2b57a27d9e
This PR changes the API so that uint64 is used for fork timestamps. It's a good choice because types.Header also uses uint64 for time. Co-authored-by: Felix Lange <fjl@twurst.com>
557 lines
18 KiB
Go
557 lines
18 KiB
Go
// Copyright 2016 The go-ethereum Authors
|
|
// This file is part of the go-ethereum library.
|
|
//
|
|
// The go-ethereum library is free software: you can redistribute it and/or modify
|
|
// it under the terms of the GNU Lesser General Public License as published by
|
|
// the Free Software Foundation, either version 3 of the License, or
|
|
// (at your option) any later version.
|
|
//
|
|
// The go-ethereum library is distributed in the hope that it will be useful,
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
// GNU Lesser General Public License for more details.
|
|
//
|
|
// You should have received a copy of the GNU Lesser General Public License
|
|
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
package light
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"math/big"
|
|
"sync"
|
|
"time"
|
|
|
|
"github.com/ethereum/go-ethereum/common"
|
|
"github.com/ethereum/go-ethereum/core"
|
|
"github.com/ethereum/go-ethereum/core/rawdb"
|
|
"github.com/ethereum/go-ethereum/core/state"
|
|
"github.com/ethereum/go-ethereum/core/txpool"
|
|
"github.com/ethereum/go-ethereum/core/types"
|
|
"github.com/ethereum/go-ethereum/ethdb"
|
|
"github.com/ethereum/go-ethereum/event"
|
|
"github.com/ethereum/go-ethereum/log"
|
|
"github.com/ethereum/go-ethereum/params"
|
|
)
|
|
|
|
const (
|
|
// chainHeadChanSize is the size of channel listening to ChainHeadEvent.
|
|
chainHeadChanSize = 10
|
|
)
|
|
|
|
// txPermanent is the number of mined blocks after a mined transaction is
|
|
// considered permanent and no rollback is expected
|
|
var txPermanent = uint64(500)
|
|
|
|
// TxPool implements the transaction pool for light clients, which keeps track
|
|
// of the status of locally created transactions, detecting if they are included
|
|
// in a block (mined) or rolled back. There are no queued transactions since we
|
|
// always receive all locally signed transactions in the same order as they are
|
|
// created.
|
|
type TxPool struct {
|
|
config *params.ChainConfig
|
|
signer types.Signer
|
|
quit chan bool
|
|
txFeed event.Feed
|
|
scope event.SubscriptionScope
|
|
chainHeadCh chan core.ChainHeadEvent
|
|
chainHeadSub event.Subscription
|
|
mu sync.RWMutex
|
|
chain *LightChain
|
|
odr OdrBackend
|
|
chainDb ethdb.Database
|
|
relay TxRelayBackend
|
|
head common.Hash
|
|
nonce map[common.Address]uint64 // "pending" nonce
|
|
pending map[common.Hash]*types.Transaction // pending transactions by tx hash
|
|
mined map[common.Hash][]*types.Transaction // mined transactions by block hash
|
|
clearIdx uint64 // earliest block nr that can contain mined tx info
|
|
|
|
istanbul bool // Fork indicator whether we are in the istanbul stage.
|
|
eip2718 bool // Fork indicator whether we are in the eip2718 stage.
|
|
shanghai bool // Fork indicator whether we are in the shanghai stage.
|
|
}
|
|
|
|
// TxRelayBackend provides an interface to the mechanism that forwards transactions to the
|
|
// ETH network. The implementations of the functions should be non-blocking.
|
|
//
|
|
// Send instructs backend to forward new transactions NewHead notifies backend about a new
|
|
// head after processed by the tx pool, including mined and rolled back transactions since
|
|
// the last event.
|
|
//
|
|
// Discard notifies backend about transactions that should be discarded either because
|
|
// they have been replaced by a re-send or because they have been mined long ago and no
|
|
// rollback is expected.
|
|
type TxRelayBackend interface {
|
|
Send(txs types.Transactions)
|
|
NewHead(head common.Hash, mined []common.Hash, rollback []common.Hash)
|
|
Discard(hashes []common.Hash)
|
|
}
|
|
|
|
// NewTxPool creates a new light transaction pool
|
|
func NewTxPool(config *params.ChainConfig, chain *LightChain, relay TxRelayBackend) *TxPool {
|
|
pool := &TxPool{
|
|
config: config,
|
|
signer: types.LatestSigner(config),
|
|
nonce: make(map[common.Address]uint64),
|
|
pending: make(map[common.Hash]*types.Transaction),
|
|
mined: make(map[common.Hash][]*types.Transaction),
|
|
quit: make(chan bool),
|
|
chainHeadCh: make(chan core.ChainHeadEvent, chainHeadChanSize),
|
|
chain: chain,
|
|
relay: relay,
|
|
odr: chain.Odr(),
|
|
chainDb: chain.Odr().Database(),
|
|
head: chain.CurrentHeader().Hash(),
|
|
clearIdx: chain.CurrentHeader().Number.Uint64(),
|
|
}
|
|
// Subscribe events from blockchain
|
|
pool.chainHeadSub = pool.chain.SubscribeChainHeadEvent(pool.chainHeadCh)
|
|
go pool.eventLoop()
|
|
|
|
return pool
|
|
}
|
|
|
|
// currentState returns the light state of the current head header
|
|
func (pool *TxPool) currentState(ctx context.Context) *state.StateDB {
|
|
return NewState(ctx, pool.chain.CurrentHeader(), pool.odr)
|
|
}
|
|
|
|
// GetNonce returns the "pending" nonce of a given address. It always queries
|
|
// the nonce belonging to the latest header too in order to detect if another
|
|
// client using the same key sent a transaction.
|
|
func (pool *TxPool) GetNonce(ctx context.Context, addr common.Address) (uint64, error) {
|
|
state := pool.currentState(ctx)
|
|
nonce := state.GetNonce(addr)
|
|
if state.Error() != nil {
|
|
return 0, state.Error()
|
|
}
|
|
sn, ok := pool.nonce[addr]
|
|
if ok && sn > nonce {
|
|
nonce = sn
|
|
}
|
|
if !ok || sn < nonce {
|
|
pool.nonce[addr] = nonce
|
|
}
|
|
return nonce, nil
|
|
}
|
|
|
|
// txStateChanges stores the recent changes between pending/mined states of
|
|
// transactions. True means mined, false means rolled back, no entry means no change
|
|
type txStateChanges map[common.Hash]bool
|
|
|
|
// setState sets the status of a tx to either recently mined or recently rolled back
|
|
func (txc txStateChanges) setState(txHash common.Hash, mined bool) {
|
|
val, ent := txc[txHash]
|
|
if ent && (val != mined) {
|
|
delete(txc, txHash)
|
|
} else {
|
|
txc[txHash] = mined
|
|
}
|
|
}
|
|
|
|
// getLists creates lists of mined and rolled back tx hashes
|
|
func (txc txStateChanges) getLists() (mined []common.Hash, rollback []common.Hash) {
|
|
for hash, val := range txc {
|
|
if val {
|
|
mined = append(mined, hash)
|
|
} else {
|
|
rollback = append(rollback, hash)
|
|
}
|
|
}
|
|
return
|
|
}
|
|
|
|
// checkMinedTxs checks newly added blocks for the currently pending transactions
|
|
// and marks them as mined if necessary. It also stores block position in the db
|
|
// and adds them to the received txStateChanges map.
|
|
func (pool *TxPool) checkMinedTxs(ctx context.Context, hash common.Hash, number uint64, txc txStateChanges) error {
|
|
// If no transactions are pending, we don't care about anything
|
|
if len(pool.pending) == 0 {
|
|
return nil
|
|
}
|
|
block, err := GetBlock(ctx, pool.odr, hash, number)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
// Gather all the local transaction mined in this block
|
|
list := pool.mined[hash]
|
|
for _, tx := range block.Transactions() {
|
|
if _, ok := pool.pending[tx.Hash()]; ok {
|
|
list = append(list, tx)
|
|
}
|
|
}
|
|
// If some transactions have been mined, write the needed data to disk and update
|
|
if list != nil {
|
|
// Retrieve all the receipts belonging to this block and write the lookup table
|
|
if _, err := GetBlockReceipts(ctx, pool.odr, hash, number); err != nil { // ODR caches, ignore results
|
|
return err
|
|
}
|
|
rawdb.WriteTxLookupEntriesByBlock(pool.chainDb, block)
|
|
|
|
// Update the transaction pool's state
|
|
for _, tx := range list {
|
|
delete(pool.pending, tx.Hash())
|
|
txc.setState(tx.Hash(), true)
|
|
}
|
|
pool.mined[hash] = list
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// rollbackTxs marks the transactions contained in recently rolled back blocks
|
|
// as rolled back. It also removes any positional lookup entries.
|
|
func (pool *TxPool) rollbackTxs(hash common.Hash, txc txStateChanges) {
|
|
batch := pool.chainDb.NewBatch()
|
|
if list, ok := pool.mined[hash]; ok {
|
|
for _, tx := range list {
|
|
txHash := tx.Hash()
|
|
rawdb.DeleteTxLookupEntry(batch, txHash)
|
|
pool.pending[txHash] = tx
|
|
txc.setState(txHash, false)
|
|
}
|
|
delete(pool.mined, hash)
|
|
}
|
|
batch.Write()
|
|
}
|
|
|
|
// reorgOnNewHead sets a new head header, processing (and rolling back if necessary)
|
|
// the blocks since the last known head and returns a txStateChanges map containing
|
|
// the recently mined and rolled back transaction hashes. If an error (context
|
|
// timeout) occurs during checking new blocks, it leaves the locally known head
|
|
// at the latest checked block and still returns a valid txStateChanges, making it
|
|
// possible to continue checking the missing blocks at the next chain head event
|
|
func (pool *TxPool) reorgOnNewHead(ctx context.Context, newHeader *types.Header) (txStateChanges, error) {
|
|
txc := make(txStateChanges)
|
|
oldh := pool.chain.GetHeaderByHash(pool.head)
|
|
newh := newHeader
|
|
// find common ancestor, create list of rolled back and new block hashes
|
|
var oldHashes, newHashes []common.Hash
|
|
for oldh.Hash() != newh.Hash() {
|
|
if oldh.Number.Uint64() >= newh.Number.Uint64() {
|
|
oldHashes = append(oldHashes, oldh.Hash())
|
|
oldh = pool.chain.GetHeader(oldh.ParentHash, oldh.Number.Uint64()-1)
|
|
}
|
|
if oldh.Number.Uint64() < newh.Number.Uint64() {
|
|
newHashes = append(newHashes, newh.Hash())
|
|
newh = pool.chain.GetHeader(newh.ParentHash, newh.Number.Uint64()-1)
|
|
if newh == nil {
|
|
// happens when CHT syncing, nothing to do
|
|
newh = oldh
|
|
}
|
|
}
|
|
}
|
|
if oldh.Number.Uint64() < pool.clearIdx {
|
|
pool.clearIdx = oldh.Number.Uint64()
|
|
}
|
|
// roll back old blocks
|
|
for _, hash := range oldHashes {
|
|
pool.rollbackTxs(hash, txc)
|
|
}
|
|
pool.head = oldh.Hash()
|
|
// check mined txs of new blocks (array is in reversed order)
|
|
for i := len(newHashes) - 1; i >= 0; i-- {
|
|
hash := newHashes[i]
|
|
if err := pool.checkMinedTxs(ctx, hash, newHeader.Number.Uint64()-uint64(i), txc); err != nil {
|
|
return txc, err
|
|
}
|
|
pool.head = hash
|
|
}
|
|
|
|
// clear old mined tx entries of old blocks
|
|
if idx := newHeader.Number.Uint64(); idx > pool.clearIdx+txPermanent {
|
|
idx2 := idx - txPermanent
|
|
if len(pool.mined) > 0 {
|
|
for i := pool.clearIdx; i < idx2; i++ {
|
|
hash := rawdb.ReadCanonicalHash(pool.chainDb, i)
|
|
if list, ok := pool.mined[hash]; ok {
|
|
hashes := make([]common.Hash, len(list))
|
|
for i, tx := range list {
|
|
hashes[i] = tx.Hash()
|
|
}
|
|
pool.relay.Discard(hashes)
|
|
delete(pool.mined, hash)
|
|
}
|
|
}
|
|
}
|
|
pool.clearIdx = idx2
|
|
}
|
|
|
|
return txc, nil
|
|
}
|
|
|
|
// blockCheckTimeout is the time limit for checking new blocks for mined
|
|
// transactions. Checking resumes at the next chain head event if timed out.
|
|
const blockCheckTimeout = time.Second * 3
|
|
|
|
// eventLoop processes chain head events and also notifies the tx relay backend
|
|
// about the new head hash and tx state changes
|
|
func (pool *TxPool) eventLoop() {
|
|
for {
|
|
select {
|
|
case ev := <-pool.chainHeadCh:
|
|
pool.setNewHead(ev.Block.Header())
|
|
// hack in order to avoid hogging the lock; this part will
|
|
// be replaced by a subsequent PR.
|
|
time.Sleep(time.Millisecond)
|
|
|
|
// System stopped
|
|
case <-pool.chainHeadSub.Err():
|
|
return
|
|
}
|
|
}
|
|
}
|
|
|
|
func (pool *TxPool) setNewHead(head *types.Header) {
|
|
pool.mu.Lock()
|
|
defer pool.mu.Unlock()
|
|
|
|
ctx, cancel := context.WithTimeout(context.Background(), blockCheckTimeout)
|
|
defer cancel()
|
|
|
|
txc, _ := pool.reorgOnNewHead(ctx, head)
|
|
m, r := txc.getLists()
|
|
pool.relay.NewHead(pool.head, m, r)
|
|
|
|
// Update fork indicator by next pending block number
|
|
next := new(big.Int).Add(head.Number, big.NewInt(1))
|
|
pool.istanbul = pool.config.IsIstanbul(next)
|
|
pool.eip2718 = pool.config.IsBerlin(next)
|
|
pool.shanghai = pool.config.IsShanghai(uint64(time.Now().Unix()))
|
|
}
|
|
|
|
// Stop stops the light transaction pool
|
|
func (pool *TxPool) Stop() {
|
|
// Unsubscribe all subscriptions registered from txpool
|
|
pool.scope.Close()
|
|
// Unsubscribe subscriptions registered from blockchain
|
|
pool.chainHeadSub.Unsubscribe()
|
|
close(pool.quit)
|
|
log.Info("Transaction pool stopped")
|
|
}
|
|
|
|
// SubscribeNewTxsEvent registers a subscription of core.NewTxsEvent and
|
|
// starts sending event to the given channel.
|
|
func (pool *TxPool) SubscribeNewTxsEvent(ch chan<- core.NewTxsEvent) event.Subscription {
|
|
return pool.scope.Track(pool.txFeed.Subscribe(ch))
|
|
}
|
|
|
|
// Stats returns the number of currently pending (locally created) transactions
|
|
func (pool *TxPool) Stats() (pending int) {
|
|
pool.mu.RLock()
|
|
defer pool.mu.RUnlock()
|
|
|
|
pending = len(pool.pending)
|
|
return
|
|
}
|
|
|
|
// validateTx checks whether a transaction is valid according to the consensus rules.
|
|
func (pool *TxPool) validateTx(ctx context.Context, tx *types.Transaction) error {
|
|
// Validate sender
|
|
var (
|
|
from common.Address
|
|
err error
|
|
)
|
|
|
|
// Validate the transaction sender and it's sig. Throw
|
|
// if the from fields is invalid.
|
|
if from, err = types.Sender(pool.signer, tx); err != nil {
|
|
return txpool.ErrInvalidSender
|
|
}
|
|
// Last but not least check for nonce errors
|
|
currentState := pool.currentState(ctx)
|
|
if n := currentState.GetNonce(from); n > tx.Nonce() {
|
|
return core.ErrNonceTooLow
|
|
}
|
|
|
|
// Check the transaction doesn't exceed the current
|
|
// block limit gas.
|
|
header := pool.chain.GetHeaderByHash(pool.head)
|
|
if header.GasLimit < tx.Gas() {
|
|
return txpool.ErrGasLimit
|
|
}
|
|
|
|
// Transactions can't be negative. This may never happen
|
|
// using RLP decoded transactions but may occur if you create
|
|
// a transaction using the RPC for example.
|
|
if tx.Value().Sign() < 0 {
|
|
return txpool.ErrNegativeValue
|
|
}
|
|
|
|
// Transactor should have enough funds to cover the costs
|
|
// cost == V + GP * GL
|
|
if b := currentState.GetBalance(from); b.Cmp(tx.Cost()) < 0 {
|
|
return core.ErrInsufficientFunds
|
|
}
|
|
|
|
// Should supply enough intrinsic gas
|
|
gas, err := core.IntrinsicGas(tx.Data(), tx.AccessList(), tx.To() == nil, true, pool.istanbul, pool.shanghai)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
if tx.Gas() < gas {
|
|
return core.ErrIntrinsicGas
|
|
}
|
|
return currentState.Error()
|
|
}
|
|
|
|
// add validates a new transaction and sets its state pending if processable.
|
|
// It also updates the locally stored nonce if necessary.
|
|
func (pool *TxPool) add(ctx context.Context, tx *types.Transaction) error {
|
|
hash := tx.Hash()
|
|
|
|
if pool.pending[hash] != nil {
|
|
return fmt.Errorf("known transaction (%x)", hash[:4])
|
|
}
|
|
err := pool.validateTx(ctx, tx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if _, ok := pool.pending[hash]; !ok {
|
|
pool.pending[hash] = tx
|
|
|
|
nonce := tx.Nonce() + 1
|
|
|
|
addr, _ := types.Sender(pool.signer, tx)
|
|
if nonce > pool.nonce[addr] {
|
|
pool.nonce[addr] = nonce
|
|
}
|
|
|
|
// Notify the subscribers. This event is posted in a goroutine
|
|
// because it's possible that somewhere during the post "Remove transaction"
|
|
// gets called which will then wait for the global tx pool lock and deadlock.
|
|
go pool.txFeed.Send(core.NewTxsEvent{Txs: types.Transactions{tx}})
|
|
}
|
|
|
|
// Print a log message if low enough level is set
|
|
log.Debug("Pooled new transaction", "hash", hash, "from", log.Lazy{Fn: func() common.Address { from, _ := types.Sender(pool.signer, tx); return from }}, "to", tx.To())
|
|
return nil
|
|
}
|
|
|
|
// Add adds a transaction to the pool if valid and passes it to the tx relay
|
|
// backend
|
|
func (pool *TxPool) Add(ctx context.Context, tx *types.Transaction) error {
|
|
pool.mu.Lock()
|
|
defer pool.mu.Unlock()
|
|
data, err := tx.MarshalBinary()
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if err := pool.add(ctx, tx); err != nil {
|
|
return err
|
|
}
|
|
//fmt.Println("Send", tx.Hash())
|
|
pool.relay.Send(types.Transactions{tx})
|
|
|
|
pool.chainDb.Put(tx.Hash().Bytes(), data)
|
|
return nil
|
|
}
|
|
|
|
// AddBatch adds all valid transactions to the pool and passes them to
|
|
// the tx relay backend
|
|
func (pool *TxPool) AddBatch(ctx context.Context, txs []*types.Transaction) {
|
|
pool.mu.Lock()
|
|
defer pool.mu.Unlock()
|
|
var sendTx types.Transactions
|
|
|
|
for _, tx := range txs {
|
|
if err := pool.add(ctx, tx); err == nil {
|
|
sendTx = append(sendTx, tx)
|
|
}
|
|
}
|
|
if len(sendTx) > 0 {
|
|
pool.relay.Send(sendTx)
|
|
}
|
|
}
|
|
|
|
// GetTransaction returns a transaction if it is contained in the pool
|
|
// and nil otherwise.
|
|
func (pool *TxPool) GetTransaction(hash common.Hash) *types.Transaction {
|
|
// check the txs first
|
|
if tx, ok := pool.pending[hash]; ok {
|
|
return tx
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// GetTransactions returns all currently processable transactions.
|
|
// The returned slice may be modified by the caller.
|
|
func (pool *TxPool) GetTransactions() (txs types.Transactions, err error) {
|
|
pool.mu.RLock()
|
|
defer pool.mu.RUnlock()
|
|
|
|
txs = make(types.Transactions, len(pool.pending))
|
|
i := 0
|
|
for _, tx := range pool.pending {
|
|
txs[i] = tx
|
|
i++
|
|
}
|
|
return txs, nil
|
|
}
|
|
|
|
// Content retrieves the data content of the transaction pool, returning all the
|
|
// pending as well as queued transactions, grouped by account and nonce.
|
|
func (pool *TxPool) Content() (map[common.Address]types.Transactions, map[common.Address]types.Transactions) {
|
|
pool.mu.RLock()
|
|
defer pool.mu.RUnlock()
|
|
|
|
// Retrieve all the pending transactions and sort by account and by nonce
|
|
pending := make(map[common.Address]types.Transactions)
|
|
for _, tx := range pool.pending {
|
|
account, _ := types.Sender(pool.signer, tx)
|
|
pending[account] = append(pending[account], tx)
|
|
}
|
|
// There are no queued transactions in a light pool, just return an empty map
|
|
queued := make(map[common.Address]types.Transactions)
|
|
return pending, queued
|
|
}
|
|
|
|
// ContentFrom retrieves the data content of the transaction pool, returning the
|
|
// pending as well as queued transactions of this address, grouped by nonce.
|
|
func (pool *TxPool) ContentFrom(addr common.Address) (types.Transactions, types.Transactions) {
|
|
pool.mu.RLock()
|
|
defer pool.mu.RUnlock()
|
|
|
|
// Retrieve the pending transactions and sort by nonce
|
|
var pending types.Transactions
|
|
for _, tx := range pool.pending {
|
|
account, _ := types.Sender(pool.signer, tx)
|
|
if account != addr {
|
|
continue
|
|
}
|
|
pending = append(pending, tx)
|
|
}
|
|
// There are no queued transactions in a light pool, just return an empty map
|
|
return pending, types.Transactions{}
|
|
}
|
|
|
|
// RemoveTransactions removes all given transactions from the pool.
|
|
func (pool *TxPool) RemoveTransactions(txs types.Transactions) {
|
|
pool.mu.Lock()
|
|
defer pool.mu.Unlock()
|
|
|
|
var hashes []common.Hash
|
|
batch := pool.chainDb.NewBatch()
|
|
for _, tx := range txs {
|
|
hash := tx.Hash()
|
|
delete(pool.pending, hash)
|
|
batch.Delete(hash.Bytes())
|
|
hashes = append(hashes, hash)
|
|
}
|
|
batch.Write()
|
|
pool.relay.Discard(hashes)
|
|
}
|
|
|
|
// RemoveTx removes the transaction with the given hash from the pool.
|
|
func (pool *TxPool) RemoveTx(hash common.Hash) {
|
|
pool.mu.Lock()
|
|
defer pool.mu.Unlock()
|
|
// delete from pending pool
|
|
delete(pool.pending, hash)
|
|
pool.chainDb.Delete(hash[:])
|
|
pool.relay.Discard([]common.Hash{hash})
|
|
}
|