bsc/eth/fetcher/block_fetcher.go

915 lines
33 KiB
Go
Raw Normal View History

2015-07-07 03:54:22 +03:00
// Copyright 2015 The go-ethereum Authors
// This file is part of the go-ethereum library.
2015-07-07 03:54:22 +03:00
//
// The go-ethereum library is free software: you can redistribute it and/or modify
2015-07-07 03:54:22 +03:00
// 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,
2015-07-07 03:54:22 +03:00
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2015-07-07 03:54:22 +03:00
// 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/>.
2015-07-07 03:54:22 +03:00
// Package fetcher contains the announcement based header, blocks or transaction synchronisation.
package fetcher
import (
"errors"
"math/rand"
"time"
"github.com/ethereum/go-ethereum/common"
[R4R] performance improvement in many aspects (#257) * focus on performance improvement in many aspects. 1. Do BlockBody verification concurrently; 2. Do calculation of intermediate root concurrently; 3. Preload accounts before processing blocks; 4. Make the snapshot layers configurable. 5. Reuse some object to reduce GC. add * rlp: improve decoder stream implementation (#22858) This commit makes various cleanup changes to rlp.Stream. * rlp: shrink Stream struct This removes a lot of unused padding space in Stream by reordering the fields. The size of Stream changes from 120 bytes to 88 bytes. Stream instances are internally cached and reused using sync.Pool, so this does not improve performance. * rlp: simplify list stack The list stack kept track of the size of the current list context as well as the current offset into it. The size had to be stored in the stack in order to subtract it from the remaining bytes of any enclosing list in ListEnd. It seems that this can be implemented in a simpler way: just subtract the size from the enclosing list context in List instead. * rlp: use atomic.Value for type cache (#22902) All encoding/decoding operations read the type cache to find the writer/decoder function responsible for a type. When analyzing CPU profiles of geth during sync, I found that the use of sync.RWMutex in cache lookups appears in the profiles. It seems we are running into CPU cache contention problems when package rlp is heavily used on all CPU cores during sync. This change makes it use atomic.Value + a writer lock instead of sync.RWMutex. In the common case where the typeinfo entry is present in the cache, we simply fetch the map and lookup the type. * rlp: optimize byte array handling (#22924) This change improves the performance of encoding/decoding [N]byte. name old time/op new time/op delta DecodeByteArrayStruct-8 336ns ± 0% 246ns ± 0% -26.98% (p=0.000 n=9+10) EncodeByteArrayStruct-8 225ns ± 1% 148ns ± 1% -34.12% (p=0.000 n=10+10) name old alloc/op new alloc/op delta DecodeByteArrayStruct-8 120B ± 0% 48B ± 0% -60.00% (p=0.000 n=10+10) EncodeByteArrayStruct-8 0.00B 0.00B ~ (all equal) * rlp: optimize big.Int decoding for size <= 32 bytes (#22927) This change grows the static integer buffer in Stream to 32 bytes, making it possible to decode 256bit integers without allocating a temporary buffer. In the recent commit 088da24, Stream struct size decreased from 120 bytes down to 88 bytes. This commit grows the struct to 112 bytes again, but the size change will not degrade performance because Stream instances are internally cached in sync.Pool. name old time/op new time/op delta DecodeBigInts-8 12.2µs ± 0% 8.6µs ± 4% -29.58% (p=0.000 n=9+10) name old speed new speed delta DecodeBigInts-8 230MB/s ± 0% 326MB/s ± 4% +42.04% (p=0.000 n=9+10) * eth/protocols/eth, les: avoid Raw() when decoding HashOrNumber (#22841) Getting the raw value is not necessary to decode this type, and decoding it directly from the stream is faster. * fix testcase * debug no lazy * fix can not repair * address comments Co-authored-by: Felix Lange <fjl@twurst.com>
2021-07-29 12:16:53 +03:00
"github.com/ethereum/go-ethereum/common/gopool"
"github.com/ethereum/go-ethereum/common/prque"
"github.com/ethereum/go-ethereum/consensus"
"github.com/ethereum/go-ethereum/core/types"
"github.com/ethereum/go-ethereum/log"
"github.com/ethereum/go-ethereum/metrics"
"github.com/ethereum/go-ethereum/trie"
)
const (
2021-07-29 12:17:11 +03:00
lightTimeout = time.Millisecond // Time allowance before an announced header is explicitly requested
arriveTimeout = 500 * time.Millisecond // Time allowance before an announced block/transaction is explicitly requested
gatherSlack = 100 * time.Millisecond // Interval used to collate almost-expired announces with fetches
fetchTimeout = 5 * time.Second // Maximum allotted time to return an explicitly requested block/transaction
reQueueBlockTimeout = 500 * time.Millisecond // Time allowance before blocks are requeued for import
)
const (
2020-06-15 12:11:06 +03:00
maxUncleDist = 11 // Maximum allowed backward distance from the chain head
maxQueueDist = 32 // Maximum allowed distance from the chain head to queue
hashLimit = 256 // Maximum number of unique blocks or headers a peer may have announced
blockLimit = 64 // Maximum number of unique blocks a peer may have delivered
)
var (
blockAnnounceInMeter = metrics.NewRegisteredMeter("eth/fetcher/block/announces/in", nil)
blockAnnounceOutTimer = metrics.NewRegisteredTimer("eth/fetcher/block/announces/out", nil)
blockAnnounceDropMeter = metrics.NewRegisteredMeter("eth/fetcher/block/announces/drop", nil)
blockAnnounceDOSMeter = metrics.NewRegisteredMeter("eth/fetcher/block/announces/dos", nil)
blockBroadcastInMeter = metrics.NewRegisteredMeter("eth/fetcher/block/broadcasts/in", nil)
blockBroadcastOutTimer = metrics.NewRegisteredTimer("eth/fetcher/block/broadcasts/out", nil)
blockBroadcastDropMeter = metrics.NewRegisteredMeter("eth/fetcher/block/broadcasts/drop", nil)
blockBroadcastDOSMeter = metrics.NewRegisteredMeter("eth/fetcher/block/broadcasts/dos", nil)
headerFetchMeter = metrics.NewRegisteredMeter("eth/fetcher/block/headers", nil)
bodyFetchMeter = metrics.NewRegisteredMeter("eth/fetcher/block/bodies", nil)
headerFilterInMeter = metrics.NewRegisteredMeter("eth/fetcher/block/filter/headers/in", nil)
headerFilterOutMeter = metrics.NewRegisteredMeter("eth/fetcher/block/filter/headers/out", nil)
bodyFilterInMeter = metrics.NewRegisteredMeter("eth/fetcher/block/filter/bodies/in", nil)
bodyFilterOutMeter = metrics.NewRegisteredMeter("eth/fetcher/block/filter/bodies/out", nil)
)
var errTerminated = errors.New("terminated")
// HeaderRetrievalFn is a callback type for retrieving a header from the local chain.
type HeaderRetrievalFn func(common.Hash) *types.Header
// blockRetrievalFn is a callback type for retrieving a block from the local chain.
type blockRetrievalFn func(common.Hash) *types.Block
// headerRequesterFn is a callback type for sending a header retrieval request.
type headerRequesterFn func(common.Hash) error
// bodyRequesterFn is a callback type for sending a body retrieval request.
type bodyRequesterFn func([]common.Hash) error
// headerVerifierFn is a callback type to verify a block's header for fast propagation.
type headerVerifierFn func(header *types.Header) error
// blockBroadcasterFn is a callback type for broadcasting a block to connected peers.
type blockBroadcasterFn func(block *types.Block, propagate bool)
// chainHeightFn is a callback type to retrieve the current chain height.
type chainHeightFn func() uint64
// headersInsertFn is a callback type to insert a batch of headers into the local chain.
type headersInsertFn func(headers []*types.Header) (int, error)
// chainInsertFn is a callback type to insert a batch of blocks into the local chain.
type chainInsertFn func(types.Blocks) (int, error)
// peerDropFn is a callback type for dropping a peer detected as malicious.
type peerDropFn func(id string)
// blockAnnounce is the hash notification of the availability of a new block in the
// network.
type blockAnnounce struct {
hash common.Hash // Hash of the block being announced
number uint64 // Number of the block being announced (0 = unknown | old protocol)
header *types.Header // Header of the block partially reassembled (new protocol)
time time.Time // Timestamp of the announcement
origin string // Identifier of the peer originating the notification
fetchHeader headerRequesterFn // Fetcher function to retrieve the header of an announced block
fetchBodies bodyRequesterFn // Fetcher function to retrieve the body of an announced block
}
// headerFilterTask represents a batch of headers needing fetcher filtering.
type headerFilterTask struct {
peer string // The source peer of block headers
headers []*types.Header // Collection of headers to filter
time time.Time // Arrival time of the headers
}
2018-06-13 17:02:28 +03:00
// bodyFilterTask represents a batch of block bodies (transactions and uncles)
// needing fetcher filtering.
type bodyFilterTask struct {
peer string // The source peer of block bodies
transactions [][]*types.Transaction // Collection of transactions per block bodies
uncles [][]*types.Header // Collection of uncles per block bodies
time time.Time // Arrival time of the blocks' contents
}
// blockOrHeaderInject represents a schedules import operation.
type blockOrHeaderInject struct {
origin string
header *types.Header // Used for light mode fetcher which only cares about header.
block *types.Block // Used for normal mode fetcher which imports full block.
}
// number returns the block number of the injected object.
func (inject *blockOrHeaderInject) number() uint64 {
if inject.header != nil {
return inject.header.Number.Uint64()
}
return inject.block.NumberU64()
}
// number returns the block hash of the injected object.
func (inject *blockOrHeaderInject) hash() common.Hash {
if inject.header != nil {
return inject.header.Hash()
}
return inject.block.Hash()
}
// BlockFetcher is responsible for accumulating block announcements from various peers
// and scheduling them for retrieval.
type BlockFetcher struct {
light bool // The indicator whether it's a light fetcher or normal one.
// Various event channels
notify chan *blockAnnounce
inject chan *blockOrHeaderInject
headerFilter chan chan *headerFilterTask
bodyFilter chan chan *bodyFilterTask
done chan common.Hash
quit chan struct{}
2021-07-29 12:17:11 +03:00
requeue chan *blockOrHeaderInject
// Announce states
announces map[string]int // Per peer blockAnnounce counts to prevent memory exhaustion
announced map[common.Hash][]*blockAnnounce // Announced blocks, scheduled for fetching
fetching map[common.Hash]*blockAnnounce // Announced blocks, currently fetching
fetched map[common.Hash][]*blockAnnounce // Blocks with headers fetched, scheduled for body retrieval
completing map[common.Hash]*blockAnnounce // Blocks with headers, currently body-completing
// Block cache
queue *prque.Prque // Queue containing the import operations (block number sorted)
queues map[string]int // Per peer block counts to prevent memory exhaustion
queued map[common.Hash]*blockOrHeaderInject // Set of already queued blocks (to dedup imports)
// Callbacks
getHeader HeaderRetrievalFn // Retrieves a header from the local chain
getBlock blockRetrievalFn // Retrieves a block from the local chain
verifyHeader headerVerifierFn // Checks if a block's headers have a valid proof of work
broadcastBlock blockBroadcasterFn // Broadcasts a block to connected peers
chainHeight chainHeightFn // Retrieves the current chain's height
insertHeaders headersInsertFn // Injects a batch of headers into the chain
insertChain chainInsertFn // Injects a batch of blocks into the chain
dropPeer peerDropFn // Drops a peer for misbehaving
// Testing hooks
announceChangeHook func(common.Hash, bool) // Method to call upon adding or deleting a hash from the blockAnnounce list
queueChangeHook func(common.Hash, bool) // Method to call upon adding or deleting a block from the import queue
fetchingHook func([]common.Hash) // Method to call upon starting a block (eth/61) or header (eth/62) fetch
completingHook func([]common.Hash) // Method to call upon starting a block body fetch (eth/62)
importedHook func(*types.Header, *types.Block) // Method to call upon successful header or block import (both eth/61 and eth/62)
}
// NewBlockFetcher creates a block fetcher to retrieve blocks based on hash announcements.
func NewBlockFetcher(light bool, getHeader HeaderRetrievalFn, getBlock blockRetrievalFn, verifyHeader headerVerifierFn, broadcastBlock blockBroadcasterFn, chainHeight chainHeightFn, insertHeaders headersInsertFn, insertChain chainInsertFn, dropPeer peerDropFn) *BlockFetcher {
return &BlockFetcher{
light: light,
notify: make(chan *blockAnnounce),
inject: make(chan *blockOrHeaderInject),
headerFilter: make(chan chan *headerFilterTask),
bodyFilter: make(chan chan *bodyFilterTask),
done: make(chan common.Hash),
quit: make(chan struct{}),
2021-07-29 12:17:11 +03:00
requeue: make(chan *blockOrHeaderInject),
announces: make(map[string]int),
announced: make(map[common.Hash][]*blockAnnounce),
fetching: make(map[common.Hash]*blockAnnounce),
fetched: make(map[common.Hash][]*blockAnnounce),
completing: make(map[common.Hash]*blockAnnounce),
queue: prque.New(nil),
queues: make(map[string]int),
queued: make(map[common.Hash]*blockOrHeaderInject),
getHeader: getHeader,
getBlock: getBlock,
verifyHeader: verifyHeader,
broadcastBlock: broadcastBlock,
chainHeight: chainHeight,
insertHeaders: insertHeaders,
insertChain: insertChain,
dropPeer: dropPeer,
}
}
2016-03-15 20:27:49 +02:00
// Start boots up the announcement based synchroniser, accepting and processing
// hash notifications and block fetches until termination requested.
func (f *BlockFetcher) Start() {
go f.loop()
}
// Stop terminates the announcement based synchroniser, canceling all pending
// operations.
func (f *BlockFetcher) Stop() {
close(f.quit)
}
// Notify announces the fetcher of the potential availability of a new block in
// the network.
func (f *BlockFetcher) Notify(peer string, hash common.Hash, number uint64, time time.Time,
headerFetcher headerRequesterFn, bodyFetcher bodyRequesterFn) error {
block := &blockAnnounce{
hash: hash,
number: number,
time: time,
origin: peer,
fetchHeader: headerFetcher,
fetchBodies: bodyFetcher,
}
select {
case f.notify <- block:
return nil
case <-f.quit:
return errTerminated
}
}
// Enqueue tries to fill gaps the fetcher's future import queue.
func (f *BlockFetcher) Enqueue(peer string, block *types.Block) error {
op := &blockOrHeaderInject{
origin: peer,
block: block,
}
select {
case f.inject <- op:
return nil
case <-f.quit:
return errTerminated
}
}
// FilterHeaders extracts all the headers that were explicitly requested by the fetcher,
// returning those that should be handled differently.
func (f *BlockFetcher) FilterHeaders(peer string, headers []*types.Header, time time.Time) []*types.Header {
log.Trace("Filtering headers", "peer", peer, "headers", len(headers))
// Send the filter channel to the fetcher
filter := make(chan *headerFilterTask)
select {
case f.headerFilter <- filter:
case <-f.quit:
return nil
}
// Request the filtering of the header list
select {
case filter <- &headerFilterTask{peer: peer, headers: headers, time: time}:
case <-f.quit:
return nil
}
// Retrieve the headers remaining after filtering
select {
case task := <-filter:
return task.headers
case <-f.quit:
return nil
}
}
// FilterBodies extracts all the block bodies that were explicitly requested by
// the fetcher, returning those that should be handled differently.
func (f *BlockFetcher) FilterBodies(peer string, transactions [][]*types.Transaction, uncles [][]*types.Header, time time.Time) ([][]*types.Transaction, [][]*types.Header) {
log.Trace("Filtering bodies", "peer", peer, "txs", len(transactions), "uncles", len(uncles))
// Send the filter channel to the fetcher
filter := make(chan *bodyFilterTask)
select {
case f.bodyFilter <- filter:
case <-f.quit:
return nil, nil
}
// Request the filtering of the body list
select {
case filter <- &bodyFilterTask{peer: peer, transactions: transactions, uncles: uncles, time: time}:
case <-f.quit:
return nil, nil
}
// Retrieve the bodies remaining after filtering
select {
case task := <-filter:
return task.transactions, task.uncles
case <-f.quit:
return nil, nil
}
}
// Loop is the main fetcher loop, checking and processing various notification
// events.
func (f *BlockFetcher) loop() {
2015-06-16 18:43:58 +03:00
// Iterate the block fetching until a quit is requested
var (
fetchTimer = time.NewTimer(0)
completeTimer = time.NewTimer(0)
)
<-fetchTimer.C // clear out the channel
<-completeTimer.C
defer fetchTimer.Stop()
defer completeTimer.Stop()
for {
// Clean up any expired block fetches
for hash, announce := range f.fetching {
if time.Since(announce.time) > fetchTimeout {
f.forgetHash(hash)
}
}
// Import any queued blocks that could potentially fit
height := f.chainHeight()
for !f.queue.Empty() {
op := f.queue.PopItem().(*blockOrHeaderInject)
hash := op.hash()
if f.queueChangeHook != nil {
f.queueChangeHook(hash, false)
}
// If too high up the chain or phase, continue later
number := op.number()
if number > height+1 {
f.queue.Push(op, -int64(number))
if f.queueChangeHook != nil {
f.queueChangeHook(hash, true)
}
break
}
// Otherwise if fresh and still unknown, try and import
if (number+maxUncleDist < height) || (f.light && f.getHeader(hash) != nil) || (!f.light && f.getBlock(hash) != nil) {
f.forgetBlock(hash)
2015-06-16 18:43:58 +03:00
continue
}
if f.light {
2021-07-29 12:17:11 +03:00
f.importHeaders(op)
} else {
2021-07-29 12:17:11 +03:00
f.importBlocks(op)
}
}
// Wait for an outside event to occur
select {
case <-f.quit:
// BlockFetcher terminating, abort all operations
return
case notification := <-f.notify:
// A block was announced, make sure the peer isn't DOSing us
blockAnnounceInMeter.Mark(1)
count := f.announces[notification.origin] + 1
if count > hashLimit {
log.Debug("Peer exceeded outstanding announces", "peer", notification.origin, "limit", hashLimit)
blockAnnounceDOSMeter.Mark(1)
break
}
// If we have a valid block number, check that it's potentially useful
if notification.number > 0 {
if dist := int64(notification.number) - int64(f.chainHeight()); dist < -maxUncleDist || dist > maxQueueDist {
log.Debug("Peer discarded announcement", "peer", notification.origin, "number", notification.number, "hash", notification.hash, "distance", dist)
blockAnnounceDropMeter.Mark(1)
break
}
}
// All is well, schedule the announce if block's not yet downloading
if _, ok := f.fetching[notification.hash]; ok {
break
}
if _, ok := f.completing[notification.hash]; ok {
break
}
f.announces[notification.origin] = count
f.announced[notification.hash] = append(f.announced[notification.hash], notification)
if f.announceChangeHook != nil && len(f.announced[notification.hash]) == 1 {
f.announceChangeHook(notification.hash, true)
}
if len(f.announced) == 1 {
f.rescheduleFetch(fetchTimer)
}
2021-07-29 12:17:11 +03:00
case op := <-f.requeue:
// Re-queue blocks that have not been written due to fork block competition
number := int64(0)
hash := ""
if op.header != nil {
number = op.header.Number.Int64()
hash = op.header.Hash().String()
} else if op.block != nil {
number = op.block.Number().Int64()
hash = op.block.Hash().String()
}
log.Info("Re-queue blocks", "number", number, "hash", hash)
f.enqueue(op.origin, op.header, op.block)
case op := <-f.inject:
// A direct block insertion was requested, try and fill any pending gaps
blockBroadcastInMeter.Mark(1)
// Now only direct block injection is allowed, drop the header injection
// here silently if we receive.
if f.light {
continue
}
f.enqueue(op.origin, nil, op.block)
case hash := <-f.done:
// A pending import finished, remove all traces of the notification
f.forgetHash(hash)
f.forgetBlock(hash)
case <-fetchTimer.C:
// At least one block's timer ran out, check for needing retrieval
request := make(map[string][]common.Hash)
for hash, announces := range f.announced {
// In current LES protocol(les2/les3), only header announce is
// available, no need to wait too much time for header broadcast.
timeout := arriveTimeout - gatherSlack
if f.light {
timeout = 0
}
if time.Since(announces[0].time) > timeout {
// Pick a random peer to retrieve from, reset all others
announce := announces[rand.Intn(len(announces))]
f.forgetHash(hash)
// If the block still didn't arrive, queue for fetching
if (f.light && f.getHeader(hash) == nil) || (!f.light && f.getBlock(hash) == nil) {
request[announce.origin] = append(request[announce.origin], hash)
f.fetching[hash] = announce
}
}
}
// Send out all block header requests
for peer, hashes := range request {
log.Trace("Fetching scheduled headers", "peer", peer, "list", hashes)
2015-06-22 20:13:18 +03:00
// Create a closure of the fetch and schedule in on a new thread
fetchHeader, hashes := f.fetching[hashes[0]].fetchHeader, hashes
[R4R] performance improvement in many aspects (#257) * focus on performance improvement in many aspects. 1. Do BlockBody verification concurrently; 2. Do calculation of intermediate root concurrently; 3. Preload accounts before processing blocks; 4. Make the snapshot layers configurable. 5. Reuse some object to reduce GC. add * rlp: improve decoder stream implementation (#22858) This commit makes various cleanup changes to rlp.Stream. * rlp: shrink Stream struct This removes a lot of unused padding space in Stream by reordering the fields. The size of Stream changes from 120 bytes to 88 bytes. Stream instances are internally cached and reused using sync.Pool, so this does not improve performance. * rlp: simplify list stack The list stack kept track of the size of the current list context as well as the current offset into it. The size had to be stored in the stack in order to subtract it from the remaining bytes of any enclosing list in ListEnd. It seems that this can be implemented in a simpler way: just subtract the size from the enclosing list context in List instead. * rlp: use atomic.Value for type cache (#22902) All encoding/decoding operations read the type cache to find the writer/decoder function responsible for a type. When analyzing CPU profiles of geth during sync, I found that the use of sync.RWMutex in cache lookups appears in the profiles. It seems we are running into CPU cache contention problems when package rlp is heavily used on all CPU cores during sync. This change makes it use atomic.Value + a writer lock instead of sync.RWMutex. In the common case where the typeinfo entry is present in the cache, we simply fetch the map and lookup the type. * rlp: optimize byte array handling (#22924) This change improves the performance of encoding/decoding [N]byte. name old time/op new time/op delta DecodeByteArrayStruct-8 336ns ± 0% 246ns ± 0% -26.98% (p=0.000 n=9+10) EncodeByteArrayStruct-8 225ns ± 1% 148ns ± 1% -34.12% (p=0.000 n=10+10) name old alloc/op new alloc/op delta DecodeByteArrayStruct-8 120B ± 0% 48B ± 0% -60.00% (p=0.000 n=10+10) EncodeByteArrayStruct-8 0.00B 0.00B ~ (all equal) * rlp: optimize big.Int decoding for size <= 32 bytes (#22927) This change grows the static integer buffer in Stream to 32 bytes, making it possible to decode 256bit integers without allocating a temporary buffer. In the recent commit 088da24, Stream struct size decreased from 120 bytes down to 88 bytes. This commit grows the struct to 112 bytes again, but the size change will not degrade performance because Stream instances are internally cached in sync.Pool. name old time/op new time/op delta DecodeBigInts-8 12.2µs ± 0% 8.6µs ± 4% -29.58% (p=0.000 n=9+10) name old speed new speed delta DecodeBigInts-8 230MB/s ± 0% 326MB/s ± 4% +42.04% (p=0.000 n=9+10) * eth/protocols/eth, les: avoid Raw() when decoding HashOrNumber (#22841) Getting the raw value is not necessary to decode this type, and decoding it directly from the stream is faster. * fix testcase * debug no lazy * fix can not repair * address comments Co-authored-by: Felix Lange <fjl@twurst.com>
2021-07-29 12:16:53 +03:00
gopool.Submit(func() {
if f.fetchingHook != nil {
f.fetchingHook(hashes)
}
for _, hash := range hashes {
headerFetchMeter.Mark(1)
fetchHeader(hash) // Suboptimal, but protocol doesn't allow batch header retrievals
}
[R4R] performance improvement in many aspects (#257) * focus on performance improvement in many aspects. 1. Do BlockBody verification concurrently; 2. Do calculation of intermediate root concurrently; 3. Preload accounts before processing blocks; 4. Make the snapshot layers configurable. 5. Reuse some object to reduce GC. add * rlp: improve decoder stream implementation (#22858) This commit makes various cleanup changes to rlp.Stream. * rlp: shrink Stream struct This removes a lot of unused padding space in Stream by reordering the fields. The size of Stream changes from 120 bytes to 88 bytes. Stream instances are internally cached and reused using sync.Pool, so this does not improve performance. * rlp: simplify list stack The list stack kept track of the size of the current list context as well as the current offset into it. The size had to be stored in the stack in order to subtract it from the remaining bytes of any enclosing list in ListEnd. It seems that this can be implemented in a simpler way: just subtract the size from the enclosing list context in List instead. * rlp: use atomic.Value for type cache (#22902) All encoding/decoding operations read the type cache to find the writer/decoder function responsible for a type. When analyzing CPU profiles of geth during sync, I found that the use of sync.RWMutex in cache lookups appears in the profiles. It seems we are running into CPU cache contention problems when package rlp is heavily used on all CPU cores during sync. This change makes it use atomic.Value + a writer lock instead of sync.RWMutex. In the common case where the typeinfo entry is present in the cache, we simply fetch the map and lookup the type. * rlp: optimize byte array handling (#22924) This change improves the performance of encoding/decoding [N]byte. name old time/op new time/op delta DecodeByteArrayStruct-8 336ns ± 0% 246ns ± 0% -26.98% (p=0.000 n=9+10) EncodeByteArrayStruct-8 225ns ± 1% 148ns ± 1% -34.12% (p=0.000 n=10+10) name old alloc/op new alloc/op delta DecodeByteArrayStruct-8 120B ± 0% 48B ± 0% -60.00% (p=0.000 n=10+10) EncodeByteArrayStruct-8 0.00B 0.00B ~ (all equal) * rlp: optimize big.Int decoding for size <= 32 bytes (#22927) This change grows the static integer buffer in Stream to 32 bytes, making it possible to decode 256bit integers without allocating a temporary buffer. In the recent commit 088da24, Stream struct size decreased from 120 bytes down to 88 bytes. This commit grows the struct to 112 bytes again, but the size change will not degrade performance because Stream instances are internally cached in sync.Pool. name old time/op new time/op delta DecodeBigInts-8 12.2µs ± 0% 8.6µs ± 4% -29.58% (p=0.000 n=9+10) name old speed new speed delta DecodeBigInts-8 230MB/s ± 0% 326MB/s ± 4% +42.04% (p=0.000 n=9+10) * eth/protocols/eth, les: avoid Raw() when decoding HashOrNumber (#22841) Getting the raw value is not necessary to decode this type, and decoding it directly from the stream is faster. * fix testcase * debug no lazy * fix can not repair * address comments Co-authored-by: Felix Lange <fjl@twurst.com>
2021-07-29 12:16:53 +03:00
})
}
// Schedule the next fetch if blocks are still pending
f.rescheduleFetch(fetchTimer)
case <-completeTimer.C:
// At least one header's timer ran out, retrieve everything
request := make(map[string][]common.Hash)
for hash, announces := range f.fetched {
// Pick a random peer to retrieve from, reset all others
announce := announces[rand.Intn(len(announces))]
f.forgetHash(hash)
// If the block still didn't arrive, queue for completion
if f.getBlock(hash) == nil {
request[announce.origin] = append(request[announce.origin], hash)
f.completing[hash] = announce
}
}
// Send out all block body requests
for peer, hashes := range request {
log.Trace("Fetching scheduled bodies", "peer", peer, "list", hashes)
// Create a closure of the fetch and schedule in on a new thread
if f.completingHook != nil {
f.completingHook(hashes)
}
bodyFetchMeter.Mark(int64(len(hashes)))
go f.completing[hashes[0]].fetchBodies(hashes)
}
// Schedule the next fetch if blocks are still pending
f.rescheduleComplete(completeTimer)
case filter := <-f.headerFilter:
// Headers arrived from a remote peer. Extract those that were explicitly
// requested by the fetcher, and return everything else so it's delivered
// to other parts of the system.
var task *headerFilterTask
select {
case task = <-filter:
case <-f.quit:
return
}
headerFilterInMeter.Mark(int64(len(task.headers)))
// Split the batch of headers into unknown ones (to return to the caller),
// known incomplete ones (requiring body retrievals) and completed blocks.
unknown, incomplete, complete, lightHeaders := []*types.Header{}, []*blockAnnounce{}, []*types.Block{}, []*blockAnnounce{}
for _, header := range task.headers {
hash := header.Hash()
// Filter fetcher-requested headers from other synchronisation algorithms
if announce := f.fetching[hash]; announce != nil && announce.origin == task.peer && f.fetched[hash] == nil && f.completing[hash] == nil && f.queued[hash] == nil {
// If the delivered header does not match the promised number, drop the announcer
if header.Number.Uint64() != announce.number {
log.Trace("Invalid block number fetched", "peer", announce.origin, "hash", header.Hash(), "announced", announce.number, "provided", header.Number)
f.dropPeer(announce.origin)
f.forgetHash(hash)
continue
}
// Collect all headers only if we are running in light
// mode and the headers are not imported by other means.
if f.light {
if f.getHeader(hash) == nil {
announce.header = header
lightHeaders = append(lightHeaders, announce)
}
f.forgetHash(hash)
continue
}
// Only keep if not imported by other means
if f.getBlock(hash) == nil {
announce.header = header
announce.time = task.time
// If the block is empty (header only), short circuit into the final import queue
if header.TxHash == types.EmptyRootHash && header.UncleHash == types.EmptyUncleHash {
log.Trace("Block empty, skipping body retrieval", "peer", announce.origin, "number", header.Number, "hash", header.Hash())
block := types.NewBlockWithHeader(header)
block.ReceivedAt = task.time
complete = append(complete, block)
f.completing[hash] = announce
continue
}
// Otherwise add to the list of blocks needing completion
incomplete = append(incomplete, announce)
} else {
log.Trace("Block already imported, discarding header", "peer", announce.origin, "number", header.Number, "hash", header.Hash())
f.forgetHash(hash)
}
} else {
// BlockFetcher doesn't know about it, add to the return list
unknown = append(unknown, header)
}
}
headerFilterOutMeter.Mark(int64(len(unknown)))
select {
case filter <- &headerFilterTask{headers: unknown, time: task.time}:
case <-f.quit:
return
}
// Schedule the retrieved headers for body completion
for _, announce := range incomplete {
hash := announce.header.Hash()
if _, ok := f.completing[hash]; ok {
continue
}
f.fetched[hash] = append(f.fetched[hash], announce)
if len(f.fetched) == 1 {
f.rescheduleComplete(completeTimer)
}
}
// Schedule the header for light fetcher import
for _, announce := range lightHeaders {
f.enqueue(announce.origin, announce.header, nil)
}
// Schedule the header-only blocks for import
for _, block := range complete {
if announce := f.completing[block.Hash()]; announce != nil {
f.enqueue(announce.origin, nil, block)
}
}
case filter := <-f.bodyFilter:
// Block bodies arrived, extract any explicitly requested blocks, return the rest
var task *bodyFilterTask
select {
case task = <-filter:
case <-f.quit:
return
}
bodyFilterInMeter.Mark(int64(len(task.transactions)))
blocks := []*types.Block{}
// abort early if there's nothing explicitly requested
if len(f.completing) > 0 {
for i := 0; i < len(task.transactions) && i < len(task.uncles); i++ {
// Match up a body to any possible completion request
var (
matched = false
uncleHash common.Hash // calculated lazily and reused
txnHash common.Hash // calculated lazily and reused
)
for hash, announce := range f.completing {
if f.queued[hash] != nil || announce.origin != task.peer {
continue
}
if uncleHash == (common.Hash{}) {
uncleHash = types.CalcUncleHash(task.uncles[i])
}
if uncleHash != announce.header.UncleHash {
continue
}
if txnHash == (common.Hash{}) {
txnHash = types.DeriveSha(types.Transactions(task.transactions[i]), trie.NewStackTrie(nil))
}
if txnHash != announce.header.TxHash {
continue
}
// Mark the body matched, reassemble if still unknown
matched = true
if f.getBlock(hash) == nil {
block := types.NewBlockWithHeader(announce.header).WithBody(task.transactions[i], task.uncles[i])
block.ReceivedAt = task.time
blocks = append(blocks, block)
} else {
f.forgetHash(hash)
}
}
if matched {
task.transactions = append(task.transactions[:i], task.transactions[i+1:]...)
task.uncles = append(task.uncles[:i], task.uncles[i+1:]...)
i--
continue
}
}
}
bodyFilterOutMeter.Mark(int64(len(task.transactions)))
select {
case filter <- task:
case <-f.quit:
return
}
// Schedule the retrieved blocks for ordered import
for _, block := range blocks {
if announce := f.completing[block.Hash()]; announce != nil {
f.enqueue(announce.origin, nil, block)
}
}
}
}
}
// rescheduleFetch resets the specified fetch timer to the next blockAnnounce timeout.
func (f *BlockFetcher) rescheduleFetch(fetch *time.Timer) {
// Short circuit if no blocks are announced
if len(f.announced) == 0 {
return
}
// Schedule announcement retrieval quickly for light mode
// since server won't send any headers to client.
if f.light {
fetch.Reset(lightTimeout)
return
}
// Otherwise find the earliest expiring announcement
earliest := time.Now()
for _, announces := range f.announced {
if earliest.After(announces[0].time) {
earliest = announces[0].time
}
}
fetch.Reset(arriveTimeout - time.Since(earliest))
}
// rescheduleComplete resets the specified completion timer to the next fetch timeout.
func (f *BlockFetcher) rescheduleComplete(complete *time.Timer) {
// Short circuit if no headers are fetched
if len(f.fetched) == 0 {
return
}
// Otherwise find the earliest expiring announcement
earliest := time.Now()
for _, announces := range f.fetched {
if earliest.After(announces[0].time) {
earliest = announces[0].time
}
}
complete.Reset(gatherSlack - time.Since(earliest))
}
// enqueue schedules a new header or block import operation, if the component
// to be imported has not yet been seen.
func (f *BlockFetcher) enqueue(peer string, header *types.Header, block *types.Block) {
var (
hash common.Hash
number uint64
)
if header != nil {
hash, number = header.Hash(), header.Number.Uint64()
} else {
hash, number = block.Hash(), block.NumberU64()
}
// Ensure the peer isn't DOSing us
count := f.queues[peer] + 1
if count > blockLimit {
log.Debug("Discarded delivered header or block, exceeded allowance", "peer", peer, "number", number, "hash", hash, "limit", blockLimit)
blockBroadcastDOSMeter.Mark(1)
f.forgetHash(hash)
return
}
// Discard any past or too distant blocks
if dist := int64(number) - int64(f.chainHeight()); dist < -maxUncleDist || dist > maxQueueDist {
log.Debug("Discarded delivered header or block, too far away", "peer", peer, "number", number, "hash", hash, "distance", dist)
blockBroadcastDropMeter.Mark(1)
f.forgetHash(hash)
return
}
// Schedule the block for future importing
if _, ok := f.queued[hash]; !ok {
op := &blockOrHeaderInject{origin: peer}
if header != nil {
op.header = header
} else {
op.block = block
}
f.queues[peer] = count
f.queued[hash] = op
f.queue.Push(op, -int64(number))
if f.queueChangeHook != nil {
f.queueChangeHook(hash, true)
}
log.Debug("Queued delivered header or block", "peer", peer, "number", number, "hash", hash, "queued", f.queue.Size())
}
}
// importHeaders spawns a new goroutine to run a header insertion into the chain.
// If the header's number is at the same height as the current import phase, it
// updates the phase states accordingly.
2021-07-29 12:17:11 +03:00
func (f *BlockFetcher) importHeaders(op *blockOrHeaderInject) {
peer := op.origin
header := op.header
hash := header.Hash()
log.Debug("Importing propagated header", "peer", peer, "number", header.Number, "hash", hash)
go func() {
defer func() { f.done <- hash }()
// If the parent's unknown, abort insertion
parent := f.getHeader(header.ParentHash)
if parent == nil {
log.Debug("Unknown parent of propagated header", "peer", peer, "number", header.Number, "hash", hash, "parent", header.ParentHash)
2021-07-29 12:17:11 +03:00
time.Sleep(reQueueBlockTimeout)
f.requeue <- op
return
}
// Validate the header and if something went wrong, drop the peer
if err := f.verifyHeader(header); err != nil && err != consensus.ErrFutureBlock {
log.Debug("Propagated header verification failed", "peer", peer, "number", header.Number, "hash", hash, "err", err)
f.dropPeer(peer)
return
}
// Run the actual import and log any issues
if _, err := f.insertHeaders([]*types.Header{header}); err != nil {
log.Debug("Propagated header import failed", "peer", peer, "number", header.Number, "hash", hash, "err", err)
return
}
// Invoke the testing hook if needed
if f.importedHook != nil {
f.importedHook(header, nil)
}
}()
}
// importBlocks spawns a new goroutine to run a block insertion into the chain. If the
2018-03-07 12:15:54 +03:00
// block's number is at the same height as the current import phase, it updates
// the phase states accordingly.
2021-07-29 12:17:11 +03:00
func (f *BlockFetcher) importBlocks(op *blockOrHeaderInject) {
peer := op.origin
block := op.block
hash := block.Hash()
// Run the import on a new thread
log.Debug("Importing propagated block", "peer", peer, "number", block.Number(), "hash", hash)
go func() {
defer func() { f.done <- hash }()
// If the parent's unknown, abort insertion
parent := f.getBlock(block.ParentHash())
if parent == nil {
log.Debug("Unknown parent of propagated block", "peer", peer, "number", block.Number(), "hash", hash, "parent", block.ParentHash())
2021-07-29 12:17:11 +03:00
time.Sleep(reQueueBlockTimeout)
f.requeue <- op
return
}
// Quickly validate the header and propagate the block if it passes
switch err := f.verifyHeader(block.Header()); err {
case nil:
// All ok, quickly propagate to our peers
blockBroadcastOutTimer.UpdateSince(block.ReceivedAt)
go f.broadcastBlock(block, true)
case consensus.ErrFutureBlock:
log.Error("Received future block", "peer", peer, "number", block.Number(), "hash", hash, "err", err)
f.dropPeer(peer)
default:
// Something went very wrong, drop the peer
log.Error("Propagated block verification failed", "peer", peer, "number", block.Number(), "hash", hash, "err", err)
f.dropPeer(peer)
return
}
// Run the actual import and log any issues
if _, err := f.insertChain(types.Blocks{block}); err != nil {
log.Debug("Propagated block import failed", "peer", peer, "number", block.Number(), "hash", hash, "err", err)
return
}
// If import succeeded, broadcast the block
blockAnnounceOutTimer.UpdateSince(block.ReceivedAt)
go f.broadcastBlock(block, false)
// Invoke the testing hook if needed
if f.importedHook != nil {
f.importedHook(nil, block)
}
}()
}
// forgetHash removes all traces of a block announcement from the fetcher's
// internal state.
func (f *BlockFetcher) forgetHash(hash common.Hash) {
// Remove all pending announces and decrement DOS counters
for _, announce := range f.announced[hash] {
f.announces[announce.origin]--
if f.announces[announce.origin] <= 0 {
delete(f.announces, announce.origin)
}
}
delete(f.announced, hash)
if f.announceChangeHook != nil {
f.announceChangeHook(hash, false)
}
// Remove any pending fetches and decrement the DOS counters
if announce := f.fetching[hash]; announce != nil {
f.announces[announce.origin]--
if f.announces[announce.origin] <= 0 {
delete(f.announces, announce.origin)
}
delete(f.fetching, hash)
}
// Remove any pending completion requests and decrement the DOS counters
for _, announce := range f.fetched[hash] {
f.announces[announce.origin]--
if f.announces[announce.origin] <= 0 {
delete(f.announces, announce.origin)
}
}
delete(f.fetched, hash)
// Remove any pending completions and decrement the DOS counters
if announce := f.completing[hash]; announce != nil {
f.announces[announce.origin]--
if f.announces[announce.origin] <= 0 {
delete(f.announces, announce.origin)
}
delete(f.completing, hash)
}
}
// forgetBlock removes all traces of a queued block from the fetcher's internal
// state.
func (f *BlockFetcher) forgetBlock(hash common.Hash) {
if insert := f.queued[hash]; insert != nil {
f.queues[insert.origin]--
if f.queues[insert.origin] == 0 {
delete(f.queues, insert.origin)
}
delete(f.queued, hash)
}
}