"required"` - GasUsed *hexutil.Uint64 `json:"gasUsed" gencodec:"required"` - Timestamp *hexutil.Uint64 `json:"timestamp" gencodec:"required"` - ExtraData *hexutil.Bytes `json:"extraData" gencodec:"required"` - BaseFeePerGas *hexutil.Big `json:"baseFeePerGas" gencodec:"required"` - BlockHash *common.Hash `json:"blockHash" gencodec:"required"` - Transactions []hexutil.Bytes `json:"transactions" gencodec:"required"` - Withdrawals []*types.Withdrawal `json:"withdrawals"` - BlobGasUsed *hexutil.Uint64 `json:"blobGasUsed"` - ExcessBlobGas *hexutil.Uint64 `json:"excessBlobGas"` - ExecutionWitness *types.ExecutionWitness `json:"executionWitness,omitempty"` + ParentHash *common.Hash `json:"parentHash" gencodec:"required"` + FeeRecipient *common.Address `json:"feeRecipient" gencodec:"required"` + StateRoot *common.Hash `json:"stateRoot" gencodec:"required"` + ReceiptsRoot *common.Hash `json:"receiptsRoot" gencodec:"required"` + LogsBloom *hexutil.Bytes `json:"logsBloom" gencodec:"required"` + Random *common.Hash `json:"prevRandao" gencodec:"required"` + Number *hexutil.Uint64 `json:"blockNumber" gencodec:"required"` + GasLimit *hexutil.Uint64 `json:"gasLimit" gencodec:"required"` + GasUsed *hexutil.Uint64 `json:"gasUsed" gencodec:"required"` + Timestamp *hexutil.Uint64 `json:"timestamp" gencodec:"required"` + ExtraData *hexutil.Bytes `json:"extraData" gencodec:"required"` + BaseFeePerGas *hexutil.Big `json:"baseFeePerGas" gencodec:"required"` + BlockHash *common.Hash `json:"blockHash" gencodec:"required"` + Transactions []hexutil.Bytes `json:"transactions" gencodec:"required"` + Withdrawals []*types.Withdrawal `json:"withdrawals"` + BlobGasUsed *hexutil.Uint64 `json:"blobGasUsed"` + ExcessBlobGas *hexutil.Uint64 `json:"excessBlobGas"` } var dec ExecutableData if err := json.Unmarshal(input, &dec); err != nil { @@ -157,8 +154,5 @@ func (e *ExecutableData) UnmarshalJSON(input []byte) error { if dec.ExcessBlobGas != nil { e.ExcessBlobGas = (*uint64)(dec.ExcessBlobGas) } - if dec.ExecutionWitness != nil { - e.ExecutionWitness = dec.ExecutionWitness - } return nil } diff --git a/beacon/engine/types.go b/beacon/engine/types.go index ddb276ab09..da9b6568f2 100644 --- a/beacon/engine/types.go +++ b/beacon/engine/types.go @@ -73,24 +73,23 @@ type payloadAttributesMarshaling struct { // ExecutableData is the data necessary to execute an EL payload. type ExecutableData struct { - ParentHash common.Hash `json:"parentHash" gencodec:"required"` - FeeRecipient common.Address `json:"feeRecipient" gencodec:"required"` - StateRoot common.Hash `json:"stateRoot" gencodec:"required"` - ReceiptsRoot common.Hash `json:"receiptsRoot" gencodec:"required"` - LogsBloom []byte `json:"logsBloom" gencodec:"required"` - Random common.Hash `json:"prevRandao" gencodec:"required"` - Number uint64 `json:"blockNumber" gencodec:"required"` - GasLimit uint64 `json:"gasLimit" gencodec:"required"` - GasUsed uint64 `json:"gasUsed" gencodec:"required"` - Timestamp uint64 `json:"timestamp" gencodec:"required"` - ExtraData []byte `json:"extraData" gencodec:"required"` - BaseFeePerGas *big.Int `json:"baseFeePerGas" gencodec:"required"` - BlockHash common.Hash `json:"blockHash" gencodec:"required"` - Transactions [][]byte `json:"transactions" gencodec:"required"` - Withdrawals []*types.Withdrawal `json:"withdrawals"` - BlobGasUsed *uint64 `json:"blobGasUsed"` - ExcessBlobGas *uint64 `json:"excessBlobGas"` - ExecutionWitness *types.ExecutionWitness `json:"executionWitness,omitempty"` + ParentHash common.Hash `json:"parentHash" gencodec:"required"` + FeeRecipient common.Address `json:"feeRecipient" gencodec:"required"` + StateRoot common.Hash `json:"stateRoot" gencodec:"required"` + ReceiptsRoot common.Hash `json:"receiptsRoot" gencodec:"required"` + LogsBloom []byte `json:"logsBloom" gencodec:"required"` + Random common.Hash `json:"prevRandao" gencodec:"required"` + Number uint64 `json:"blockNumber" gencodec:"required"` + GasLimit uint64 `json:"gasLimit" gencodec:"required"` + GasUsed uint64 `json:"gasUsed" gencodec:"required"` + Timestamp uint64 `json:"timestamp" gencodec:"required"` + ExtraData []byte `json:"extraData" gencodec:"required"` + BaseFeePerGas *big.Int `json:"baseFeePerGas" gencodec:"required"` + BlockHash common.Hash `json:"blockHash" gencodec:"required"` + Transactions [][]byte `json:"transactions" gencodec:"required"` + Withdrawals []*types.Withdrawal `json:"withdrawals"` + BlobGasUsed *uint64 `json:"blobGasUsed"` + ExcessBlobGas *uint64 `json:"excessBlobGas"` } // JSON type overrides for executableData. @@ -316,8 +315,7 @@ func ExecutableDataToBlockNoHash(data ExecutableData, versionedHashes []common.H RequestsHash: requestsHash, } return types.NewBlockWithHeader(header). - WithBody(types.Body{Transactions: txs, Uncles: nil, Withdrawals: data.Withdrawals}). - WithWitness(data.ExecutionWitness), + WithBody(types.Body{Transactions: txs, Uncles: nil, Withdrawals: data.Withdrawals}), nil } @@ -325,24 +323,23 @@ func ExecutableDataToBlockNoHash(data ExecutableData, versionedHashes []common.H // fields from the given block. It assumes the given block is post-merge block. func BlockToExecutableData(block *types.Block, fees *big.Int, sidecars []*types.BlobTxSidecar, requests [][]byte) *ExecutionPayloadEnvelope { data := &ExecutableData{ - BlockHash: block.Hash(), - ParentHash: block.ParentHash(), - FeeRecipient: block.Coinbase(), - StateRoot: block.Root(), - Number: block.NumberU64(), - GasLimit: block.GasLimit(), - GasUsed: block.GasUsed(), - BaseFeePerGas: block.BaseFee(), - Timestamp: block.Time(), - ReceiptsRoot: block.ReceiptHash(), - LogsBloom: block.Bloom().Bytes(), - Transactions: encodeTransactions(block.Transactions()), - Random: block.MixDigest(), - ExtraData: block.Extra(), - Withdrawals: block.Withdrawals(), - BlobGasUsed: block.BlobGasUsed(), - ExcessBlobGas: block.ExcessBlobGas(), - ExecutionWitness: block.ExecutionWitness(), + BlockHash: block.Hash(), + ParentHash: block.ParentHash(), + FeeRecipient: block.Coinbase(), + StateRoot: block.Root(), + Number: block.NumberU64(), + GasLimit: block.GasLimit(), + GasUsed: block.GasUsed(), + BaseFeePerGas: block.BaseFee(), + Timestamp: block.Time(), + ReceiptsRoot: block.ReceiptHash(), + LogsBloom: block.Bloom().Bytes(), + Transactions: encodeTransactions(block.Transactions()), + Random: block.MixDigest(), + ExtraData: block.Extra(), + Withdrawals: block.Withdrawals(), + BlobGasUsed: block.BlobGasUsed(), + ExcessBlobGas: block.ExcessBlobGas(), } // Add blobs. diff --git a/beacon/light/sync/head_sync.go b/beacon/light/sync/head_sync.go index 5e41258053..7189767d9c 100644 --- a/beacon/light/sync/head_sync.go +++ b/beacon/light/sync/head_sync.go @@ -105,6 +105,7 @@ func (s *HeadSync) Process(requester request.Requester, events []request.Event) delete(s.serverHeads, event.Server) delete(s.unvalidatedOptimistic, event.Server) delete(s.unvalidatedFinality, event.Server) + delete(s.reqFinalityEpoch, event.Server) } } } diff --git a/cmd/geth/main.go b/cmd/geth/main.go index b294ee593e..96f9f58dde 100644 --- a/cmd/geth/main.go +++ b/cmd/geth/main.go @@ -251,8 +251,6 @@ func init() { utils.ShowDeprecated, // See snapshot.go snapshotCommand, - // See verkle.go - verkleCommand, } if logTestCommand != nil { app.Commands = append(app.Commands, logTestCommand) diff --git a/cmd/geth/verkle.go b/cmd/geth/verkle.go deleted file mode 100644 index c064d70aba..0000000000 --- a/cmd/geth/verkle.go +++ /dev/null @@ -1,214 +0,0 @@ -// Copyright 2022 The go-ethereum Authors -// This file is part of go-ethereum. -// -// go-ethereum is free software: you can redistribute it and/or modify -// it under the terms of the GNU General Public License as published by -// the Free Software Foundation, either version 3 of the License, or -// (at your option) any later version. -// -// go-ethereum 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 General Public License for more details. -// -// You should have received a copy of the GNU General Public License -// along with go-ethereum. If not, see . - -package main - -import ( - "bytes" - "encoding/hex" - "errors" - "fmt" - "os" - "slices" - - "github.com/ethereum/go-ethereum/cmd/utils" - "github.com/ethereum/go-ethereum/common" - "github.com/ethereum/go-ethereum/core/rawdb" - "github.com/ethereum/go-ethereum/log" - "github.com/ethereum/go-verkle" - "github.com/urfave/cli/v2" -) - -var ( - zero [32]byte - - verkleCommand = &cli.Command{ - Name: "verkle", - Usage: "A set of experimental verkle tree management commands", - Description: "", - Subcommands: []*cli.Command{ - { - Name: "verify", - Usage: "verify the conversion of a MPT into a verkle tree", - ArgsUsage: "", - Action: verifyVerkle, - Flags: slices.Concat(utils.NetworkFlags, utils.DatabaseFlags), - Description: ` -geth verkle verify -This command takes a root commitment and attempts to rebuild the tree. - `, - }, - { - Name: "dump", - Usage: "Dump a verkle tree to a DOT file", - ArgsUsage: " [ ...]", - Action: expandVerkle, - Flags: slices.Concat(utils.NetworkFlags, utils.DatabaseFlags), - Description: ` -geth verkle dump [ ...] -This command will produce a dot file representing the tree, rooted at . -in which key1, key2, ... are expanded. - `, - }, - }, - } -) - -// recurse into each child to ensure they can be loaded from the db. The tree isn't rebuilt -// (only its nodes are loaded) so there is no need to flush them, the garbage collector should -// take care of that for us. -func checkChildren(root verkle.VerkleNode, resolver verkle.NodeResolverFn) error { - switch node := root.(type) { - case *verkle.InternalNode: - for i, child := range node.Children() { - childC := child.Commit().Bytes() - - if bytes.Equal(childC[:], zero[:]) { - continue - } - childS, err := resolver(childC[:]) - if err != nil { - return fmt.Errorf("could not find child %x in db: %w", childC, err) - } - // depth is set to 0, the tree isn't rebuilt so it's not a problem - childN, err := verkle.ParseNode(childS, 0) - if err != nil { - return fmt.Errorf("decode error child %x in db: %w", child.Commitment().Bytes(), err) - } - if err := checkChildren(childN, resolver); err != nil { - return fmt.Errorf("%x%w", i, err) // write the path to the erroring node - } - } - case *verkle.LeafNode: - // sanity check: ensure at least one value is non-zero - - for i := 0; i < verkle.NodeWidth; i++ { - if len(node.Value(i)) != 0 { - return nil - } - } - return errors.New("both balance and nonce are 0") - case verkle.Empty: - // nothing to do - default: - return fmt.Errorf("unsupported type encountered %v", root) - } - - return nil -} - -func verifyVerkle(ctx *cli.Context) error { - stack, _ := makeConfigNode(ctx) - defer stack.Close() - - chaindb := utils.MakeChainDatabase(ctx, stack, true) - defer chaindb.Close() - headBlock := rawdb.ReadHeadBlock(chaindb) - if headBlock == nil { - log.Error("Failed to load head block") - return errors.New("no head block") - } - if ctx.NArg() > 1 { - log.Error("Too many arguments given") - return errors.New("too many arguments") - } - var ( - rootC common.Hash - err error - ) - if ctx.NArg() == 1 { - rootC, err = parseRoot(ctx.Args().First()) - if err != nil { - log.Error("Failed to resolve state root", "error", err) - return err - } - log.Info("Rebuilding the tree", "root", rootC) - } else { - rootC = headBlock.Root() - log.Info("Rebuilding the tree", "root", rootC, "number", headBlock.NumberU64()) - } - - serializedRoot, err := chaindb.Get(rootC[:]) - if err != nil { - return err - } - root, err := verkle.ParseNode(serializedRoot, 0) - if err != nil { - return err - } - - if err := checkChildren(root, chaindb.Get); err != nil { - log.Error("Could not rebuild the tree from the database", "err", err) - return err - } - - log.Info("Tree was rebuilt from the database") - return nil -} - -func expandVerkle(ctx *cli.Context) error { - stack, _ := makeConfigNode(ctx) - defer stack.Close() - - chaindb := utils.MakeChainDatabase(ctx, stack, true) - defer chaindb.Close() - var ( - rootC common.Hash - keylist [][]byte - err error - ) - if ctx.NArg() >= 2 { - rootC, err = parseRoot(ctx.Args().First()) - if err != nil { - log.Error("Failed to resolve state root", "error", err) - return err - } - keylist = make([][]byte, 0, ctx.Args().Len()-1) - args := ctx.Args().Slice() - for i := range args[1:] { - key, err := hex.DecodeString(args[i+1]) - log.Info("decoded key", "arg", args[i+1], "key", key) - if err != nil { - return fmt.Errorf("error decoding key #%d: %w", i+1, err) - } - keylist = append(keylist, key) - } - log.Info("Rebuilding the tree", "root", rootC) - } else { - return fmt.Errorf("usage: %s root key1 [key 2...]", ctx.App.Name) - } - - serializedRoot, err := chaindb.Get(rootC[:]) - if err != nil { - return err - } - root, err := verkle.ParseNode(serializedRoot, 0) - if err != nil { - return err - } - - for i, key := range keylist { - log.Info("Reading key", "index", i, "key", key) - root.Get(key, chaindb.Get) - } - - if err := os.WriteFile("dump.dot", []byte(verkle.ToDot(root)), 0600); err != nil { - log.Error("Failed to dump file", "err", err) - } else { - log.Info("Tree was dumped to file", "file", "dump.dot") - } - return nil -} diff --git a/cmd/keeper/go.mod b/cmd/keeper/go.mod index 8402382a9b..a42be042aa 100644 --- a/cmd/keeper/go.mod +++ b/cmd/keeper/go.mod @@ -14,13 +14,11 @@ require ( github.com/cespare/xxhash/v2 v2.3.0 // indirect github.com/consensys/gnark-crypto v0.18.1 // indirect github.com/crate-crypto/go-eth-kzg v1.4.0 // indirect - github.com/crate-crypto/go-ipa v0.0.0-20240724233137-53bbb0ceb27a // indirect github.com/deckarep/golang-set/v2 v2.6.0 // indirect github.com/decred/dcrd/dcrec/secp256k1/v4 v4.0.1 // indirect github.com/emicklei/dot v1.6.2 // indirect github.com/ethereum/c-kzg-4844/v2 v2.1.5 // indirect github.com/ethereum/go-bigmodexpfix v0.0.0-20250911101455-f9e208c548ab // indirect - github.com/ethereum/go-verkle v0.2.2 // indirect github.com/ferranbt/fastssz v0.1.4 // indirect github.com/go-ole/go-ole v1.3.0 // indirect github.com/gofrs/flock v0.12.1 // indirect diff --git a/cmd/keeper/go.sum b/cmd/keeper/go.sum index 4f4c0dbba0..133a3b10b1 100644 --- a/cmd/keeper/go.sum +++ b/cmd/keeper/go.sum @@ -30,8 +30,6 @@ github.com/consensys/gnark-crypto v0.18.1 h1:RyLV6UhPRoYYzaFnPQA4qK3DyuDgkTgskDd github.com/consensys/gnark-crypto v0.18.1/go.mod h1:L3mXGFTe1ZN+RSJ+CLjUt9x7PNdx8ubaYfDROyp2Z8c= github.com/crate-crypto/go-eth-kzg v1.4.0 h1:WzDGjHk4gFg6YzV0rJOAsTK4z3Qkz5jd4RE3DAvPFkg= github.com/crate-crypto/go-eth-kzg v1.4.0/go.mod h1:J9/u5sWfznSObptgfa92Jq8rTswn6ahQWEuiLHOjCUI= -github.com/crate-crypto/go-ipa v0.0.0-20240724233137-53bbb0ceb27a h1:W8mUrRp6NOVl3J+MYp5kPMoUZPp7aOYHtaua31lwRHg= -github.com/crate-crypto/go-ipa v0.0.0-20240724233137-53bbb0ceb27a/go.mod h1:sTwzHBvIzm2RfVCGNEBZgRyjwK40bVoun3ZnGOCafNM= github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c= github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38= github.com/deckarep/golang-set/v2 v2.6.0 h1:XfcQbWM1LlMB8BsJ8N9vW5ehnnPVIw0je80NsVHagjM= @@ -46,8 +44,6 @@ github.com/ethereum/c-kzg-4844/v2 v2.1.5 h1:aVtoLK5xwJ6c5RiqO8g8ptJ5KU+2Hdquf6G3 github.com/ethereum/c-kzg-4844/v2 v2.1.5/go.mod h1:u59hRTTah4Co6i9fDWtiCjTrblJv0UwsqZKCc0GfgUs= github.com/ethereum/go-bigmodexpfix v0.0.0-20250911101455-f9e208c548ab h1:rvv6MJhy07IMfEKuARQ9TKojGqLVNxQajaXEp/BoqSk= github.com/ethereum/go-bigmodexpfix v0.0.0-20250911101455-f9e208c548ab/go.mod h1:IuLm4IsPipXKF7CW5Lzf68PIbZ5yl7FFd74l/E0o9A8= -github.com/ethereum/go-verkle v0.2.2 h1:I2W0WjnrFUIzzVPwm8ykY+7pL2d4VhlsePn4j7cnFk8= -github.com/ethereum/go-verkle v0.2.2/go.mod h1:M3b90YRnzqKyyzBEWJGqj8Qff4IDeXnzFw0P9bFw3uk= github.com/ferranbt/fastssz v0.1.4 h1:OCDB+dYDEQDvAgtAGnTSidK1Pe2tW3nFV40XyMkTeDY= github.com/ferranbt/fastssz v0.1.4/go.mod h1:Ea3+oeoRGGLGm5shYAeDgu6PGUlcvQhE2fILyD9+tGg= github.com/getsentry/sentry-go v0.27.0 h1:Pv98CIbtB3LkMWmXi4Joa5OOcwbmnX88sF5qbK3r3Ps= diff --git a/consensus/beacon/consensus.go b/consensus/beacon/consensus.go index dbba73947f..eed27407a5 100644 --- a/consensus/beacon/consensus.go +++ b/consensus/beacon/consensus.go @@ -365,46 +365,7 @@ func (beacon *Beacon) FinalizeAndAssemble(chain consensus.ChainHeaderReader, hea header.Root = state.IntermediateRoot(true) // Assemble the final block. - block := types.NewBlock(header, body, receipts, trie.NewStackTrie(nil)) - - // Create the block witness and attach to block. - // This step needs to happen as late as possible to catch all access events. - if chain.Config().IsVerkle(header.Number, header.Time) { - keys := state.AccessEvents().Keys() - - // Open the pre-tree to prove the pre-state against - parent := chain.GetHeaderByNumber(header.Number.Uint64() - 1) - if parent == nil { - return nil, fmt.Errorf("nil parent header for block %d", header.Number) - } - preTrie, err := state.Database().OpenTrie(parent.Root) - if err != nil { - return nil, fmt.Errorf("error opening pre-state tree root: %w", err) - } - postTrie := state.GetTrie() - if postTrie == nil { - return nil, errors.New("post-state tree is not available") - } - vktPreTrie, okpre := preTrie.(*trie.VerkleTrie) - vktPostTrie, okpost := postTrie.(*trie.VerkleTrie) - - // The witness is only attached iff both parent and current block are - // using verkle tree. - if okpre && okpost { - if len(keys) > 0 { - verkleProof, stateDiff, err := vktPreTrie.Proof(vktPostTrie, keys) - if err != nil { - return nil, fmt.Errorf("error generating verkle proof for block %d: %w", header.Number, err) - } - block = block.WithWitness(&types.ExecutionWitness{ - StateDiff: stateDiff, - VerkleProof: verkleProof, - }) - } - } - } - - return block, nil + return types.NewBlock(header, body, receipts, trie.NewStackTrie(nil)), nil } // Seal generates a new sealing request for the given input block and pushes diff --git a/console/prompt/prompter.go b/console/prompt/prompter.go index 2a20b6906a..5a0a89e76a 100644 --- a/console/prompt/prompter.go +++ b/console/prompt/prompter.go @@ -142,7 +142,7 @@ func (p *terminalPrompter) PromptPassword(prompt string) (passwd string, err err // PromptConfirm displays the given prompt to the user and requests a boolean // choice to be made, returning that choice. func (p *terminalPrompter) PromptConfirm(prompt string) (bool, error) { - input, err := p.Prompt(prompt + " [y/n] ") + input, err := p.PromptInput(prompt + " [y/n] ") if len(input) > 0 && strings.EqualFold(input[:1], "y") { return true, nil } diff --git a/core/state/access_events.go b/core/state/access_events.go index 0575c9898a..86f44bd623 100644 --- a/core/state/access_events.go +++ b/core/state/access_events.go @@ -23,7 +23,7 @@ import ( "github.com/ethereum/go-ethereum/common" "github.com/ethereum/go-ethereum/common/math" "github.com/ethereum/go-ethereum/params" - "github.com/ethereum/go-ethereum/trie/utils" + "github.com/ethereum/go-ethereum/trie/bintrie" "github.com/holiman/uint256" ) @@ -45,15 +45,12 @@ var zeroTreeIndex uint256.Int type AccessEvents struct { branches map[branchAccessKey]mode chunks map[chunkAccessKey]mode - - pointCache *utils.PointCache } -func NewAccessEvents(pointCache *utils.PointCache) *AccessEvents { +func NewAccessEvents() *AccessEvents { return &AccessEvents{ - branches: make(map[branchAccessKey]mode), - chunks: make(map[chunkAccessKey]mode), - pointCache: pointCache, + branches: make(map[branchAccessKey]mode), + chunks: make(map[chunkAccessKey]mode), } } @@ -75,8 +72,11 @@ func (ae *AccessEvents) Keys() [][]byte { // TODO: consider if parallelizing this is worth it, probably depending on len(ae.chunks). keys := make([][]byte, 0, len(ae.chunks)) for chunk := range ae.chunks { - basePoint := ae.pointCache.Get(chunk.addr[:]) - key := utils.GetTreeKeyWithEvaluatedAddress(basePoint, &chunk.treeIndex, chunk.leafKey) + var offset [32]byte + treeIndexBytes := chunk.treeIndex.Bytes32() + copy(offset[:31], treeIndexBytes[1:]) + offset[31] = chunk.leafKey + key := bintrie.GetBinaryTreeKey(chunk.addr, offset[:]) keys = append(keys, key) } return keys @@ -84,9 +84,8 @@ func (ae *AccessEvents) Keys() [][]byte { func (ae *AccessEvents) Copy() *AccessEvents { cpy := &AccessEvents{ - branches: maps.Clone(ae.branches), - chunks: maps.Clone(ae.chunks), - pointCache: ae.pointCache, + branches: maps.Clone(ae.branches), + chunks: maps.Clone(ae.chunks), } return cpy } @@ -95,12 +94,12 @@ func (ae *AccessEvents) Copy() *AccessEvents { // member fields of an account. func (ae *AccessEvents) AddAccount(addr common.Address, isWrite bool, availableGas uint64) uint64 { var gas uint64 // accumulate the consumed gas - consumed, expected := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.BasicDataLeafKey, isWrite, availableGas) + consumed, expected := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.BasicDataLeafKey, isWrite, availableGas) if consumed < expected { return expected } gas += consumed - consumed, expected = ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.CodeHashLeafKey, isWrite, availableGas-consumed) + consumed, expected = ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.CodeHashLeafKey, isWrite, availableGas-consumed) if consumed < expected { return expected + gas } @@ -112,7 +111,7 @@ func (ae *AccessEvents) AddAccount(addr common.Address, isWrite bool, availableG // cold member fields of an account, that need to be touched when making a message // call to that account. func (ae *AccessEvents) MessageCallGas(destination common.Address, availableGas uint64) uint64 { - _, expected := ae.touchAddressAndChargeGas(destination, zeroTreeIndex, utils.BasicDataLeafKey, false, availableGas) + _, expected := ae.touchAddressAndChargeGas(destination, zeroTreeIndex, bintrie.BasicDataLeafKey, false, availableGas) if expected == 0 { expected = params.WarmStorageReadCostEIP2929 } @@ -122,11 +121,11 @@ func (ae *AccessEvents) MessageCallGas(destination common.Address, availableGas // ValueTransferGas returns the gas to be charged for each of the currently // cold balance member fields of the caller and the callee accounts. func (ae *AccessEvents) ValueTransferGas(callerAddr, targetAddr common.Address, availableGas uint64) uint64 { - _, expected1 := ae.touchAddressAndChargeGas(callerAddr, zeroTreeIndex, utils.BasicDataLeafKey, true, availableGas) + _, expected1 := ae.touchAddressAndChargeGas(callerAddr, zeroTreeIndex, bintrie.BasicDataLeafKey, true, availableGas) if expected1 > availableGas { return expected1 } - _, expected2 := ae.touchAddressAndChargeGas(targetAddr, zeroTreeIndex, utils.BasicDataLeafKey, true, availableGas-expected1) + _, expected2 := ae.touchAddressAndChargeGas(targetAddr, zeroTreeIndex, bintrie.BasicDataLeafKey, true, availableGas-expected1) if expected1+expected2 == 0 { return params.WarmStorageReadCostEIP2929 } @@ -138,8 +137,8 @@ func (ae *AccessEvents) ValueTransferGas(callerAddr, targetAddr common.Address, // address collision is done before the transfer, and so no write // are guaranteed to happen at this point. func (ae *AccessEvents) ContractCreatePreCheckGas(addr common.Address, availableGas uint64) uint64 { - consumed, expected1 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.BasicDataLeafKey, false, availableGas) - _, expected2 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.CodeHashLeafKey, false, availableGas-consumed) + consumed, expected1 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.BasicDataLeafKey, false, availableGas) + _, expected2 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.CodeHashLeafKey, false, availableGas-consumed) return expected1 + expected2 } @@ -147,9 +146,9 @@ func (ae *AccessEvents) ContractCreatePreCheckGas(addr common.Address, available // a contract creation. func (ae *AccessEvents) ContractCreateInitGas(addr common.Address, availableGas uint64) (uint64, uint64) { var gas uint64 - consumed, expected1 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.BasicDataLeafKey, true, availableGas) + consumed, expected1 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.BasicDataLeafKey, true, availableGas) gas += consumed - consumed, expected2 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.CodeHashLeafKey, true, availableGas-consumed) + consumed, expected2 := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.CodeHashLeafKey, true, availableGas-consumed) gas += consumed return gas, expected1 + expected2 } @@ -157,20 +156,20 @@ func (ae *AccessEvents) ContractCreateInitGas(addr common.Address, availableGas // AddTxOrigin adds the member fields of the sender account to the access event list, // so that cold accesses are not charged, since they are covered by the 21000 gas. func (ae *AccessEvents) AddTxOrigin(originAddr common.Address) { - ae.touchAddressAndChargeGas(originAddr, zeroTreeIndex, utils.BasicDataLeafKey, true, gomath.MaxUint64) - ae.touchAddressAndChargeGas(originAddr, zeroTreeIndex, utils.CodeHashLeafKey, false, gomath.MaxUint64) + ae.touchAddressAndChargeGas(originAddr, zeroTreeIndex, bintrie.BasicDataLeafKey, true, gomath.MaxUint64) + ae.touchAddressAndChargeGas(originAddr, zeroTreeIndex, bintrie.CodeHashLeafKey, false, gomath.MaxUint64) } // AddTxDestination adds the member fields of the sender account to the access event list, // so that cold accesses are not charged, since they are covered by the 21000 gas. func (ae *AccessEvents) AddTxDestination(addr common.Address, sendsValue, doesntExist bool) { - ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.BasicDataLeafKey, sendsValue, gomath.MaxUint64) - ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.CodeHashLeafKey, doesntExist, gomath.MaxUint64) + ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.BasicDataLeafKey, sendsValue, gomath.MaxUint64) + ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.CodeHashLeafKey, doesntExist, gomath.MaxUint64) } // SlotGas returns the amount of gas to be charged for a cold storage access. func (ae *AccessEvents) SlotGas(addr common.Address, slot common.Hash, isWrite bool, availableGas uint64, chargeWarmCosts bool) uint64 { - treeIndex, subIndex := utils.StorageIndex(slot.Bytes()) + treeIndex, subIndex := bintrie.StorageIndex(slot.Bytes()) _, expected := ae.touchAddressAndChargeGas(addr, *treeIndex, subIndex, isWrite, availableGas) if expected == 0 && chargeWarmCosts { expected = params.WarmStorageReadCostEIP2929 @@ -313,7 +312,7 @@ func (ae *AccessEvents) CodeChunksRangeGas(contractAddr common.Address, startPC, // Note that an access in write mode implies an access in read mode, whereas an // access in read mode does not imply an access in write mode. func (ae *AccessEvents) BasicDataGas(addr common.Address, isWrite bool, availableGas uint64, chargeWarmCosts bool) uint64 { - _, expected := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.BasicDataLeafKey, isWrite, availableGas) + _, expected := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.BasicDataLeafKey, isWrite, availableGas) if expected == 0 && chargeWarmCosts { if availableGas < params.WarmStorageReadCostEIP2929 { return availableGas @@ -329,7 +328,7 @@ func (ae *AccessEvents) BasicDataGas(addr common.Address, isWrite bool, availabl // Note that an access in write mode implies an access in read mode, whereas an access in // read mode does not imply an access in write mode. func (ae *AccessEvents) CodeHashGas(addr common.Address, isWrite bool, availableGas uint64, chargeWarmCosts bool) uint64 { - _, expected := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, utils.CodeHashLeafKey, isWrite, availableGas) + _, expected := ae.touchAddressAndChargeGas(addr, zeroTreeIndex, bintrie.CodeHashLeafKey, isWrite, availableGas) if expected == 0 && chargeWarmCosts { if availableGas < params.WarmStorageReadCostEIP2929 { return availableGas diff --git a/core/state/access_events_test.go b/core/state/access_events_test.go index e80859a0b4..0b39130e8d 100644 --- a/core/state/access_events_test.go +++ b/core/state/access_events_test.go @@ -22,7 +22,6 @@ import ( "github.com/ethereum/go-ethereum/common" "github.com/ethereum/go-ethereum/params" - "github.com/ethereum/go-ethereum/trie/utils" ) var ( @@ -38,7 +37,7 @@ func init() { } func TestAccountHeaderGas(t *testing.T) { - ae := NewAccessEvents(utils.NewPointCache(1024)) + ae := NewAccessEvents() // Check cold read cost gas := ae.BasicDataGas(testAddr, false, math.MaxUint64, false) @@ -93,7 +92,7 @@ func TestAccountHeaderGas(t *testing.T) { // TestContractCreateInitGas checks that the gas cost of contract creation is correctly // calculated. func TestContractCreateInitGas(t *testing.T) { - ae := NewAccessEvents(utils.NewPointCache(1024)) + ae := NewAccessEvents() var testAddr [20]byte for i := byte(0); i < 20; i++ { @@ -116,7 +115,7 @@ func TestContractCreateInitGas(t *testing.T) { // TestMessageCallGas checks that the gas cost of message calls is correctly // calculated. func TestMessageCallGas(t *testing.T) { - ae := NewAccessEvents(utils.NewPointCache(1024)) + ae := NewAccessEvents() // Check cold read cost, without a value gas := ae.MessageCallGas(testAddr, math.MaxUint64) diff --git a/core/state/database.go b/core/state/database.go index ae177d964f..1e8fc9d5c9 100644 --- a/core/state/database.go +++ b/core/state/database.go @@ -31,7 +31,6 @@ import ( "github.com/ethereum/go-ethereum/trie/bintrie" "github.com/ethereum/go-ethereum/trie/transitiontrie" "github.com/ethereum/go-ethereum/trie/trienode" - "github.com/ethereum/go-ethereum/trie/utils" "github.com/ethereum/go-ethereum/triedb" ) @@ -41,9 +40,6 @@ const ( // Cache size granted for caching clean code. codeCacheSize = 256 * 1024 * 1024 - - // Number of address->curve point associations to keep. - pointCacheSize = 4096 ) // Database wraps access to tries and contract code. @@ -57,9 +53,6 @@ type Database interface { // OpenStorageTrie opens the storage trie of an account. OpenStorageTrie(stateRoot common.Hash, address common.Address, root common.Hash, trie Trie) (Trie, error) - // PointCache returns the cache holding points used in verkle tree key computation - PointCache() *utils.PointCache - // TrieDB returns the underlying trie database for managing trie nodes. TrieDB() *triedb.Database @@ -161,7 +154,6 @@ type CachingDB struct { snap *snapshot.Tree codeCache *lru.SizeConstrainedCache[common.Hash, []byte] codeSizeCache *lru.Cache[common.Hash, int] - pointCache *utils.PointCache // Transition-specific fields TransitionStatePerRoot *lru.Cache[common.Hash, *overlay.TransitionState] @@ -175,7 +167,6 @@ func NewDatabase(triedb *triedb.Database, snap *snapshot.Tree) *CachingDB { snap: snap, codeCache: lru.NewSizeConstrainedCache[common.Hash, []byte](codeCacheSize), codeSizeCache: lru.NewCache[common.Hash, int](codeSizeCacheSize), - pointCache: utils.NewPointCache(pointCacheSize), TransitionStatePerRoot: lru.NewCache[common.Hash, *overlay.TransitionState](1000), } } @@ -211,7 +202,7 @@ func (db *CachingDB) Reader(stateRoot common.Hash) (Reader, error) { } // Configure the trie reader, which is expected to be available as the // gatekeeper unless the state is corrupted. - tr, err := newTrieReader(stateRoot, db.triedb, db.pointCache) + tr, err := newTrieReader(stateRoot, db.triedb) if err != nil { return nil, err } @@ -289,11 +280,6 @@ func (db *CachingDB) TrieDB() *triedb.Database { return db.triedb } -// PointCache returns the cache of evaluated curve points. -func (db *CachingDB) PointCache() *utils.PointCache { - return db.pointCache -} - // Snapshot returns the underlying state snapshot. func (db *CachingDB) Snapshot() *snapshot.Tree { return db.snap @@ -304,8 +290,6 @@ func mustCopyTrie(t Trie) Trie { switch t := t.(type) { case *trie.StateTrie: return t.Copy() - case *trie.VerkleTrie: - return t.Copy() case *transitiontrie.TransitionTrie: return t.Copy() default: diff --git a/core/state/database_history.go b/core/state/database_history.go index 314c56c470..f9c4a69f2f 100644 --- a/core/state/database_history.go +++ b/core/state/database_history.go @@ -25,7 +25,6 @@ import ( "github.com/ethereum/go-ethereum/core/types" "github.com/ethereum/go-ethereum/ethdb" "github.com/ethereum/go-ethereum/rlp" - "github.com/ethereum/go-ethereum/trie/utils" "github.com/ethereum/go-ethereum/triedb" "github.com/ethereum/go-ethereum/triedb/pathdb" ) @@ -105,7 +104,6 @@ type HistoricDB struct { triedb *triedb.Database codeCache *lru.SizeConstrainedCache[common.Hash, []byte] codeSizeCache *lru.Cache[common.Hash, int] - pointCache *utils.PointCache } // NewHistoricDatabase creates a historic state database. @@ -115,7 +113,6 @@ func NewHistoricDatabase(disk ethdb.KeyValueStore, triedb *triedb.Database) *His triedb: triedb, codeCache: lru.NewSizeConstrainedCache[common.Hash, []byte](codeCacheSize), codeSizeCache: lru.NewCache[common.Hash, int](codeSizeCacheSize), - pointCache: utils.NewPointCache(pointCacheSize), } } @@ -139,11 +136,6 @@ func (db *HistoricDB) OpenStorageTrie(stateRoot common.Hash, address common.Addr return nil, errors.New("not implemented") } -// PointCache returns the cache holding points used in verkle tree key computation -func (db *HistoricDB) PointCache() *utils.PointCache { - return db.pointCache -} - // TrieDB returns the underlying trie database for managing trie nodes. func (db *HistoricDB) TrieDB() *triedb.Database { return db.triedb diff --git a/core/state/reader.go b/core/state/reader.go index c912ca28da..38228f8453 100644 --- a/core/state/reader.go +++ b/core/state/reader.go @@ -33,7 +33,6 @@ import ( "github.com/ethereum/go-ethereum/trie" "github.com/ethereum/go-ethereum/trie/bintrie" "github.com/ethereum/go-ethereum/trie/transitiontrie" - "github.com/ethereum/go-ethereum/trie/utils" "github.com/ethereum/go-ethereum/triedb" "github.com/ethereum/go-ethereum/triedb/database" ) @@ -267,7 +266,7 @@ type trieReader struct { // newTrieReader constructs a trie reader of the specific state. An error will be // returned if the associated trie specified by root is not existent. -func newTrieReader(root common.Hash, db *triedb.Database, cache *utils.PointCache) (*trieReader, error) { +func newTrieReader(root common.Hash, db *triedb.Database) (*trieReader, error) { var ( tr Trie err error diff --git a/core/state/state_object.go b/core/state/state_object.go index 91623a838b..411d5fb5b5 100644 --- a/core/state/state_object.go +++ b/core/state/state_object.go @@ -29,6 +29,7 @@ import ( "github.com/ethereum/go-ethereum/log" "github.com/ethereum/go-ethereum/rlp" "github.com/ethereum/go-ethereum/trie" + "github.com/ethereum/go-ethereum/trie/bintrie" "github.com/ethereum/go-ethereum/trie/transitiontrie" "github.com/ethereum/go-ethereum/trie/trienode" "github.com/holiman/uint256" @@ -498,8 +499,8 @@ func (s *stateObject) deepCopy(db *StateDB) *stateObject { } switch s.trie.(type) { - case *trie.VerkleTrie: - // Verkle uses only one tree, and the copy has already been + case *bintrie.BinaryTrie: + // UBT uses only one tree, and the copy has already been // made in mustCopyTrie. obj.trie = db.trie case *transitiontrie.TransitionTrie: diff --git a/core/state/statedb.go b/core/state/statedb.go index 7c6b8bbdfc..c239d66233 100644 --- a/core/state/statedb.go +++ b/core/state/statedb.go @@ -38,7 +38,6 @@ import ( "github.com/ethereum/go-ethereum/params" "github.com/ethereum/go-ethereum/trie" "github.com/ethereum/go-ethereum/trie/trienode" - "github.com/ethereum/go-ethereum/trie/utils" "github.com/holiman/uint256" "golang.org/x/sync/errgroup" ) @@ -188,7 +187,7 @@ func NewWithReader(root common.Hash, db Database, reader Reader) (*StateDB, erro transientStorage: newTransientStorage(), } if db.TrieDB().IsVerkle() { - sdb.accessEvents = NewAccessEvents(db.PointCache()) + sdb.accessEvents = NewAccessEvents() } return sdb, nil } @@ -1495,11 +1494,6 @@ func (s *StateDB) markUpdate(addr common.Address) { s.mutations[addr].typ = update } -// PointCache returns the point cache used by verkle tree. -func (s *StateDB) PointCache() *utils.PointCache { - return s.db.PointCache() -} - // Witness retrieves the current state witness being collected. func (s *StateDB) Witness() *stateless.Witness { return s.witness diff --git a/core/state/statedb_hooked.go b/core/state/statedb_hooked.go index 50acc03aa8..33a2016784 100644 --- a/core/state/statedb_hooked.go +++ b/core/state/statedb_hooked.go @@ -25,7 +25,6 @@ import ( "github.com/ethereum/go-ethereum/core/types" "github.com/ethereum/go-ethereum/crypto" "github.com/ethereum/go-ethereum/params" - "github.com/ethereum/go-ethereum/trie/utils" "github.com/holiman/uint256" ) @@ -133,10 +132,6 @@ func (s *hookedStateDB) AddSlotToAccessList(addr common.Address, slot common.Has s.inner.AddSlotToAccessList(addr, slot) } -func (s *hookedStateDB) PointCache() *utils.PointCache { - return s.inner.PointCache() -} - func (s *hookedStateDB) Prepare(rules params.Rules, sender, coinbase common.Address, dest *common.Address, precompiles []common.Address, txAccesses types.AccessList) { s.inner.Prepare(rules, sender, coinbase, dest, precompiles, txAccesses) } diff --git a/core/types/block.go b/core/types/block.go index b5b6468a13..c52c05a4c7 100644 --- a/core/types/block.go +++ b/core/types/block.go @@ -31,7 +31,6 @@ import ( "github.com/ethereum/go-ethereum/common" "github.com/ethereum/go-ethereum/common/hexutil" "github.com/ethereum/go-ethereum/rlp" - "github.com/ethereum/go-verkle" ) // A BlockNonce is a 64-bit hash which proves (combined with the @@ -61,13 +60,6 @@ func (n *BlockNonce) UnmarshalText(input []byte) error { return hexutil.UnmarshalFixedText("BlockNonce", input, n[:]) } -// ExecutionWitness represents the witness + proof used in a verkle context, -// to provide the ability to execute a block statelessly. -type ExecutionWitness struct { - StateDiff verkle.StateDiff `json:"stateDiff"` - VerkleProof *verkle.VerkleProof `json:"verkleProof"` -} - //go:generate go run github.com/fjl/gencodec -type Header -field-override headerMarshaling -out gen_header_json.go //go:generate go run ../../rlp/rlpgen -type Header -out gen_header_rlp.go @@ -209,11 +201,6 @@ type Block struct { transactions Transactions withdrawals Withdrawals - // witness is not an encoded part of the block body. - // It is held in Block in order for easy relaying to the places - // that process it. - witness *ExecutionWitness - // caches hash atomic.Pointer[common.Hash] size atomic.Uint64 @@ -429,9 +416,6 @@ func (b *Block) BlobGasUsed() *uint64 { return blobGasUsed } -// ExecutionWitness returns the verkle execution witneess + proof for a block -func (b *Block) ExecutionWitness() *ExecutionWitness { return b.witness } - // Size returns the true RLP encoded storage size of the block, either by encoding // and returning it, or returning a previously cached value. func (b *Block) Size() uint64 { @@ -494,7 +478,6 @@ func (b *Block) WithSeal(header *Header) *Block { transactions: b.transactions, uncles: b.uncles, withdrawals: b.withdrawals, - witness: b.witness, } } @@ -506,7 +489,6 @@ func (b *Block) WithBody(body Body) *Block { transactions: slices.Clone(body.Transactions), uncles: make([]*Header, len(body.Uncles)), withdrawals: slices.Clone(body.Withdrawals), - witness: b.witness, } for i := range body.Uncles { block.uncles[i] = CopyHeader(body.Uncles[i]) @@ -514,16 +496,6 @@ func (b *Block) WithBody(body Body) *Block { return block } -func (b *Block) WithWitness(witness *ExecutionWitness) *Block { - return &Block{ - header: b.header, - transactions: b.transactions, - uncles: b.uncles, - withdrawals: b.withdrawals, - witness: witness, - } -} - // Hash returns the keccak256 hash of b's header. // The hash is computed on the first call and cached thereafter. func (b *Block) Hash() common.Hash { diff --git a/core/vm/evm.go b/core/vm/evm.go index 8975c791c8..25a3318c02 100644 --- a/core/vm/evm.go +++ b/core/vm/evm.go @@ -214,7 +214,7 @@ func (evm *EVM) SetJumpDestCache(jumpDests JumpDestCache) { // This is not threadsafe and should only be done very cautiously. func (evm *EVM) SetTxContext(txCtx TxContext) { if evm.chainRules.IsEIP4762 { - txCtx.AccessEvents = state.NewAccessEvents(evm.StateDB.PointCache()) + txCtx.AccessEvents = state.NewAccessEvents() } evm.TxContext = txCtx } diff --git a/core/vm/interface.go b/core/vm/interface.go index d7f4c10e1f..e2f6a65189 100644 --- a/core/vm/interface.go +++ b/core/vm/interface.go @@ -23,7 +23,6 @@ import ( "github.com/ethereum/go-ethereum/core/tracing" "github.com/ethereum/go-ethereum/core/types" "github.com/ethereum/go-ethereum/params" - "github.com/ethereum/go-ethereum/trie/utils" "github.com/holiman/uint256" ) @@ -84,9 +83,6 @@ type StateDB interface { // even if the feature/fork is not active yet AddSlotToAccessList(addr common.Address, slot common.Hash) - // PointCache returns the point cache used in computations - PointCache() *utils.PointCache - Prepare(rules params.Rules, sender, coinbase common.Address, dest *common.Address, precompiles []common.Address, txAccesses types.AccessList) RevertToSnapshot(int) diff --git a/eth/downloader/beaconsync.go b/eth/downloader/beaconsync.go index 750c224230..914e1dfada 100644 --- a/eth/downloader/beaconsync.go +++ b/eth/downloader/beaconsync.go @@ -36,7 +36,6 @@ type beaconBackfiller struct { downloader *Downloader // Downloader to direct via this callback implementation success func() // Callback to run on successful sync cycle completion filling bool // Flag whether the downloader is backfilling or not - filled *types.Header // Last header filled by the last terminated sync loop started chan struct{} // Notification channel whether the downloader inited lock sync.Mutex // Mutex protecting the sync lock } @@ -56,13 +55,15 @@ func (b *beaconBackfiller) suspend() *types.Header { // If no filling is running, don't waste cycles b.lock.Lock() filling := b.filling - filled := b.filled started := b.started b.lock.Unlock() if !filling { + // Sync cycle was inactive, retrieve and return the latest snap block + // as the filled header. log.Debug("Backfiller was inactive") - return filled // Return the filled header on the previous sync completion + + return b.downloader.blockchain.CurrentSnapBlock() } // A previous filling should be running, though it may happen that it hasn't // yet started (being done on a new goroutine). Many concurrent beacon head @@ -77,7 +78,6 @@ func (b *beaconBackfiller) suspend() *types.Header { log.Debug("Backfiller has been suspended") // Sync cycle was just terminated, retrieve and return the last filled header. - // Can't use `filled` as that contains a stale value from before cancellation. return b.downloader.blockchain.CurrentSnapBlock() } @@ -92,7 +92,6 @@ func (b *beaconBackfiller) resume() { return } b.filling = true - b.filled = nil b.started = make(chan struct{}) b.lock.Unlock() @@ -103,7 +102,6 @@ func (b *beaconBackfiller) resume() { defer func() { b.lock.Lock() b.filling = false - b.filled = b.downloader.blockchain.CurrentSnapBlock() b.lock.Unlock() }() // If the downloader fails, report an error as in beacon chain mode there @@ -113,7 +111,7 @@ func (b *beaconBackfiller) resume() { return } // Synchronization succeeded. Since this happens async, notify the outer - // context to disable snap syncing and enable transaction propagation. + // context to enable transaction propagation. if b.success != nil { b.success() } @@ -188,6 +186,8 @@ func (d *Downloader) findBeaconAncestor() (uint64, error) { log.Error("Failed to retrieve beacon bounds", "err", err) return 0, err } + log.Debug("Searching beacon ancestor", "local", number, "beaconhead", beaconHead.Number, "beacontail", beaconTail.Number) + var linked bool switch d.getMode() { case ethconfig.FullSync: @@ -241,6 +241,7 @@ func (d *Downloader) findBeaconAncestor() (uint64, error) { } start = check } + log.Debug("Found beacon ancestor", "number", start) return start, nil } diff --git a/eth/downloader/downloader.go b/eth/downloader/downloader.go index e16014be95..caeb3d64dd 100644 --- a/eth/downloader/downloader.go +++ b/eth/downloader/downloader.go @@ -248,7 +248,7 @@ func New(stateDb ethdb.Database, mode ethconfig.SyncMode, mux *event.TypeMux, ch syncStartBlock: chain.CurrentSnapBlock().Number.Uint64(), } // Create the post-merge skeleton syncer and start the process - dl.skeleton = newSkeleton(stateDb, dl.peers, dropPeer, newBeaconBackfiller(dl, success)) + dl.skeleton = newSkeleton(stateDb, dl.peers, dropPeer, newBeaconBackfiller(dl, success), chain) go dl.stateFetcher() return dl diff --git a/eth/downloader/skeleton.go b/eth/downloader/skeleton.go index c498ac84ec..e693bfc066 100644 --- a/eth/downloader/skeleton.go +++ b/eth/downloader/skeleton.go @@ -207,6 +207,7 @@ type backfiller interface { type skeleton struct { db ethdb.Database // Database backing the skeleton filler backfiller // Chain syncer suspended/resumed by head events + chain chainReader // Underlying block chain peers *peerSet // Set of peers we can sync from idles map[string]*peerConnection // Set of idle peers in the current sync cycle @@ -231,12 +232,19 @@ type skeleton struct { syncStarting func() // callback triggered after a sync cycle is inited but before started } +// chainReader wraps the method to retrieve the head of the local chain. +type chainReader interface { + // CurrentSnapBlock retrieves the head snap block from the local chain. + CurrentSnapBlock() *types.Header +} + // newSkeleton creates a new sync skeleton that tracks a potentially dangling // header chain until it's linked into an existing set of blocks. -func newSkeleton(db ethdb.Database, peers *peerSet, drop peerDropFn, filler backfiller) *skeleton { +func newSkeleton(db ethdb.Database, peers *peerSet, drop peerDropFn, filler backfiller, chain chainReader) *skeleton { sk := &skeleton{ db: db, filler: filler, + chain: chain, peers: peers, drop: drop, requests: make(map[uint64]*headerRequest), @@ -354,6 +362,29 @@ func (s *skeleton) Sync(head *types.Header, final *types.Header, force bool) err } } +// linked returns the flag indicating whether the skeleton has been linked with +// the local chain. +func (s *skeleton) linked(number uint64, hash common.Hash) bool { + linked := rawdb.HasHeader(s.db, hash, number) && + rawdb.HasBody(s.db, hash, number) && + rawdb.HasReceipts(s.db, hash, number) + + // Ensure the skeleton chain links to the local chain below the chain head. + // This accounts for edge cases where leftover chain segments above the head + // may still link to the skeleton chain. In such cases, synchronization is + // likely to fail due to potentially missing segments in the middle. + // + // You can try to produce the edge case by these steps: + // - sync the chain + // - debug.setHead(`0x1`) + // - kill the geth process (the chain segment will be left with chain head rewound) + // - restart + if s.chain.CurrentSnapBlock() != nil { + linked = linked && s.chain.CurrentSnapBlock().Number.Uint64() >= number + } + return linked +} + // sync is the internal version of Sync that executes a single sync cycle, either // until some termination condition is reached, or until the current cycle merges // with a previously aborted run. @@ -378,10 +409,7 @@ func (s *skeleton) sync(head *types.Header) (*types.Header, error) { // If the sync is already done, resume the backfiller. When the loop stops, // terminate the backfiller too. - linked := len(s.progress.Subchains) == 1 && - rawdb.HasHeader(s.db, s.progress.Subchains[0].Next, s.scratchHead) && - rawdb.HasBody(s.db, s.progress.Subchains[0].Next, s.scratchHead) && - rawdb.HasReceipts(s.db, s.progress.Subchains[0].Next, s.scratchHead) + linked := len(s.progress.Subchains) == 1 && s.linked(s.scratchHead, s.progress.Subchains[0].Next) if linked { s.filler.resume() } @@ -497,12 +525,7 @@ func (s *skeleton) sync(head *types.Header) (*types.Header, error) { // is still running, it will pick it up. If it already terminated, // a new cycle needs to be spun up. if linked { - linked = len(s.progress.Subchains) == 1 && - rawdb.HasHeader(s.db, s.progress.Subchains[0].Next, s.scratchHead) && - rawdb.HasBody(s.db, s.progress.Subchains[0].Next, s.scratchHead) && - rawdb.HasReceipts(s.db, s.progress.Subchains[0].Next, s.scratchHead) - - if linked { + if len(s.progress.Subchains) == 1 && s.linked(s.scratchHead, s.progress.Subchains[0].Next) { // The skeleton chain has been extended and is still linked with the local // chain, try to re-schedule the backfiller if it's already terminated. s.filler.resume() @@ -946,6 +969,45 @@ func (s *skeleton) revertRequest(req *headerRequest) { s.scratchOwners[(s.scratchHead-req.head)/requestHeaders] = "" } +// mergeSubchains is invoked once certain beacon headers have been persisted locally +// and the subchains should be merged in case there are some overlaps between. An +// indicator will be returned if the last subchain is merged with previous subchain. +func (s *skeleton) mergeSubchains() bool { + // If the subchain extended into the next subchain, we need to handle + // the overlap. Since there could be many overlaps, do this in a loop. + var merged bool + for len(s.progress.Subchains) > 1 && s.progress.Subchains[1].Head >= s.progress.Subchains[0].Tail { + // Extract some stats from the second subchain + head := s.progress.Subchains[1].Head + tail := s.progress.Subchains[1].Tail + next := s.progress.Subchains[1].Next + + // Since we just overwrote part of the next subchain, we need to trim + // its head independent of matching or mismatching content + if s.progress.Subchains[1].Tail >= s.progress.Subchains[0].Tail { + // Fully overwritten, get rid of the subchain as a whole + log.Debug("Previous subchain fully overwritten", "head", head, "tail", tail, "next", next) + s.progress.Subchains = append(s.progress.Subchains[:1], s.progress.Subchains[2:]...) + continue + } else { + // Partially overwritten, trim the head to the overwritten size + log.Debug("Previous subchain partially overwritten", "head", head, "tail", tail, "next", next) + s.progress.Subchains[1].Head = s.progress.Subchains[0].Tail - 1 + } + // If the old subchain is an extension of the new one, merge the two + // and let the skeleton syncer restart (to clean internal state) + if rawdb.ReadSkeletonHeader(s.db, s.progress.Subchains[1].Head).Hash() == s.progress.Subchains[0].Next { + log.Debug("Previous subchain merged", "head", head, "tail", tail, "next", next) + s.progress.Subchains[0].Tail = s.progress.Subchains[1].Tail + s.progress.Subchains[0].Next = s.progress.Subchains[1].Next + + s.progress.Subchains = append(s.progress.Subchains[:1], s.progress.Subchains[2:]...) + merged = true + } + } + return merged +} + func (s *skeleton) processResponse(res *headerResponse) (linked bool, merged bool) { res.peer.log.Trace("Processing header response", "head", res.headers[0].Number, "hash", res.headers[0].Hash(), "count", len(res.headers)) @@ -1019,10 +1081,9 @@ func (s *skeleton) processResponse(res *headerResponse) (linked bool, merged boo // processing is done, so it's just one more "needless" check. // // The weird cascading checks are done to minimize the database reads. - linked = rawdb.HasHeader(s.db, header.ParentHash, header.Number.Uint64()-1) && - rawdb.HasBody(s.db, header.ParentHash, header.Number.Uint64()-1) && - rawdb.HasReceipts(s.db, header.ParentHash, header.Number.Uint64()-1) + linked = s.linked(header.Number.Uint64()-1, header.ParentHash) if linked { + log.Debug("Primary subchain linked", "number", header.Number.Uint64()-1, "hash", header.ParentHash) break } } @@ -1036,6 +1097,9 @@ func (s *skeleton) processResponse(res *headerResponse) (linked bool, merged boo // If the beacon chain was linked to the local chain, completely swap out // all internal progress and abort header synchronization. if linked { + // Merge all overlapped subchains beforehand + s.mergeSubchains() + // Linking into the local chain should also mean that there are no // leftover subchains, but in the case of importing the blocks via // the engine API, we will not push the subchains forward. This will @@ -1093,41 +1157,10 @@ func (s *skeleton) processResponse(res *headerResponse) (linked bool, merged boo s.scratchHead -= uint64(consumed) - // If the subchain extended into the next subchain, we need to handle - // the overlap. Since there could be many overlaps (come on), do this - // in a loop. - for len(s.progress.Subchains) > 1 && s.progress.Subchains[1].Head >= s.progress.Subchains[0].Tail { - // Extract some stats from the second subchain - head := s.progress.Subchains[1].Head - tail := s.progress.Subchains[1].Tail - next := s.progress.Subchains[1].Next - - // Since we just overwrote part of the next subchain, we need to trim - // its head independent of matching or mismatching content - if s.progress.Subchains[1].Tail >= s.progress.Subchains[0].Tail { - // Fully overwritten, get rid of the subchain as a whole - log.Debug("Previous subchain fully overwritten", "head", head, "tail", tail, "next", next) - s.progress.Subchains = append(s.progress.Subchains[:1], s.progress.Subchains[2:]...) - continue - } else { - // Partially overwritten, trim the head to the overwritten size - log.Debug("Previous subchain partially overwritten", "head", head, "tail", tail, "next", next) - s.progress.Subchains[1].Head = s.progress.Subchains[0].Tail - 1 - } - // If the old subchain is an extension of the new one, merge the two - // and let the skeleton syncer restart (to clean internal state) - if rawdb.ReadSkeletonHeader(s.db, s.progress.Subchains[1].Head).Hash() == s.progress.Subchains[0].Next { - log.Debug("Previous subchain merged", "head", head, "tail", tail, "next", next) - s.progress.Subchains[0].Tail = s.progress.Subchains[1].Tail - s.progress.Subchains[0].Next = s.progress.Subchains[1].Next - - s.progress.Subchains = append(s.progress.Subchains[:1], s.progress.Subchains[2:]...) - merged = true - } - } // If subchains were merged, all further available headers in the scratch // space are invalid since we skipped ahead. Stop processing the scratch // space to avoid dropping peers thinking they delivered invalid data. + merged = s.mergeSubchains() if merged { break } @@ -1158,15 +1191,17 @@ func (s *skeleton) processResponse(res *headerResponse) (linked bool, merged boo // due to the downloader backfilling past the tracked tail. func (s *skeleton) cleanStales(filled *types.Header) error { number := filled.Number.Uint64() - log.Trace("Cleaning stale beacon headers", "filled", number, "hash", filled.Hash()) + log.Debug("Cleaning stale beacon headers", "filled", number, "hash", filled.Hash()) - // If the filled header is below the linked subchain, something's corrupted - // internally. Report and error and refuse to do anything. + // If the filled header is below the subchain, it means the skeleton is not + // linked with local chain yet, don't bother to do cleanup. if number+1 < s.progress.Subchains[0].Tail { - return fmt.Errorf("filled header below beacon header tail: %d < %d", number, s.progress.Subchains[0].Tail) + log.Debug("filled header below beacon header tail", "filled", number, "tail", s.progress.Subchains[0].Tail) + return nil } // If nothing in subchain is filled, don't bother to do cleanup. if number+1 == s.progress.Subchains[0].Tail { + log.Debug("Skeleton chain not yet consumed", "filled", number, "hash", filled.Hash(), "tail", s.progress.Subchains[0].Tail) return nil } // If the latest fill was on a different subchain, it means the backfiller diff --git a/eth/downloader/skeleton_test.go b/eth/downloader/skeleton_test.go index 4aa97cf1f7..5c54b4b5c2 100644 --- a/eth/downloader/skeleton_test.go +++ b/eth/downloader/skeleton_test.go @@ -20,6 +20,7 @@ import ( "encoding/json" "errors" "fmt" + "math" "math/big" "sync/atomic" "testing" @@ -71,6 +72,12 @@ func (hf *hookedBackfiller) resume() { } } +type fakeChainReader struct{} + +func (fc *fakeChainReader) CurrentSnapBlock() *types.Header { + return &types.Header{Number: big.NewInt(math.MaxInt64)} +} + // skeletonTestPeer is a mock peer that can only serve header requests from a // pre-perated header chain (which may be arbitrarily wrong for testing). // @@ -369,7 +376,7 @@ func TestSkeletonSyncInit(t *testing.T) { // Create a skeleton sync and run a cycle wait := make(chan struct{}) - skeleton := newSkeleton(db, newPeerSet(), nil, newHookedBackfiller()) + skeleton := newSkeleton(db, newPeerSet(), nil, newHookedBackfiller(), &fakeChainReader{}) skeleton.syncStarting = func() { close(wait) } skeleton.Sync(tt.head, nil, true) @@ -472,7 +479,7 @@ func TestSkeletonSyncExtend(t *testing.T) { // Create a skeleton sync and run a cycle wait := make(chan struct{}) - skeleton := newSkeleton(db, newPeerSet(), nil, newHookedBackfiller()) + skeleton := newSkeleton(db, newPeerSet(), nil, newHookedBackfiller(), &fakeChainReader{}) skeleton.syncStarting = func() { close(wait) } skeleton.Sync(tt.head, nil, true) @@ -885,7 +892,7 @@ func TestSkeletonSyncRetrievals(t *testing.T) { } } // Create a skeleton sync and run a cycle - skeleton := newSkeleton(db, peerset, drop, filler) + skeleton := newSkeleton(db, peerset, drop, filler, &fakeChainReader{}) skeleton.Sync(tt.head, nil, true) // Wait a bit (bleah) for the initial sync loop to go to idle. This might diff --git a/go.mod b/go.mod index aff1d53923..66f3a3ffa5 100644 --- a/go.mod +++ b/go.mod @@ -15,7 +15,6 @@ require ( github.com/cockroachdb/pebble v1.1.5 github.com/consensys/gnark-crypto v0.18.1 github.com/crate-crypto/go-eth-kzg v1.4.0 - github.com/crate-crypto/go-ipa v0.0.0-20240724233137-53bbb0ceb27a github.com/davecgh/go-spew v1.1.1 github.com/dchest/siphash v1.2.3 github.com/deckarep/golang-set/v2 v2.6.0 @@ -24,7 +23,6 @@ require ( github.com/dop251/goja v0.0.0-20230605162241-28ee0ee714f3 github.com/ethereum/c-kzg-4844/v2 v2.1.5 github.com/ethereum/go-bigmodexpfix v0.0.0-20250911101455-f9e208c548ab - github.com/ethereum/go-verkle v0.2.2 github.com/fatih/color v1.16.0 github.com/ferranbt/fastssz v0.1.4 github.com/fsnotify/fsnotify v1.6.0 diff --git a/go.sum b/go.sum index 503e0975d6..ad066abc03 100644 --- a/go.sum +++ b/go.sum @@ -81,8 +81,6 @@ github.com/cpuguy83/go-md2man/v2 v2.0.5 h1:ZtcqGrnekaHpVLArFSe4HK5DoKx1T0rq2DwVB github.com/cpuguy83/go-md2man/v2 v2.0.5/go.mod h1:tgQtvFlXSQOSOSIRvRPT7W67SCa46tRHOmNcaadrF8o= github.com/crate-crypto/go-eth-kzg v1.4.0 h1:WzDGjHk4gFg6YzV0rJOAsTK4z3Qkz5jd4RE3DAvPFkg= github.com/crate-crypto/go-eth-kzg v1.4.0/go.mod h1:J9/u5sWfznSObptgfa92Jq8rTswn6ahQWEuiLHOjCUI= -github.com/crate-crypto/go-ipa v0.0.0-20240724233137-53bbb0ceb27a h1:W8mUrRp6NOVl3J+MYp5kPMoUZPp7aOYHtaua31lwRHg= -github.com/crate-crypto/go-ipa v0.0.0-20240724233137-53bbb0ceb27a/go.mod h1:sTwzHBvIzm2RfVCGNEBZgRyjwK40bVoun3ZnGOCafNM= github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E= github.com/cyberdelia/templates v0.0.0-20141128023046-ca7fffd4298c/go.mod h1:GyV+0YP4qX0UQ7r2MoYZ+AvYDp12OF5yg4q8rGnyNh4= github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38= @@ -117,8 +115,6 @@ github.com/ethereum/c-kzg-4844/v2 v2.1.5 h1:aVtoLK5xwJ6c5RiqO8g8ptJ5KU+2Hdquf6G3 github.com/ethereum/c-kzg-4844/v2 v2.1.5/go.mod h1:u59hRTTah4Co6i9fDWtiCjTrblJv0UwsqZKCc0GfgUs= github.com/ethereum/go-bigmodexpfix v0.0.0-20250911101455-f9e208c548ab h1:rvv6MJhy07IMfEKuARQ9TKojGqLVNxQajaXEp/BoqSk= github.com/ethereum/go-bigmodexpfix v0.0.0-20250911101455-f9e208c548ab/go.mod h1:IuLm4IsPipXKF7CW5Lzf68PIbZ5yl7FFd74l/E0o9A8= -github.com/ethereum/go-verkle v0.2.2 h1:I2W0WjnrFUIzzVPwm8ykY+7pL2d4VhlsePn4j7cnFk8= -github.com/ethereum/go-verkle v0.2.2/go.mod h1:M3b90YRnzqKyyzBEWJGqj8Qff4IDeXnzFw0P9bFw3uk= github.com/fatih/color v1.16.0 h1:zmkK9Ngbjj+K0yRhTVONQh1p/HknKYSlNT+vZCzyokM= github.com/fatih/color v1.16.0/go.mod h1:fL2Sau1YI5c0pdGEVCbKQbLXB6edEj1ZgiY4NijnWvE= github.com/ferranbt/fastssz v0.1.4 h1:OCDB+dYDEQDvAgtAGnTSidK1Pe2tW3nFV40XyMkTeDY= diff --git a/params/protocol_params.go b/params/protocol_params.go index e8b044f450..bb506af015 100644 --- a/params/protocol_params.go +++ b/params/protocol_params.go @@ -32,7 +32,7 @@ const ( MaximumExtraDataSize uint64 = 32 // Maximum size extra data may be after Genesis. ExpByteGas uint64 = 10 // Times ceil(log256(exponent)) for the EXP instruction. - SloadGas uint64 = 50 // Multiplied by the number of 32-byte words that are copied (round up) for any *COPY operation and added. + SloadGas uint64 = 50 // CallValueTransferGas uint64 = 9000 // Paid for CALL when the value transfer is non-zero. CallNewAccountGas uint64 = 25000 // Paid for CALL when the destination address didn't exist prior. TxGas uint64 = 21000 // Per transaction not creating a contract. NOTE: Not payable on data of calls between transactions. @@ -82,7 +82,7 @@ const ( CallCreateDepth uint64 = 1024 // Maximum depth of call/create stack. ExpGas uint64 = 10 // Once per EXP instruction LogGas uint64 = 375 // Per LOG* operation. - CopyGas uint64 = 3 // + CopyGas uint64 = 3 // Multiplied by the number of 32-byte words that are copied (round up) for any *COPY operation and added. StackLimit uint64 = 1024 // Maximum size of VM stack allowed. TierStepGas uint64 = 0 // Once per operation, for a selection of them. LogTopicGas uint64 = 375 // Multiplied by the * of the LOG*, per LOG transaction. e.g. LOG0 incurs 0 * c_txLogTopicGas, LOG4 incurs 4 * c_txLogTopicGas. diff --git a/trie/bintrie/key_encoding.go b/trie/bintrie/key_encoding.go index cda797521a..5a93fcde9a 100644 --- a/trie/bintrie/key_encoding.go +++ b/trie/bintrie/key_encoding.go @@ -33,8 +33,17 @@ const ( ) var ( - zeroHash = common.Hash{} - codeOffset = uint256.NewInt(128) + zeroInt = uint256.NewInt(0) + zeroHash = common.Hash{} + verkleNodeWidthLog2 = 8 + headerStorageOffset = uint256.NewInt(64) + codeOffset = uint256.NewInt(128) + codeStorageDelta = uint256.NewInt(0).Sub(codeOffset, headerStorageOffset) + mainStorageOffsetLshVerkleNodeWidth = new(uint256.Int).Lsh(uint256.NewInt(1), 248-uint(verkleNodeWidthLog2)) + CodeOffset = uint256.NewInt(128) + VerkleNodeWidth = uint256.NewInt(256) + HeaderStorageOffset = uint256.NewInt(64) + VerkleNodeWidthLog2 = 8 ) func GetBinaryTreeKey(addr common.Address, key []byte) []byte { @@ -83,3 +92,38 @@ func GetBinaryTreeKeyCodeChunk(address common.Address, chunknr *uint256.Int) []b chunkOffset := new(uint256.Int).Add(codeOffset, chunknr).Bytes() return GetBinaryTreeKey(address, chunkOffset) } + +func StorageIndex(storageKey []byte) (*uint256.Int, byte) { + // If the storage slot is in the header, we need to add the header offset. + var key uint256.Int + key.SetBytes(storageKey) + if key.Cmp(codeStorageDelta) < 0 { + // This addition is always safe; it can't ever overflow since pos. - -package utils - -import ( - "encoding/binary" - "sync" - - "github.com/crate-crypto/go-ipa/bandersnatch/fr" - "github.com/ethereum/go-ethereum/common/lru" - "github.com/ethereum/go-ethereum/metrics" - "github.com/ethereum/go-verkle" - "github.com/holiman/uint256" -) - -const ( - BasicDataLeafKey = 0 - CodeHashLeafKey = 1 - - BasicDataVersionOffset = 0 - BasicDataCodeSizeOffset = 5 - BasicDataNonceOffset = 8 - BasicDataBalanceOffset = 16 -) - -var ( - zero = uint256.NewInt(0) - verkleNodeWidthLog2 = 8 - headerStorageOffset = uint256.NewInt(64) - codeOffset = uint256.NewInt(128) - verkleNodeWidth = uint256.NewInt(256) - codeStorageDelta = uint256.NewInt(0).Sub(codeOffset, headerStorageOffset) - mainStorageOffsetLshVerkleNodeWidth = new(uint256.Int).Lsh(uint256.NewInt(1), 248-uint(verkleNodeWidthLog2)) - CodeOffset = uint256.NewInt(128) - VerkleNodeWidth = uint256.NewInt(256) - HeaderStorageOffset = uint256.NewInt(64) - VerkleNodeWidthLog2 = 8 - - index0Point *verkle.Point // pre-computed commitment of polynomial [2+256*64] - - // cacheHitGauge is the metric to track how many cache hit occurred. - cacheHitGauge = metrics.NewRegisteredGauge("trie/verkle/cache/hit", nil) - - // cacheMissGauge is the metric to track how many cache miss occurred. - cacheMissGauge = metrics.NewRegisteredGauge("trie/verkle/cache/miss", nil) -) - -func init() { - // The byte array is the Marshalled output of the point computed as such: - // - // var ( - // config = verkle.GetConfig() - // fr verkle.Fr - // ) - // verkle.FromLEBytes(&fr, []byte{2, 64}) - // point := config.CommitToPoly([]verkle.Fr{fr}, 1) - index0Point = new(verkle.Point) - err := index0Point.SetBytes([]byte{34, 25, 109, 242, 193, 5, 144, 224, 76, 52, 189, 92, 197, 126, 9, 145, 27, 152, 199, 130, 165, 3, 210, 27, 193, 131, 142, 28, 110, 26, 16, 191}) - if err != nil { - panic(err) - } -} - -// PointCache is the LRU cache for storing evaluated address commitment. -type PointCache struct { - lru lru.BasicLRU[string, *verkle.Point] - lock sync.RWMutex -} - -// NewPointCache returns the cache with specified size. -func NewPointCache(maxItems int) *PointCache { - return &PointCache{ - lru: lru.NewBasicLRU[string, *verkle.Point](maxItems), - } -} - -// Get returns the cached commitment for the specified address, or computing -// it on the flight. -func (c *PointCache) Get(addr []byte) *verkle.Point { - c.lock.Lock() - defer c.lock.Unlock() - - p, ok := c.lru.Get(string(addr)) - if ok { - cacheHitGauge.Inc(1) - return p - } - cacheMissGauge.Inc(1) - p = evaluateAddressPoint(addr) - c.lru.Add(string(addr), p) - return p -} - -// GetStem returns the first 31 bytes of the tree key as the tree stem. It only -// works for the account metadata whose treeIndex is 0. -func (c *PointCache) GetStem(addr []byte) []byte { - p := c.Get(addr) - return pointToHash(p, 0)[:31] -} - -// GetTreeKey performs both the work of the spec's get_tree_key function, and that -// of pedersen_hash: it builds the polynomial in pedersen_hash without having to -// create a mostly zero-filled buffer and "type cast" it to a 128-long 16-byte -// array. Since at most the first 5 coefficients of the polynomial will be non-zero, -// these 5 coefficients are created directly. -func GetTreeKey(address []byte, treeIndex *uint256.Int, subIndex byte) []byte { - if len(address) < 32 { - var aligned [32]byte - address = append(aligned[:32-len(address)], address...) - } - // poly = [2+256*64, address_le_low, address_le_high, tree_index_le_low, tree_index_le_high] - var poly [5]fr.Element - - // 32-byte address, interpreted as two little endian - // 16-byte numbers. - verkle.FromLEBytes(&poly[1], address[:16]) - verkle.FromLEBytes(&poly[2], address[16:]) - - // treeIndex must be interpreted as a 32-byte aligned little-endian integer. - // e.g: if treeIndex is 0xAABBCC, we need the byte representation to be 0xCCBBAA00...00. - // poly[3] = LE({CC,BB,AA,00...0}) (16 bytes), poly[4]=LE({00,00,...}) (16 bytes). - // - // To avoid unnecessary endianness conversions for go-ipa, we do some trick: - // - poly[3]'s byte representation is the same as the *top* 16 bytes (trieIndexBytes[16:]) of - // 32-byte aligned big-endian representation (BE({00,...,AA,BB,CC})). - // - poly[4]'s byte representation is the same as the *low* 16 bytes (trieIndexBytes[:16]) of - // the 32-byte aligned big-endian representation (BE({00,00,...}). - trieIndexBytes := treeIndex.Bytes32() - verkle.FromBytes(&poly[3], trieIndexBytes[16:]) - verkle.FromBytes(&poly[4], trieIndexBytes[:16]) - - cfg := verkle.GetConfig() - ret := cfg.CommitToPoly(poly[:], 0) - - // add a constant point corresponding to poly[0]=[2+256*64]. - ret.Add(ret, index0Point) - - return pointToHash(ret, subIndex) -} - -// GetTreeKeyWithEvaluatedAddress is basically identical to GetTreeKey, the only -// difference is a part of polynomial is already evaluated. -// -// Specifically, poly = [2+256*64, address_le_low, address_le_high] is already -// evaluated. -func GetTreeKeyWithEvaluatedAddress(evaluated *verkle.Point, treeIndex *uint256.Int, subIndex byte) []byte { - var poly [5]fr.Element - - // little-endian, 32-byte aligned treeIndex - var index [32]byte - for i := 0; i < len(treeIndex); i++ { - binary.LittleEndian.PutUint64(index[i*8:(i+1)*8], treeIndex[i]) - } - verkle.FromLEBytes(&poly[3], index[:16]) - verkle.FromLEBytes(&poly[4], index[16:]) - - cfg := verkle.GetConfig() - ret := cfg.CommitToPoly(poly[:], 0) - - // add the pre-evaluated address - ret.Add(ret, evaluated) - - return pointToHash(ret, subIndex) -} - -// BasicDataKey returns the verkle tree key of the basic data field for -// the specified account. -func BasicDataKey(address []byte) []byte { - return GetTreeKey(address, zero, BasicDataLeafKey) -} - -// CodeHashKey returns the verkle tree key of the code hash field for -// the specified account. -func CodeHashKey(address []byte) []byte { - return GetTreeKey(address, zero, CodeHashLeafKey) -} - -func codeChunkIndex(chunk *uint256.Int) (*uint256.Int, byte) { - var ( - chunkOffset = new(uint256.Int).Add(codeOffset, chunk) - treeIndex, subIndexMod = new(uint256.Int).DivMod(chunkOffset, verkleNodeWidth, new(uint256.Int)) - ) - return treeIndex, byte(subIndexMod.Uint64()) -} - -// CodeChunkKey returns the verkle tree key of the code chunk for the -// specified account. -func CodeChunkKey(address []byte, chunk *uint256.Int) []byte { - treeIndex, subIndex := codeChunkIndex(chunk) - return GetTreeKey(address, treeIndex, subIndex) -} - -func GetTreeKeyCodeChunkIndices(chunk *uint256.Int) (*uint256.Int, byte) { - chunkOffset := new(uint256.Int).Add(CodeOffset, chunk) - treeIndex := new(uint256.Int).Div(chunkOffset, VerkleNodeWidth) - subIndexMod := new(uint256.Int).Mod(chunkOffset, VerkleNodeWidth) - var subIndex byte - if len(subIndexMod) != 0 { - subIndex = byte(subIndexMod[0]) - } - return treeIndex, subIndex -} - -func GetTreeKeyCodeChunk(address []byte, chunk *uint256.Int) []byte { - treeIndex, subIndex := GetTreeKeyCodeChunkIndices(chunk) - return GetTreeKey(address, treeIndex, subIndex) -} - -func StorageIndex(storageKey []byte) (*uint256.Int, byte) { - // If the storage slot is in the header, we need to add the header offset. - var key uint256.Int - key.SetBytes(storageKey) - if key.Cmp(codeStorageDelta) < 0 { - // This addition is always safe; it can't ever overflow since pos. - -package utils - -import ( - "bytes" - "testing" - - "github.com/ethereum/go-verkle" - "github.com/holiman/uint256" -) - -func TestTreeKey(t *testing.T) { - var ( - address = []byte{0x01} - addressEval = evaluateAddressPoint(address) - smallIndex = uint256.NewInt(1) - largeIndex = uint256.NewInt(10000) - smallStorage = []byte{0x1} - largeStorage = bytes.Repeat([]byte{0xff}, 16) - ) - if !bytes.Equal(BasicDataKey(address), BasicDataKeyWithEvaluatedAddress(addressEval)) { - t.Fatal("Unmatched basic data key") - } - if !bytes.Equal(CodeHashKey(address), CodeHashKeyWithEvaluatedAddress(addressEval)) { - t.Fatal("Unmatched code hash key") - } - if !bytes.Equal(CodeChunkKey(address, smallIndex), CodeChunkKeyWithEvaluatedAddress(addressEval, smallIndex)) { - t.Fatal("Unmatched code chunk key") - } - if !bytes.Equal(CodeChunkKey(address, largeIndex), CodeChunkKeyWithEvaluatedAddress(addressEval, largeIndex)) { - t.Fatal("Unmatched code chunk key") - } - if !bytes.Equal(StorageSlotKey(address, smallStorage), StorageSlotKeyWithEvaluatedAddress(addressEval, smallStorage)) { - t.Fatal("Unmatched storage slot key") - } - if !bytes.Equal(StorageSlotKey(address, largeStorage), StorageSlotKeyWithEvaluatedAddress(addressEval, largeStorage)) { - t.Fatal("Unmatched storage slot key") - } -} - -// goos: darwin -// goarch: amd64 -// pkg: github.com/ethereum/go-ethereum/trie/utils -// cpu: VirtualApple @ 2.50GHz -// BenchmarkTreeKey -// BenchmarkTreeKey-8 398731 2961 ns/op 32 B/op 1 allocs/op -func BenchmarkTreeKey(b *testing.B) { - // Initialize the IPA settings which can be pretty expensive. - verkle.GetConfig() - - b.ReportAllocs() - b.ResetTimer() - - for i := 0; i < b.N; i++ { - BasicDataKey([]byte{0x01}) - } -} - -// goos: darwin -// goarch: amd64 -// pkg: github.com/ethereum/go-ethereum/trie/utils -// cpu: VirtualApple @ 2.50GHz -// BenchmarkTreeKeyWithEvaluation -// BenchmarkTreeKeyWithEvaluation-8 513855 2324 ns/op 32 B/op 1 allocs/op -func BenchmarkTreeKeyWithEvaluation(b *testing.B) { - // Initialize the IPA settings which can be pretty expensive. - verkle.GetConfig() - - addr := []byte{0x01} - eval := evaluateAddressPoint(addr) - - b.ReportAllocs() - b.ResetTimer() - for i := 0; i < b.N; i++ { - BasicDataKeyWithEvaluatedAddress(eval) - } -} - -// goos: darwin -// goarch: amd64 -// pkg: github.com/ethereum/go-ethereum/trie/utils -// cpu: VirtualApple @ 2.50GHz -// BenchmarkStorageKey -// BenchmarkStorageKey-8 230516 4584 ns/op 96 B/op 3 allocs/op -func BenchmarkStorageKey(b *testing.B) { - // Initialize the IPA settings which can be pretty expensive. - verkle.GetConfig() - - b.ReportAllocs() - b.ResetTimer() - - for i := 0; i < b.N; i++ { - StorageSlotKey([]byte{0x01}, bytes.Repeat([]byte{0xff}, 32)) - } -} - -// goos: darwin -// goarch: amd64 -// pkg: github.com/ethereum/go-ethereum/trie/utils -// cpu: VirtualApple @ 2.50GHz -// BenchmarkStorageKeyWithEvaluation -// BenchmarkStorageKeyWithEvaluation-8 320125 3753 ns/op 96 B/op 3 allocs/op -func BenchmarkStorageKeyWithEvaluation(b *testing.B) { - // Initialize the IPA settings which can be pretty expensive. - verkle.GetConfig() - - addr := []byte{0x01} - eval := evaluateAddressPoint(addr) - - b.ReportAllocs() - b.ResetTimer() - for i := 0; i < b.N; i++ { - StorageSlotKeyWithEvaluatedAddress(eval, bytes.Repeat([]byte{0xff}, 32)) - } -} diff --git a/trie/verkle.go b/trie/verkle.go deleted file mode 100644 index 70793330c5..0000000000 --- a/trie/verkle.go +++ /dev/null @@ -1,458 +0,0 @@ -// Copyright 2023 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 trie - -import ( - "bytes" - "encoding/binary" - "errors" - "fmt" - - "github.com/ethereum/go-ethereum/common" - "github.com/ethereum/go-ethereum/core/types" - "github.com/ethereum/go-ethereum/ethdb" - "github.com/ethereum/go-ethereum/trie/trienode" - "github.com/ethereum/go-ethereum/trie/utils" - "github.com/ethereum/go-ethereum/triedb/database" - "github.com/ethereum/go-verkle" - "github.com/holiman/uint256" -) - -var ( - errInvalidRootType = errors.New("invalid node type for root") -) - -// VerkleTrie is a wrapper around VerkleNode that implements the trie.Trie -// interface so that Verkle trees can be reused verbatim. -type VerkleTrie struct { - root verkle.VerkleNode - cache *utils.PointCache - reader *Reader - tracer *PrevalueTracer -} - -// NewVerkleTrie constructs a verkle tree based on the specified root hash. -func NewVerkleTrie(root common.Hash, db database.NodeDatabase, cache *utils.PointCache) (*VerkleTrie, error) { - reader, err := NewReader(root, common.Hash{}, db) - if err != nil { - return nil, err - } - t := &VerkleTrie{ - root: verkle.New(), - cache: cache, - reader: reader, - tracer: NewPrevalueTracer(), - } - // Parse the root verkle node if it's not empty. - if root != types.EmptyVerkleHash && root != types.EmptyRootHash { - blob, err := t.nodeResolver(nil) - if err != nil { - return nil, err - } - node, err := verkle.ParseNode(blob, 0) - if err != nil { - return nil, err - } - t.root = node - } - return t, nil -} - -// GetKey returns the sha3 preimage of a hashed key that was previously used -// to store a value. -func (t *VerkleTrie) GetKey(key []byte) []byte { - return key -} - -// GetAccount implements state.Trie, retrieving the account with the specified -// account address. If the specified account is not in the verkle tree, nil will -// be returned. If the tree is corrupted, an error will be returned. -func (t *VerkleTrie) GetAccount(addr common.Address) (*types.StateAccount, error) { - var ( - acc = &types.StateAccount{} - values [][]byte - err error - ) - switch n := t.root.(type) { - case *verkle.InternalNode: - values, err = n.GetValuesAtStem(t.cache.GetStem(addr[:]), t.nodeResolver) - if err != nil { - return nil, fmt.Errorf("GetAccount (%x) error: %v", addr, err) - } - default: - return nil, errInvalidRootType - } - if values == nil { - return nil, nil - } - basicData := values[utils.BasicDataLeafKey] - acc.Nonce = binary.BigEndian.Uint64(basicData[utils.BasicDataNonceOffset:]) - acc.Balance = new(uint256.Int).SetBytes(basicData[utils.BasicDataBalanceOffset : utils.BasicDataBalanceOffset+16]) - acc.CodeHash = values[utils.CodeHashLeafKey] - - // TODO account.Root is leave as empty. How should we handle the legacy account? - return acc, nil -} - -// PrefetchAccount attempts to resolve specific accounts from the database -// to accelerate subsequent trie operations. -func (t *VerkleTrie) PrefetchAccount(addresses []common.Address) error { - for _, addr := range addresses { - if _, err := t.GetAccount(addr); err != nil { - return err - } - } - return nil -} - -// GetStorage implements state.Trie, retrieving the storage slot with the specified -// account address and storage key. If the specified slot is not in the verkle tree, -// nil will be returned. If the tree is corrupted, an error will be returned. -func (t *VerkleTrie) GetStorage(addr common.Address, key []byte) ([]byte, error) { - k := utils.StorageSlotKeyWithEvaluatedAddress(t.cache.Get(addr.Bytes()), key) - val, err := t.root.Get(k, t.nodeResolver) - if err != nil { - return nil, err - } - return common.TrimLeftZeroes(val), nil -} - -// PrefetchStorage attempts to resolve specific storage slots from the database -// to accelerate subsequent trie operations. -func (t *VerkleTrie) PrefetchStorage(addr common.Address, keys [][]byte) error { - for _, key := range keys { - if _, err := t.GetStorage(addr, key); err != nil { - return err - } - } - return nil -} - -// UpdateAccount implements state.Trie, writing the provided account into the tree. -// If the tree is corrupted, an error will be returned. -func (t *VerkleTrie) UpdateAccount(addr common.Address, acc *types.StateAccount, codeLen int) error { - var ( - err error - basicData [32]byte - values = make([][]byte, verkle.NodeWidth) - stem = t.cache.GetStem(addr[:]) - ) - - // Code size is encoded in BasicData as a 3-byte big-endian integer. Spare bytes are present - // before the code size to support bigger integers in the future. PutUint32(...) requires - // 4 bytes, so we need to shift the offset 1 byte to the left. - binary.BigEndian.PutUint32(basicData[utils.BasicDataCodeSizeOffset-1:], uint32(codeLen)) - binary.BigEndian.PutUint64(basicData[utils.BasicDataNonceOffset:], acc.Nonce) - if acc.Balance.ByteLen() > 16 { - panic("balance too large") - } - acc.Balance.WriteToSlice(basicData[utils.BasicDataBalanceOffset : utils.BasicDataBalanceOffset+16]) - values[utils.BasicDataLeafKey] = basicData[:] - values[utils.CodeHashLeafKey] = acc.CodeHash[:] - - switch root := t.root.(type) { - case *verkle.InternalNode: - err = root.InsertValuesAtStem(stem, values, t.nodeResolver) - default: - return errInvalidRootType - } - if err != nil { - return fmt.Errorf("UpdateAccount (%x) error: %v", addr, err) - } - - return nil -} - -// UpdateStorage implements state.Trie, writing the provided storage slot into -// the tree. If the tree is corrupted, an error will be returned. -func (t *VerkleTrie) UpdateStorage(address common.Address, key, value []byte) error { - // Left padding the slot value to 32 bytes. - var v [32]byte - if len(value) >= 32 { - copy(v[:], value[:32]) - } else { - copy(v[32-len(value):], value[:]) - } - k := utils.StorageSlotKeyWithEvaluatedAddress(t.cache.Get(address.Bytes()), key) - return t.root.Insert(k, v[:], t.nodeResolver) -} - -// DeleteAccount leaves the account untouched, as no account deletion can happen -// in verkle. -// There is a special corner case, in which an account that is prefunded, CREATE2-d -// and then SELFDESTRUCT-d should see its funds drained. EIP161 says that account -// should be removed, but this is verboten by the verkle spec. This contains a -// workaround in which the method checks for this corner case, and if so, overwrites -// the balance with 0. This will be removed once the spec has been clarified. -func (t *VerkleTrie) DeleteAccount(addr common.Address) error { - k := utils.BasicDataKeyWithEvaluatedAddress(t.cache.Get(addr.Bytes())) - values, err := t.root.(*verkle.InternalNode).GetValuesAtStem(k, t.nodeResolver) - if err != nil { - return fmt.Errorf("Error getting data at %x in delete: %w", k, err) - } - var prefunded bool - for i, v := range values { - switch i { - case 0: - prefunded = len(v) == 32 - case 1: - prefunded = len(v) == 32 && bytes.Equal(v, types.EmptyCodeHash[:]) - default: - prefunded = v == nil - } - if !prefunded { - break - } - } - if prefunded { - t.root.Insert(k, common.Hash{}.Bytes(), t.nodeResolver) - } - return nil -} - -// RollBackAccount removes the account info + code from the tree, unlike DeleteAccount -// that will overwrite it with 0s. The first 64 storage slots are also removed. -func (t *VerkleTrie) RollBackAccount(addr common.Address) error { - var ( - evaluatedAddr = t.cache.Get(addr.Bytes()) - basicDataKey = utils.BasicDataKeyWithEvaluatedAddress(evaluatedAddr) - ) - basicDataBytes, err := t.root.Get(basicDataKey, t.nodeResolver) - if err != nil { - return fmt.Errorf("rollback: error finding code size: %w", err) - } - if len(basicDataBytes) == 0 { - return errors.New("rollback: basic data is not existent") - } - // The code size is encoded in BasicData as a 3-byte big-endian integer. Spare bytes are present - // before the code size to support bigger integers in the future. - // LittleEndian.Uint32(...) expects 4-bytes, so we need to shift the offset 1-byte to the left. - codeSize := binary.BigEndian.Uint32(basicDataBytes[utils.BasicDataCodeSizeOffset-1:]) - - // Delete the account header + first 64 slots + first 128 code chunks - _, err = t.root.(*verkle.InternalNode).DeleteAtStem(basicDataKey[:31], t.nodeResolver) - if err != nil { - return fmt.Errorf("error rolling back account header: %w", err) - } - - // Delete all further code - for i, chunknr := uint64(31*128), uint64(128); i < uint64(codeSize); i, chunknr = i+31*256, chunknr+256 { - // evaluate group key at the start of a new group - offset := uint256.NewInt(chunknr) - key := utils.CodeChunkKeyWithEvaluatedAddress(evaluatedAddr, offset) - - if _, err = t.root.(*verkle.InternalNode).DeleteAtStem(key[:], t.nodeResolver); err != nil { - return fmt.Errorf("error deleting code chunk stem (addr=%x, offset=%d) error: %w", addr[:], offset, err) - } - } - return nil -} - -// DeleteStorage implements state.Trie, deleting the specified storage slot from -// the trie. If the storage slot was not existent in the trie, no error will be -// returned. If the trie is corrupted, an error will be returned. -func (t *VerkleTrie) DeleteStorage(addr common.Address, key []byte) error { - var zero [32]byte - k := utils.StorageSlotKeyWithEvaluatedAddress(t.cache.Get(addr.Bytes()), key) - return t.root.Insert(k, zero[:], t.nodeResolver) -} - -// Hash returns the root hash of the tree. It does not write to the database and -// can be used even if the tree doesn't have one. -func (t *VerkleTrie) Hash() common.Hash { - return t.root.Commit().Bytes() -} - -// Commit writes all nodes to the tree's memory database. -func (t *VerkleTrie) Commit(_ bool) (common.Hash, *trienode.NodeSet) { - root := t.root.(*verkle.InternalNode) - nodes, err := root.BatchSerialize() - if err != nil { - // Error return from this function indicates error in the code logic - // of BatchSerialize, and we fail catastrophically if this is the case. - panic(fmt.Errorf("BatchSerialize failed: %v", err)) - } - nodeset := trienode.NewNodeSet(common.Hash{}) - for _, node := range nodes { - // Hash parameter is not used in pathdb - nodeset.AddNode(node.Path, trienode.NewNodeWithPrev(common.Hash{}, node.SerializedBytes, t.tracer.Get(node.Path))) - } - // Serialize root commitment form - return t.Hash(), nodeset -} - -// NodeIterator implements state.Trie, returning an iterator that returns -// nodes of the trie. Iteration starts at the key after the given start key. -// -// TODO(gballet, rjl493456442) implement it. -func (t *VerkleTrie) NodeIterator(startKey []byte) (NodeIterator, error) { - // TODO(@CPerezz): remove. - return nil, errors.New("not implemented") -} - -// Prove implements state.Trie, constructing a Merkle proof for key. The result -// contains all encoded nodes on the path to the value at key. The value itself -// is also included in the last node and can be retrieved by verifying the proof. -// -// If the trie does not contain a value for key, the returned proof contains all -// nodes of the longest existing prefix of the key (at least the root), ending -// with the node that proves the absence of the key. -// -// TODO(gballet, rjl493456442) implement it. -func (t *VerkleTrie) Prove(key []byte, proofDb ethdb.KeyValueWriter) error { - panic("not implemented") -} - -// Copy returns a deep-copied verkle tree. -func (t *VerkleTrie) Copy() *VerkleTrie { - return &VerkleTrie{ - root: t.root.Copy(), - cache: t.cache, - reader: t.reader, - tracer: t.tracer.Copy(), - } -} - -// IsVerkle indicates if the trie is a Verkle trie. -func (t *VerkleTrie) IsVerkle() bool { - return true -} - -// Proof builds and returns the verkle multiproof for keys, built against -// the pre tree. The post tree is passed in order to add the post values -// to that proof. -func (t *VerkleTrie) Proof(posttrie *VerkleTrie, keys [][]byte) (*verkle.VerkleProof, verkle.StateDiff, error) { - var postroot verkle.VerkleNode - if posttrie != nil { - postroot = posttrie.root - } - proof, _, _, _, err := verkle.MakeVerkleMultiProof(t.root, postroot, keys, t.nodeResolver) - if err != nil { - return nil, nil, err - } - p, kvps, err := verkle.SerializeProof(proof) - if err != nil { - return nil, nil, err - } - return p, kvps, nil -} - -// ChunkedCode represents a sequence of 32-bytes chunks of code (31 bytes of which -// are actual code, and 1 byte is the pushdata offset). -type ChunkedCode []byte - -// Copy the values here so as to avoid an import cycle -const ( - PUSH1 = byte(0x60) - PUSH32 = byte(0x7f) -) - -// ChunkifyCode generates the chunked version of an array representing EVM bytecode -func ChunkifyCode(code []byte) ChunkedCode { - var ( - chunkOffset = 0 // offset in the chunk - chunkCount = len(code) / 31 - codeOffset = 0 // offset in the code - ) - if len(code)%31 != 0 { - chunkCount++ - } - chunks := make([]byte, chunkCount*32) - for i := 0; i < chunkCount; i++ { - // number of bytes to copy, 31 unless the end of the code has been reached. - end := 31 * (i + 1) - if len(code) < end { - end = len(code) - } - copy(chunks[i*32+1:], code[31*i:end]) // copy the code itself - - // chunk offset = taken from the last chunk. - if chunkOffset > 31 { - // skip offset calculation if push data covers the whole chunk - chunks[i*32] = 31 - chunkOffset = 1 - continue - } - chunks[32*i] = byte(chunkOffset) - chunkOffset = 0 - - // Check each instruction and update the offset it should be 0 unless - // a PUSH-N overflows. - for ; codeOffset < end; codeOffset++ { - if code[codeOffset] >= PUSH1 && code[codeOffset] <= PUSH32 { - codeOffset += int(code[codeOffset] - PUSH1 + 1) - if codeOffset+1 >= 31*(i+1) { - codeOffset++ - chunkOffset = codeOffset - 31*(i+1) - break - } - } - } - } - return chunks -} - -// UpdateContractCode implements state.Trie, writing the provided contract code -// into the trie. -// Note that the code-size *must* be already saved by a previous UpdateAccount call. -func (t *VerkleTrie) UpdateContractCode(addr common.Address, codeHash common.Hash, code []byte) error { - var ( - chunks = ChunkifyCode(code) - values [][]byte - key []byte - err error - ) - for i, chunknr := 0, uint64(0); i < len(chunks); i, chunknr = i+32, chunknr+1 { - groupOffset := (chunknr + 128) % 256 - if groupOffset == 0 /* start of new group */ || chunknr == 0 /* first chunk in header group */ { - values = make([][]byte, verkle.NodeWidth) - key = utils.CodeChunkKeyWithEvaluatedAddress(t.cache.Get(addr.Bytes()), uint256.NewInt(chunknr)) - } - values[groupOffset] = chunks[i : i+32] - - if groupOffset == 255 || len(chunks)-i <= 32 { - switch root := t.root.(type) { - case *verkle.InternalNode: - err = root.InsertValuesAtStem(key[:31], values, t.nodeResolver) - if err != nil { - return fmt.Errorf("UpdateContractCode (addr=%x) error: %w", addr[:], err) - } - default: - return errInvalidRootType - } - } - } - return nil -} - -func (t *VerkleTrie) ToDot() string { - return verkle.ToDot(t.root) -} - -func (t *VerkleTrie) nodeResolver(path []byte) ([]byte, error) { - blob, err := t.reader.Node(path, common.Hash{}) - if err != nil { - return nil, err - } - t.tracer.Put(path, blob) - return blob, nil -} - -// Witness returns a set containing all trie nodes that have been accessed. -func (t *VerkleTrie) Witness() map[string][]byte { - panic("not implemented") -} diff --git a/trie/verkle_test.go b/trie/verkle_test.go deleted file mode 100644 index 1832e3db13..0000000000 --- a/trie/verkle_test.go +++ /dev/null @@ -1,173 +0,0 @@ -// Copyright 2023 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 trie - -import ( - "bytes" - "reflect" - "testing" - - "github.com/ethereum/go-ethereum/common" - "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/trie/utils" - "github.com/holiman/uint256" -) - -var ( - accounts = map[common.Address]*types.StateAccount{ - {1}: { - Nonce: 100, - Balance: uint256.NewInt(100), - CodeHash: common.Hash{0x1}.Bytes(), - }, - {2}: { - Nonce: 200, - Balance: uint256.NewInt(200), - CodeHash: common.Hash{0x2}.Bytes(), - }, - } - storages = map[common.Address]map[common.Hash][]byte{ - {1}: { - common.Hash{10}: []byte{10}, - common.Hash{11}: []byte{11}, - common.MaxHash: []byte{0xff}, - }, - {2}: { - common.Hash{20}: []byte{20}, - common.Hash{21}: []byte{21}, - common.MaxHash: []byte{0xff}, - }, - } -) - -func TestVerkleTreeReadWrite(t *testing.T) { - db := newTestDatabase(rawdb.NewMemoryDatabase(), rawdb.PathScheme) - tr, _ := NewVerkleTrie(types.EmptyVerkleHash, db, utils.NewPointCache(100)) - - for addr, acct := range accounts { - if err := tr.UpdateAccount(addr, acct, 0); err != nil { - t.Fatalf("Failed to update account, %v", err) - } - for key, val := range storages[addr] { - if err := tr.UpdateStorage(addr, key.Bytes(), val); err != nil { - t.Fatalf("Failed to update storage, %v", err) - } - } - } - - for addr, acct := range accounts { - stored, err := tr.GetAccount(addr) - if err != nil { - t.Fatalf("Failed to get account, %v", err) - } - if !reflect.DeepEqual(stored, acct) { - t.Fatal("account is not matched") - } - for key, val := range storages[addr] { - stored, err := tr.GetStorage(addr, key.Bytes()) - if err != nil { - t.Fatalf("Failed to get storage, %v", err) - } - if !bytes.Equal(stored, val) { - t.Fatal("storage is not matched") - } - } - } -} - -func TestVerkleRollBack(t *testing.T) { - db := newTestDatabase(rawdb.NewMemoryDatabase(), rawdb.PathScheme) - tr, _ := NewVerkleTrie(types.EmptyVerkleHash, db, utils.NewPointCache(100)) - - for addr, acct := range accounts { - // create more than 128 chunks of code - code := make([]byte, 129*32) - for i := 0; i < len(code); i += 2 { - code[i] = 0x60 - code[i+1] = byte(i % 256) - } - if err := tr.UpdateAccount(addr, acct, len(code)); err != nil { - t.Fatalf("Failed to update account, %v", err) - } - for key, val := range storages[addr] { - if err := tr.UpdateStorage(addr, key.Bytes(), val); err != nil { - t.Fatalf("Failed to update storage, %v", err) - } - } - hash := crypto.Keccak256Hash(code) - if err := tr.UpdateContractCode(addr, hash, code); err != nil { - t.Fatalf("Failed to update contract, %v", err) - } - } - - // Check that things were created - for addr, acct := range accounts { - stored, err := tr.GetAccount(addr) - if err != nil { - t.Fatalf("Failed to get account, %v", err) - } - if !reflect.DeepEqual(stored, acct) { - t.Fatal("account is not matched") - } - for key, val := range storages[addr] { - stored, err := tr.GetStorage(addr, key.Bytes()) - if err != nil { - t.Fatalf("Failed to get storage, %v", err) - } - if !bytes.Equal(stored, val) { - t.Fatal("storage is not matched") - } - } - } - - // ensure there is some code in the 2nd group of the 1st account - keyOf2ndGroup := utils.CodeChunkKeyWithEvaluatedAddress(tr.cache.Get(common.Address{1}.Bytes()), uint256.NewInt(128)) - chunk, err := tr.root.Get(keyOf2ndGroup, nil) - if err != nil { - t.Fatalf("Failed to get account, %v", err) - } - if len(chunk) == 0 { - t.Fatal("account was not created ") - } - - // Rollback first account and check that it is gone - addr1 := common.Address{1} - err = tr.RollBackAccount(addr1) - if err != nil { - t.Fatalf("error rolling back address 1: %v", err) - } - - // ensure the account is gone - stored, err := tr.GetAccount(addr1) - if err != nil { - t.Fatalf("Failed to get account, %v", err) - } - if stored != nil { - t.Fatal("account was not deleted") - } - - // ensure that the last code chunk is also gone from the tree - chunk, err = tr.root.Get(keyOf2ndGroup, nil) - if err != nil { - t.Fatalf("Failed to get account, %v", err) - } - if len(chunk) != 0 { - t.Fatal("account was not deleted") - } -} diff --git a/triedb/pathdb/history_index.go b/triedb/pathdb/history_index.go index 87b6e377af..cc5cd204b4 100644 --- a/triedb/pathdb/history_index.go +++ b/triedb/pathdb/history_index.go @@ -163,12 +163,15 @@ type indexWriter struct { db ethdb.KeyValueReader } -// newIndexWriter constructs the index writer for the specified state. -func newIndexWriter(db ethdb.KeyValueReader, state stateIdent) (*indexWriter, error) { +// newIndexWriter constructs the index writer for the specified state. Additionally, +// it takes an integer as the limit and prunes all existing elements above that ID. +// It's essential as the recovery mechanism after unclean shutdown during the history +// indexing. +func newIndexWriter(db ethdb.KeyValueReader, state stateIdent, limit uint64) (*indexWriter, error) { blob := readStateIndex(state, db) if len(blob) == 0 { desc := newIndexBlockDesc(0) - bw, _ := newBlockWriter(nil, desc) + bw, _ := newBlockWriter(nil, desc, 0 /* useless if the block is empty */) return &indexWriter{ descList: []*indexBlockDesc{desc}, bw: bw, @@ -180,15 +183,27 @@ func newIndexWriter(db ethdb.KeyValueReader, state stateIdent) (*indexWriter, er if err != nil { return nil, err } + // Trim trailing blocks whose elements all exceed the limit. + for i := len(descList) - 1; i > 0 && descList[i].max > limit; i-- { + // The previous block has the elements that exceed the limit, + // therefore the current block can be entirely dropped. + if descList[i-1].max >= limit { + descList = descList[:i] + } + } + // Take the last block for appending new elements lastDesc := descList[len(descList)-1] indexBlock := readStateIndexBlock(state, db, lastDesc.id) - bw, err := newBlockWriter(indexBlock, lastDesc) + + // Construct the writer for the last block. All elements in this block + // that exceed the limit will be truncated. + bw, err := newBlockWriter(indexBlock, lastDesc, limit) if err != nil { return nil, err } return &indexWriter{ descList: descList, - lastID: lastDesc.max, + lastID: bw.last(), bw: bw, state: state, db: db, @@ -221,7 +236,7 @@ func (w *indexWriter) rotate() error { desc = newIndexBlockDesc(w.bw.desc.id + 1) ) w.frozen = append(w.frozen, w.bw) - w.bw, err = newBlockWriter(nil, desc) + w.bw, err = newBlockWriter(nil, desc, 0 /* useless if the block is empty */) if err != nil { return err } @@ -271,13 +286,13 @@ type indexDeleter struct { } // newIndexDeleter constructs the index deleter for the specified state. -func newIndexDeleter(db ethdb.KeyValueReader, state stateIdent) (*indexDeleter, error) { +func newIndexDeleter(db ethdb.KeyValueReader, state stateIdent, limit uint64) (*indexDeleter, error) { blob := readStateIndex(state, db) if len(blob) == 0 { // TODO(rjl493456442) we can probably return an error here, // deleter with no data is meaningless. desc := newIndexBlockDesc(0) - bw, _ := newBlockWriter(nil, desc) + bw, _ := newBlockWriter(nil, desc, 0 /* useless if the block is empty */) return &indexDeleter{ descList: []*indexBlockDesc{desc}, bw: bw, @@ -289,22 +304,34 @@ func newIndexDeleter(db ethdb.KeyValueReader, state stateIdent) (*indexDeleter, if err != nil { return nil, err } + // Trim trailing blocks whose elements all exceed the limit. + for i := len(descList) - 1; i > 0 && descList[i].max > limit; i-- { + // The previous block has the elements that exceed the limit, + // therefore the current block can be entirely dropped. + if descList[i-1].max >= limit { + descList = descList[:i] + } + } + // Take the block for deleting element from lastDesc := descList[len(descList)-1] indexBlock := readStateIndexBlock(state, db, lastDesc.id) - bw, err := newBlockWriter(indexBlock, lastDesc) + + // Construct the writer for the last block. All elements in this block + // that exceed the limit will be truncated. + bw, err := newBlockWriter(indexBlock, lastDesc, limit) if err != nil { return nil, err } return &indexDeleter{ descList: descList, - lastID: lastDesc.max, + lastID: bw.last(), bw: bw, state: state, db: db, }, nil } -// empty returns an flag indicating whether the state index is empty. +// empty returns whether the state index is empty. func (d *indexDeleter) empty() bool { return d.bw.empty() && len(d.descList) == 1 } @@ -337,7 +364,7 @@ func (d *indexDeleter) pop(id uint64) error { // Open the previous block writer for deleting lastDesc := d.descList[len(d.descList)-1] indexBlock := readStateIndexBlock(d.state, d.db, lastDesc.id) - bw, err := newBlockWriter(indexBlock, lastDesc) + bw, err := newBlockWriter(indexBlock, lastDesc, lastDesc.max) if err != nil { return err } diff --git a/triedb/pathdb/history_index_block.go b/triedb/pathdb/history_index_block.go index 7b59c8e882..13f16b4cf3 100644 --- a/triedb/pathdb/history_index_block.go +++ b/triedb/pathdb/history_index_block.go @@ -21,13 +21,15 @@ import ( "errors" "fmt" "math" + + "github.com/ethereum/go-ethereum/log" ) const ( - indexBlockDescSize = 14 // The size of index block descriptor - indexBlockEntriesCap = 4096 // The maximum number of entries can be grouped in a block - indexBlockRestartLen = 256 // The restart interval length of index block - historyIndexBatch = 512 * 1024 // The number of state history indexes for constructing or deleting as batch + indexBlockDescSize = 14 // The size of index block descriptor + indexBlockEntriesCap = 4096 // The maximum number of entries can be grouped in a block + indexBlockRestartLen = 256 // The restart interval length of index block + historyIndexBatch = 8 * 1024 * 1024 // The number of state history indexes for constructing or deleting as batch ) // indexBlockDesc represents a descriptor for an index block, which contains a @@ -180,7 +182,11 @@ type blockWriter struct { data []byte // Aggregated encoded data slice } -func newBlockWriter(blob []byte, desc *indexBlockDesc) (*blockWriter, error) { +// newBlockWriter constructs a block writer. In addition to the existing data +// and block description, it takes an element ID and prunes all existing elements +// above that ID. It's essential as the recovery mechanism after unclean shutdown +// during the history indexing. +func newBlockWriter(blob []byte, desc *indexBlockDesc, limit uint64) (*blockWriter, error) { if len(blob) == 0 { return &blockWriter{ desc: desc, @@ -191,11 +197,22 @@ func newBlockWriter(blob []byte, desc *indexBlockDesc) (*blockWriter, error) { if err != nil { return nil, err } - return &blockWriter{ + writer := &blockWriter{ desc: desc, restarts: restarts, data: data, // safe to own the slice - }, nil + } + var trimmed int + for !writer.empty() && writer.last() > limit { + if err := writer.pop(writer.last()); err != nil { + return nil, err + } + trimmed += 1 + } + if trimmed > 0 { + log.Debug("Truncated extraneous elements", "count", trimmed, "limit", limit) + } + return writer, nil } // append adds a new element to the block. The new element must be greater than @@ -271,6 +288,7 @@ func (b *blockWriter) sectionLast(section int) uint64 { // sectionSearch looks up the specified value in the given section, // the position and the preceding value will be returned if found. +// It assumes that the preceding element exists in the section. func (b *blockWriter) sectionSearch(section int, n uint64) (found bool, prev uint64, pos int) { b.scanSection(section, func(v uint64, p int) bool { if n == v { @@ -295,7 +313,6 @@ func (b *blockWriter) pop(id uint64) error { } // If there is only one entry left, the entire block should be reset if b.desc.entries == 1 { - //b.desc.min = 0 b.desc.max = 0 b.desc.entries = 0 b.restarts = nil @@ -331,6 +348,15 @@ func (b *blockWriter) full() bool { return b.desc.full() } +// last returns the last element in the block. It should only be called when +// writer is not empty, otherwise the returned data is meaningless. +func (b *blockWriter) last() uint64 { + if b.empty() { + return 0 + } + return b.desc.max +} + // finish finalizes the index block encoding by appending the encoded restart points // and the restart counter to the end of the block. // diff --git a/triedb/pathdb/history_index_block_test.go b/triedb/pathdb/history_index_block_test.go index c251cea2ec..f8c6d3ab87 100644 --- a/triedb/pathdb/history_index_block_test.go +++ b/triedb/pathdb/history_index_block_test.go @@ -28,7 +28,7 @@ func TestBlockReaderBasic(t *testing.T) { elements := []uint64{ 1, 5, 10, 11, 20, } - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) for i := 0; i < len(elements); i++ { bw.append(elements[i]) } @@ -66,7 +66,7 @@ func TestBlockReaderLarge(t *testing.T) { } slices.Sort(elements) - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) for i := 0; i < len(elements); i++ { bw.append(elements[i]) } @@ -95,7 +95,7 @@ func TestBlockReaderLarge(t *testing.T) { } func TestBlockWriterBasic(t *testing.T) { - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) if !bw.empty() { t.Fatal("expected empty block") } @@ -103,11 +103,13 @@ func TestBlockWriterBasic(t *testing.T) { if err := bw.append(1); err == nil { t.Fatal("out-of-order insertion is not expected") } + var maxElem uint64 for i := 0; i < 10; i++ { bw.append(uint64(i + 3)) + maxElem = uint64(i + 3) } - bw, err := newBlockWriter(bw.finish(), newIndexBlockDesc(0)) + bw, err := newBlockWriter(bw.finish(), newIndexBlockDesc(0), maxElem) if err != nil { t.Fatalf("Failed to construct the block writer, %v", err) } @@ -119,8 +121,71 @@ func TestBlockWriterBasic(t *testing.T) { bw.finish() } +func TestBlockWriterWithLimit(t *testing.T) { + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) + + var maxElem uint64 + for i := 0; i < indexBlockRestartLen*2; i++ { + bw.append(uint64(i + 1)) + maxElem = uint64(i + 1) + } + + suites := []struct { + limit uint64 + expMax uint64 + }{ + // nothing to truncate + { + maxElem, maxElem, + }, + // truncate the last element + { + maxElem - 1, maxElem - 1, + }, + // truncation around the restart boundary + { + uint64(indexBlockRestartLen + 1), + uint64(indexBlockRestartLen + 1), + }, + // truncation around the restart boundary + { + uint64(indexBlockRestartLen), + uint64(indexBlockRestartLen), + }, + { + uint64(1), uint64(1), + }, + // truncate the entire block, it's in theory invalid + { + uint64(0), uint64(0), + }, + } + for i, suite := range suites { + desc := *bw.desc + block, err := newBlockWriter(bw.finish(), &desc, suite.limit) + if err != nil { + t.Fatalf("Failed to construct the block writer, %v", err) + } + if block.desc.max != suite.expMax { + t.Fatalf("Test %d, unexpected max value, got %d, want %d", i, block.desc.max, suite.expMax) + } + + // Re-fill the elements + var maxElem uint64 + for elem := suite.limit + 1; elem < indexBlockRestartLen*4; elem++ { + if err := block.append(elem); err != nil { + t.Fatalf("Failed to append value %d: %v", elem, err) + } + maxElem = elem + } + if block.desc.max != maxElem { + t.Fatalf("Test %d, unexpected max value, got %d, want %d", i, block.desc.max, maxElem) + } + } +} + func TestBlockWriterDelete(t *testing.T) { - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) for i := 0; i < 10; i++ { bw.append(uint64(i + 1)) } @@ -147,7 +212,7 @@ func TestBlcokWriterDeleteWithData(t *testing.T) { elements := []uint64{ 1, 5, 10, 11, 20, } - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) for i := 0; i < len(elements); i++ { bw.append(elements[i]) } @@ -158,7 +223,7 @@ func TestBlcokWriterDeleteWithData(t *testing.T) { max: 20, entries: 5, } - bw, err := newBlockWriter(bw.finish(), desc) + bw, err := newBlockWriter(bw.finish(), desc, elements[len(elements)-1]) if err != nil { t.Fatalf("Failed to construct block writer %v", err) } @@ -201,15 +266,18 @@ func TestBlcokWriterDeleteWithData(t *testing.T) { } func TestCorruptedIndexBlock(t *testing.T) { - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) + + var maxElem uint64 for i := 0; i < 10; i++ { bw.append(uint64(i + 1)) + maxElem = uint64(i + 1) } buf := bw.finish() // Mutate the buffer manually buf[len(buf)-1]++ - _, err := newBlockWriter(buf, newIndexBlockDesc(0)) + _, err := newBlockWriter(buf, newIndexBlockDesc(0), maxElem) if err == nil { t.Fatal("Corrupted index block data is not detected") } @@ -218,7 +286,7 @@ func TestCorruptedIndexBlock(t *testing.T) { // BenchmarkParseIndexBlock benchmarks the performance of parseIndexBlock. func BenchmarkParseIndexBlock(b *testing.B) { // Generate a realistic index block blob - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) for i := 0; i < 4096; i++ { bw.append(uint64(i * 2)) } @@ -238,13 +306,15 @@ func BenchmarkBlockWriterAppend(b *testing.B) { b.ReportAllocs() b.ResetTimer() - desc := newIndexBlockDesc(0) - writer, _ := newBlockWriter(nil, desc) + var blockID uint32 + desc := newIndexBlockDesc(blockID) + writer, _ := newBlockWriter(nil, desc, 0) for i := 0; i < b.N; i++ { if writer.full() { - desc = newIndexBlockDesc(0) - writer, _ = newBlockWriter(nil, desc) + blockID += 1 + desc = newIndexBlockDesc(blockID) + writer, _ = newBlockWriter(nil, desc, 0) } if err := writer.append(writer.desc.max + 1); err != nil { b.Error(err) diff --git a/triedb/pathdb/history_index_iterator_test.go b/triedb/pathdb/history_index_iterator_test.go index da60dc6e8f..f0dd3fee4a 100644 --- a/triedb/pathdb/history_index_iterator_test.go +++ b/triedb/pathdb/history_index_iterator_test.go @@ -33,7 +33,7 @@ func makeTestIndexBlock(count int) ([]byte, []uint64) { marks = make(map[uint64]bool) elements []uint64 ) - bw, _ := newBlockWriter(nil, newIndexBlockDesc(0)) + bw, _ := newBlockWriter(nil, newIndexBlockDesc(0), 0) for i := 0; i < count; i++ { n := uint64(rand.Uint32()) if marks[n] { @@ -67,7 +67,7 @@ func makeTestIndexBlocks(db ethdb.KeyValueStore, stateIdent stateIdent, count in } sort.Slice(elements, func(i, j int) bool { return elements[i] < elements[j] }) - iw, _ := newIndexWriter(db, stateIdent) + iw, _ := newIndexWriter(db, stateIdent, 0) for i := 0; i < len(elements); i++ { iw.append(elements[i]) } diff --git a/triedb/pathdb/history_index_test.go b/triedb/pathdb/history_index_test.go index be9b7c4049..42cb04b001 100644 --- a/triedb/pathdb/history_index_test.go +++ b/triedb/pathdb/history_index_test.go @@ -33,7 +33,7 @@ func TestIndexReaderBasic(t *testing.T) { 1, 5, 10, 11, 20, } db := rawdb.NewMemoryDatabase() - bw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa})) + bw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), 0) for i := 0; i < len(elements); i++ { bw.append(elements[i]) } @@ -75,7 +75,7 @@ func TestIndexReaderLarge(t *testing.T) { slices.Sort(elements) db := rawdb.NewMemoryDatabase() - bw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa})) + bw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), 0) for i := 0; i < len(elements); i++ { bw.append(elements[i]) } @@ -122,19 +122,21 @@ func TestEmptyIndexReader(t *testing.T) { func TestIndexWriterBasic(t *testing.T) { db := rawdb.NewMemoryDatabase() - iw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa})) + iw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), 0) iw.append(2) if err := iw.append(1); err == nil { t.Fatal("out-of-order insertion is not expected") } + var maxElem uint64 for i := 0; i < 10; i++ { iw.append(uint64(i + 3)) + maxElem = uint64(i + 3) } batch := db.NewBatch() iw.finish(batch) batch.Write() - iw, err := newIndexWriter(db, newAccountIdent(common.Hash{0xa})) + iw, err := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), maxElem) if err != nil { t.Fatalf("Failed to construct the block writer, %v", err) } @@ -146,18 +148,87 @@ func TestIndexWriterBasic(t *testing.T) { iw.finish(db.NewBatch()) } -func TestIndexWriterDelete(t *testing.T) { +func TestIndexWriterWithLimit(t *testing.T) { db := rawdb.NewMemoryDatabase() - iw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa})) + iw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), 0) + + var maxElem uint64 + for i := 0; i < indexBlockEntriesCap*2; i++ { + iw.append(uint64(i + 1)) + maxElem = uint64(i + 1) + } + batch := db.NewBatch() + iw.finish(batch) + batch.Write() + + suites := []struct { + limit uint64 + expMax uint64 + }{ + // nothing to truncate + { + maxElem, maxElem, + }, + // truncate the last element + { + maxElem - 1, maxElem - 1, + }, + // truncation around the block boundary + { + uint64(indexBlockEntriesCap + 1), + uint64(indexBlockEntriesCap + 1), + }, + // truncation around the block boundary + { + uint64(indexBlockEntriesCap), + uint64(indexBlockEntriesCap), + }, + { + uint64(1), uint64(1), + }, + // truncate the entire index, it's in theory invalid + { + uint64(0), uint64(0), + }, + } + for i, suite := range suites { + iw, err := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), suite.limit) + if err != nil { + t.Fatalf("Failed to construct the index writer, %v", err) + } + if iw.lastID != suite.expMax { + t.Fatalf("Test %d, unexpected max value, got %d, want %d", i, iw.lastID, suite.expMax) + } + + // Re-fill the elements + var maxElem uint64 + for elem := suite.limit + 1; elem < indexBlockEntriesCap*4; elem++ { + if err := iw.append(elem); err != nil { + t.Fatalf("Failed to append value %d: %v", elem, err) + } + maxElem = elem + } + if iw.lastID != maxElem { + t.Fatalf("Test %d, unexpected max value, got %d, want %d", i, iw.lastID, maxElem) + } + } +} + +func TestIndexDeleterBasic(t *testing.T) { + db := rawdb.NewMemoryDatabase() + iw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), 0) + + var maxElem uint64 for i := 0; i < indexBlockEntriesCap*4; i++ { iw.append(uint64(i + 1)) + maxElem = uint64(i + 1) } batch := db.NewBatch() iw.finish(batch) batch.Write() // Delete unknown id, the request should be rejected - id, _ := newIndexDeleter(db, newAccountIdent(common.Hash{0xa})) + id, _ := newIndexDeleter(db, newAccountIdent(common.Hash{0xa}), maxElem) if err := id.pop(indexBlockEntriesCap * 5); err == nil { t.Fatal("Expect error to occur for unknown id") } @@ -168,10 +239,66 @@ func TestIndexWriterDelete(t *testing.T) { if id.lastID != uint64(i-1) { t.Fatalf("Unexpected lastID, want: %d, got: %d", uint64(i-1), iw.lastID) } - if rand.Intn(10) == 0 { - batch := db.NewBatch() - id.finish(batch) - batch.Write() + } +} + +func TestIndexDeleterWithLimit(t *testing.T) { + db := rawdb.NewMemoryDatabase() + iw, _ := newIndexWriter(db, newAccountIdent(common.Hash{0xa}), 0) + + var maxElem uint64 + for i := 0; i < indexBlockEntriesCap*2; i++ { + iw.append(uint64(i + 1)) + maxElem = uint64(i + 1) + } + batch := db.NewBatch() + iw.finish(batch) + batch.Write() + + suites := []struct { + limit uint64 + expMax uint64 + }{ + // nothing to truncate + { + maxElem, maxElem, + }, + // truncate the last element + { + maxElem - 1, maxElem - 1, + }, + // truncation around the block boundary + { + uint64(indexBlockEntriesCap + 1), + uint64(indexBlockEntriesCap + 1), + }, + // truncation around the block boundary + { + uint64(indexBlockEntriesCap), + uint64(indexBlockEntriesCap), + }, + { + uint64(1), uint64(1), + }, + // truncate the entire index, it's in theory invalid + { + uint64(0), uint64(0), + }, + } + for i, suite := range suites { + id, err := newIndexDeleter(db, newAccountIdent(common.Hash{0xa}), suite.limit) + if err != nil { + t.Fatalf("Failed to construct the index writer, %v", err) + } + if id.lastID != suite.expMax { + t.Fatalf("Test %d, unexpected max value, got %d, want %d", i, id.lastID, suite.expMax) + } + + // Keep removing elements + for elem := id.lastID; elem > 0; elem-- { + if err := id.pop(elem); err != nil { + t.Fatalf("Failed to pop value %d: %v", elem, err) + } } } } diff --git a/triedb/pathdb/history_indexer.go b/triedb/pathdb/history_indexer.go index 893ccd6523..9af7a96dc6 100644 --- a/triedb/pathdb/history_indexer.go +++ b/triedb/pathdb/history_indexer.go @@ -40,11 +40,6 @@ const ( stateHistoryIndexVersion = stateHistoryIndexV0 // the current state index version trienodeHistoryIndexV0 = uint8(0) // initial version of trienode index structure trienodeHistoryIndexVersion = trienodeHistoryIndexV0 // the current trienode index version - - // estimations for calculating the batch size for atomic database commit - estimatedStateHistoryIndexSize = 3 // The average size of each state history index entry is approximately 2–3 bytes - estimatedTrienodeHistoryIndexSize = 3 // The average size of each trienode history index entry is approximately 2-3 bytes - estimatedIndexBatchSizeFactor = 32 // The factor counts for the write amplification for each entry ) // indexVersion returns the latest index version for the given history type. @@ -155,22 +150,6 @@ func (b *batchIndexer) process(h history, id uint64) error { return b.finish(false) } -// makeBatch constructs a database batch based on the number of pending entries. -// The batch size is roughly estimated to minimize repeated resizing rounds, -// as accurately predicting the exact size is technically challenging. -func (b *batchIndexer) makeBatch() ethdb.Batch { - var size int - switch b.typ { - case typeStateHistory: - size = estimatedStateHistoryIndexSize - case typeTrienodeHistory: - size = estimatedTrienodeHistoryIndexSize - default: - panic(fmt.Sprintf("unknown history type %d", b.typ)) - } - return b.db.NewBatchWithSize(size * estimatedIndexBatchSizeFactor * b.pending) -} - // finish writes the accumulated state indexes into the disk if either the // memory limitation is reached or it's requested forcibly. func (b *batchIndexer) finish(force bool) error { @@ -181,17 +160,38 @@ func (b *batchIndexer) finish(force bool) error { return nil } var ( - batch = b.makeBatch() - batchMu sync.RWMutex - start = time.Now() - eg errgroup.Group + start = time.Now() + eg errgroup.Group + + batch = b.db.NewBatchWithSize(ethdb.IdealBatchSize) + batchSize int + batchMu sync.RWMutex + + writeBatch = func(fn func(batch ethdb.Batch)) error { + batchMu.Lock() + defer batchMu.Unlock() + + fn(batch) + if batch.ValueSize() >= ethdb.IdealBatchSize { + batchSize += batch.ValueSize() + if err := batch.Write(); err != nil { + return err + } + batch.Reset() + } + return nil + } ) eg.SetLimit(runtime.NumCPU()) + var indexed uint64 + if metadata := loadIndexMetadata(b.db, b.typ); metadata != nil { + indexed = metadata.Last + } for ident, list := range b.index { eg.Go(func() error { if !b.delete { - iw, err := newIndexWriter(b.db, ident) + iw, err := newIndexWriter(b.db, ident, indexed) if err != nil { return err } @@ -200,11 +200,11 @@ func (b *batchIndexer) finish(force bool) error { return err } } - batchMu.Lock() - iw.finish(batch) - batchMu.Unlock() + return writeBatch(func(batch ethdb.Batch) { + iw.finish(batch) + }) } else { - id, err := newIndexDeleter(b.db, ident) + id, err := newIndexDeleter(b.db, ident, indexed) if err != nil { return err } @@ -213,11 +213,10 @@ func (b *batchIndexer) finish(force bool) error { return err } } - batchMu.Lock() - id.finish(batch) - batchMu.Unlock() + return writeBatch(func(batch ethdb.Batch) { + id.finish(batch) + }) } - return nil }) } if err := eg.Wait(); err != nil { @@ -233,10 +232,12 @@ func (b *batchIndexer) finish(force bool) error { storeIndexMetadata(batch, b.typ, b.lastID-1) } } + batchSize += batch.ValueSize() + if err := batch.Write(); err != nil { return err } - log.Debug("Committed batch indexer", "type", b.typ, "entries", len(b.index), "records", b.pending, "elapsed", common.PrettyDuration(time.Since(start))) + log.Debug("Committed batch indexer", "type", b.typ, "entries", len(b.index), "records", b.pending, "size", common.StorageSize(batchSize), "elapsed", common.PrettyDuration(time.Since(start))) b.pending = 0 b.index = make(map[stateIdent][]uint64) return nil