go-ethereum/eth/protocols/snap/handler.go

572 lines
19 KiB
Go
Raw Normal View History

// Copyright 2020 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 snap
import (
"bytes"
"fmt"
"time"
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/core"
"github.com/ethereum/go-ethereum/core/types"
"github.com/ethereum/go-ethereum/log"
"github.com/ethereum/go-ethereum/metrics"
"github.com/ethereum/go-ethereum/p2p"
"github.com/ethereum/go-ethereum/p2p/enode"
"github.com/ethereum/go-ethereum/p2p/enr"
"github.com/ethereum/go-ethereum/trie"
"github.com/ethereum/go-ethereum/trie/trienode"
)
const (
// softResponseLimit is the target maximum size of replies to data retrievals.
softResponseLimit = 2 * 1024 * 1024
// maxCodeLookups is the maximum number of bytecodes to serve. This number is
// there to limit the number of disk lookups.
maxCodeLookups = 1024
// stateLookupSlack defines the ratio by how much a state response can exceed
// the requested limit in order to try and avoid breaking up contracts into
// multiple packages and proving them.
stateLookupSlack = 0.1
// maxTrieNodeLookups is the maximum number of state trie nodes to serve. This
// number is there to limit the number of disk lookups.
maxTrieNodeLookups = 1024
// maxTrieNodeTimeSpent is the maximum time we should spend on looking up trie nodes.
// If we spend too much time, then it's a fairly high chance of timing out
// at the remote side, which means all the work is in vain.
maxTrieNodeTimeSpent = 5 * time.Second
)
// Handler is a callback to invoke from an outside runner after the boilerplate
// exchanges have passed.
type Handler func(peer *Peer) error
// Backend defines the data retrieval methods to serve remote requests and the
// callback methods to invoke on remote deliveries.
type Backend interface {
// Chain retrieves the blockchain object to serve data.
Chain() *core.BlockChain
// RunPeer is invoked when a peer joins on the `eth` protocol. The handler
// should do any peer maintenance work, handshakes and validations. If all
// is passed, control should be given back to the `handler` to process the
// inbound messages going forward.
RunPeer(peer *Peer, handler Handler) error
// PeerInfo retrieves all known `snap` information about a peer.
PeerInfo(id enode.ID) interface{}
// Handle is a callback to be invoked when a data packet is received from
// the remote peer. Only packets not consumed by the protocol handler will
// be forwarded to the backend.
Handle(peer *Peer, packet Packet) error
}
// MakeProtocols constructs the P2P protocol definitions for `snap`.
func MakeProtocols(backend Backend, dnsdisc enode.Iterator) []p2p.Protocol {
// Filter the discovery iterator for nodes advertising snap support.
dnsdisc = enode.Filter(dnsdisc, func(n *enode.Node) bool {
var snap enrEntry
return n.Load(&snap) == nil
})
protocols := make([]p2p.Protocol, len(ProtocolVersions))
for i, version := range ProtocolVersions {
version := version // Closure
protocols[i] = p2p.Protocol{
Name: ProtocolName,
Version: version,
Length: protocolLengths[version],
Run: func(p *p2p.Peer, rw p2p.MsgReadWriter) error {
return backend.RunPeer(NewPeer(version, p, rw), func(peer *Peer) error {
return Handle(backend, peer)
})
},
NodeInfo: func() interface{} {
return nodeInfo(backend.Chain())
},
PeerInfo: func(id enode.ID) interface{} {
return backend.PeerInfo(id)
},
Attributes: []enr.Entry{&enrEntry{}},
DialCandidates: dnsdisc,
}
}
return protocols
}
// Handle is the callback invoked to manage the life cycle of a `snap` peer.
// When this function terminates, the peer is disconnected.
func Handle(backend Backend, peer *Peer) error {
for {
if err := HandleMessage(backend, peer); err != nil {
peer.Log().Debug("Message handling failed in `snap`", "err", err)
return err
}
}
}
// HandleMessage is invoked whenever an inbound message is received from a
2021-03-19 02:49:24 -05:00
// remote peer on the `snap` protocol. The remote connection is torn down upon
// returning any error.
func HandleMessage(backend Backend, peer *Peer) error {
// Read the next message from the remote peer, and ensure it's fully consumed
msg, err := peer.rw.ReadMsg()
if err != nil {
return err
}
if msg.Size > maxMessageSize {
return fmt.Errorf("%w: %v > %v", errMsgTooLarge, msg.Size, maxMessageSize)
}
defer msg.Discard()
start := time.Now()
2022-10-11 02:37:00 -05:00
// Track the amount of time it takes to serve the request and run the handler
if metrics.Enabled {
h := fmt.Sprintf("%s/%s/%d/%#02x", p2p.HandleHistName, ProtocolName, peer.Version(), msg.Code)
defer func(start time.Time) {
sampler := func() metrics.Sample {
return metrics.ResettingSample(
metrics.NewExpDecaySample(1028, 0.015),
)
}
metrics.GetOrRegisterHistogramLazy(h, nil, sampler).Update(time.Since(start).Microseconds())
}(start)
}
// Handle the message depending on its contents
switch {
case msg.Code == GetAccountRangeMsg:
// Decode the account retrieval request
var req GetAccountRangePacket
if err := msg.Decode(&req); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
// Service the request, potentially returning nothing in case of errors
accounts, proofs := ServiceGetAccountRangeQuery(backend.Chain(), &req)
// Send back anything accumulated (or empty in case of errors)
return p2p.Send(peer.rw, AccountRangeMsg, &AccountRangePacket{
ID: req.ID,
Accounts: accounts,
Proof: proofs,
})
case msg.Code == AccountRangeMsg:
// A range of accounts arrived to one of our previous requests
res := new(AccountRangePacket)
if err := msg.Decode(res); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
// Ensure the range is monotonically increasing
for i := 1; i < len(res.Accounts); i++ {
if bytes.Compare(res.Accounts[i-1].Hash[:], res.Accounts[i].Hash[:]) >= 0 {
return fmt.Errorf("accounts not monotonically increasing: #%d [%x] vs #%d [%x]", i-1, res.Accounts[i-1].Hash[:], i, res.Accounts[i].Hash[:])
}
}
requestTracker.Fulfil(peer.id, peer.version, AccountRangeMsg, res.ID)
return backend.Handle(peer, res)
case msg.Code == GetStorageRangesMsg:
// Decode the storage retrieval request
var req GetStorageRangesPacket
if err := msg.Decode(&req); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
// Service the request, potentially returning nothing in case of errors
slots, proofs := ServiceGetStorageRangesQuery(backend.Chain(), &req)
// Send back anything accumulated (or empty in case of errors)
return p2p.Send(peer.rw, StorageRangesMsg, &StorageRangesPacket{
ID: req.ID,
Slots: slots,
Proof: proofs,
})
case msg.Code == StorageRangesMsg:
// A range of storage slots arrived to one of our previous requests
res := new(StorageRangesPacket)
if err := msg.Decode(res); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
eth/protocols/snap: generate storage trie from full dirty snap data (#22668) * eth/protocols/snap: generate storage trie from full dirty snap data * eth/protocols/snap: get rid of some more dead code * eth/protocols/snap: less frequent logs, also log during trie generation * eth/protocols/snap: implement dirty account range stack-hashing * eth/protocols/snap: don't loop on account trie generation * eth/protocols/snap: fix account format in trie * core, eth, ethdb: glue snap packets together, but not chunks * eth/protocols/snap: print completion log for snap phase * eth/protocols/snap: extended tests * eth/protocols/snap: make testcase pass * eth/protocols/snap: fix account stacktrie commit without defer * ethdb: fix key counts on reset * eth/protocols: fix typos * eth/protocols/snap: make better use of delivered data (#44) * eth/protocols/snap: make better use of delivered data * squashme * eth/protocols/snap: reduce chunking * squashme * eth/protocols/snap: reduce chunking further * eth/protocols/snap: break out hash range calculations * eth/protocols/snap: use sort.Search instead of looping * eth/protocols/snap: prevent crash on storage response with no keys * eth/protocols/snap: nitpicks all around * eth/protocols/snap: clear heal need on 1-chunk storage completion * eth/protocols/snap: fix range chunker, add tests Co-authored-by: Péter Szilágyi <peterke@gmail.com> * trie: fix test API error * eth/protocols/snap: fix some further liter issues * eth/protocols/snap: fix accidental batch reuse Co-authored-by: Martin Holst Swende <martin@swende.se>
2021-04-27 09:19:59 -05:00
// Ensure the ranges are monotonically increasing
for i, slots := range res.Slots {
for j := 1; j < len(slots); j++ {
if bytes.Compare(slots[j-1].Hash[:], slots[j].Hash[:]) >= 0 {
return fmt.Errorf("storage slots not monotonically increasing for account #%d: #%d [%x] vs #%d [%x]", i, j-1, slots[j-1].Hash[:], j, slots[j].Hash[:])
}
}
}
requestTracker.Fulfil(peer.id, peer.version, StorageRangesMsg, res.ID)
return backend.Handle(peer, res)
case msg.Code == GetByteCodesMsg:
// Decode bytecode retrieval request
var req GetByteCodesPacket
if err := msg.Decode(&req); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
// Service the request, potentially returning nothing in case of errors
codes := ServiceGetByteCodesQuery(backend.Chain(), &req)
// Send back anything accumulated (or empty in case of errors)
return p2p.Send(peer.rw, ByteCodesMsg, &ByteCodesPacket{
ID: req.ID,
Codes: codes,
})
case msg.Code == ByteCodesMsg:
// A batch of byte codes arrived to one of our previous requests
res := new(ByteCodesPacket)
if err := msg.Decode(res); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
requestTracker.Fulfil(peer.id, peer.version, ByteCodesMsg, res.ID)
return backend.Handle(peer, res)
case msg.Code == GetTrieNodesMsg:
// Decode trie node retrieval request
var req GetTrieNodesPacket
if err := msg.Decode(&req); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
// Service the request, potentially returning nothing in case of errors
nodes, err := ServiceGetTrieNodesQuery(backend.Chain(), &req, start)
if err != nil {
return err
}
// Send back anything accumulated (or empty in case of errors)
return p2p.Send(peer.rw, TrieNodesMsg, &TrieNodesPacket{
ID: req.ID,
Nodes: nodes,
})
case msg.Code == TrieNodesMsg:
// A batch of trie nodes arrived to one of our previous requests
res := new(TrieNodesPacket)
if err := msg.Decode(res); err != nil {
return fmt.Errorf("%w: message %v: %v", errDecode, msg, err)
}
requestTracker.Fulfil(peer.id, peer.version, TrieNodesMsg, res.ID)
return backend.Handle(peer, res)
default:
return fmt.Errorf("%w: %v", errInvalidMsgCode, msg.Code)
}
}
// ServiceGetAccountRangeQuery assembles the response to an account range query.
// It is exposed to allow external packages to test protocol behavior.
func ServiceGetAccountRangeQuery(chain *core.BlockChain, req *GetAccountRangePacket) ([]*AccountData, [][]byte) {
if req.Bytes > softResponseLimit {
req.Bytes = softResponseLimit
}
// Retrieve the requested state and bail out if non existent
tr, err := trie.New(trie.StateTrieID(req.Root), chain.TrieDB())
if err != nil {
return nil, nil
}
it, err := chain.Snapshots().AccountIterator(req.Root, req.Origin)
if err != nil {
return nil, nil
}
// Iterate over the requested range and pile accounts up
var (
accounts []*AccountData
size uint64
last common.Hash
)
for it.Next() {
hash, account := it.Hash(), common.CopyBytes(it.Account())
// Track the returned interval for the Merkle proofs
last = hash
// Assemble the reply item
size += uint64(common.HashLength + len(account))
accounts = append(accounts, &AccountData{
Hash: hash,
Body: account,
})
// If we've exceeded the request threshold, abort
if bytes.Compare(hash[:], req.Limit[:]) >= 0 {
break
}
if size > req.Bytes {
break
}
}
it.Release()
// Generate the Merkle proofs for the first and last account
proof := trienode.NewProofSet()
if err := tr.Prove(req.Origin[:], proof); err != nil {
log.Warn("Failed to prove account range", "origin", req.Origin, "err", err)
return nil, nil
}
if last != (common.Hash{}) {
if err := tr.Prove(last[:], proof); err != nil {
log.Warn("Failed to prove account range", "last", last, "err", err)
return nil, nil
}
}
return accounts, proof.List()
}
func ServiceGetStorageRangesQuery(chain *core.BlockChain, req *GetStorageRangesPacket) ([][]*StorageData, [][]byte) {
if req.Bytes > softResponseLimit {
req.Bytes = softResponseLimit
}
// TODO(karalabe): Do we want to enforce > 0 accounts and 1 account if origin is set?
2022-10-11 02:37:00 -05:00
// TODO(karalabe): - Logging locally is not ideal as remote faults annoy the local user
// TODO(karalabe): - Dropping the remote peer is less flexible wrt client bugs (slow is better than non-functional)
// Calculate the hard limit at which to abort, even if mid storage trie
hardLimit := uint64(float64(req.Bytes) * (1 + stateLookupSlack))
// Retrieve storage ranges until the packet limit is reached
var (
slots [][]*StorageData
proofs [][]byte
size uint64
)
for _, account := range req.Accounts {
// If we've exceeded the requested data limit, abort without opening
// a new storage range (that we'd need to prove due to exceeded size)
if size >= req.Bytes {
break
}
// The first account might start from a different origin and end sooner
var origin common.Hash
if len(req.Origin) > 0 {
origin, req.Origin = common.BytesToHash(req.Origin), nil
}
var limit = common.MaxHash
if len(req.Limit) > 0 {
limit, req.Limit = common.BytesToHash(req.Limit), nil
}
// Retrieve the requested state and bail out if non existent
it, err := chain.Snapshots().StorageIterator(req.Root, account, origin)
if err != nil {
return nil, nil
}
// Iterate over the requested range and pile slots up
var (
storage []*StorageData
last common.Hash
abort bool
)
for it.Next() {
if size >= hardLimit {
abort = true
break
}
hash, slot := it.Hash(), common.CopyBytes(it.Slot())
// Track the returned interval for the Merkle proofs
last = hash
// Assemble the reply item
size += uint64(common.HashLength + len(slot))
storage = append(storage, &StorageData{
Hash: hash,
Body: slot,
})
// If we've exceeded the request threshold, abort
if bytes.Compare(hash[:], limit[:]) >= 0 {
break
}
}
if len(storage) > 0 {
slots = append(slots, storage)
}
it.Release()
// Generate the Merkle proofs for the first and last storage slot, but
// only if the response was capped. If the entire storage trie included
// in the response, no need for any proofs.
if origin != (common.Hash{}) || (abort && len(storage) > 0) {
// Request started at a non-zero hash or was capped prematurely, add
// the endpoint Merkle proofs
accTrie, err := trie.NewStateTrie(trie.StateTrieID(req.Root), chain.TrieDB())
if err != nil {
return nil, nil
}
acc, err := accTrie.GetAccountByHash(account)
if err != nil || acc == nil {
return nil, nil
}
id := trie.StorageTrieID(req.Root, account, acc.Root)
stTrie, err := trie.NewStateTrie(id, chain.TrieDB())
if err != nil {
return nil, nil
}
proof := trienode.NewProofSet()
if err := stTrie.Prove(origin[:], proof); err != nil {
log.Warn("Failed to prove storage range", "origin", req.Origin, "err", err)
return nil, nil
}
if last != (common.Hash{}) {
if err := stTrie.Prove(last[:], proof); err != nil {
log.Warn("Failed to prove storage range", "last", last, "err", err)
return nil, nil
}
}
proofs = append(proofs, proof.List()...)
// Proof terminates the reply as proofs are only added if a node
// refuses to serve more data (exception when a contract fetch is
// finishing, but that's that).
break
}
}
return slots, proofs
}
// ServiceGetByteCodesQuery assembles the response to a byte codes query.
// It is exposed to allow external packages to test protocol behavior.
func ServiceGetByteCodesQuery(chain *core.BlockChain, req *GetByteCodesPacket) [][]byte {
if req.Bytes > softResponseLimit {
req.Bytes = softResponseLimit
}
if len(req.Hashes) > maxCodeLookups {
req.Hashes = req.Hashes[:maxCodeLookups]
}
// Retrieve bytecodes until the packet size limit is reached
var (
codes [][]byte
bytes uint64
)
for _, hash := range req.Hashes {
if hash == types.EmptyCodeHash {
// Peers should not request the empty code, but if they do, at
// least sent them back a correct response without db lookups
codes = append(codes, []byte{})
} else if blob, err := chain.ContractCodeWithPrefix(hash); err == nil {
codes = append(codes, blob)
bytes += uint64(len(blob))
}
if bytes > req.Bytes {
break
}
}
return codes
}
// ServiceGetTrieNodesQuery assembles the response to a trie nodes query.
// It is exposed to allow external packages to test protocol behavior.
func ServiceGetTrieNodesQuery(chain *core.BlockChain, req *GetTrieNodesPacket, start time.Time) ([][]byte, error) {
if req.Bytes > softResponseLimit {
req.Bytes = softResponseLimit
}
// Make sure we have the state associated with the request
triedb := chain.TrieDB()
accTrie, err := trie.NewStateTrie(trie.StateTrieID(req.Root), triedb)
if err != nil {
// We don't have the requested state available, bail out
return nil, nil
}
// The 'snap' might be nil, in which case we cannot serve storage slots.
snap := chain.Snapshots().Snapshot(req.Root)
// Retrieve trie nodes until the packet size limit is reached
var (
nodes [][]byte
bytes uint64
loads int // Trie hash expansions to count database reads
)
for _, pathset := range req.Paths {
switch len(pathset) {
case 0:
// Ensure we penalize invalid requests
return nil, fmt.Errorf("%w: zero-item pathset requested", errBadRequest)
case 1:
// If we're only retrieving an account trie node, fetch it directly
blob, resolved, err := accTrie.GetNode(pathset[0])
loads += resolved // always account database reads, even for failures
if err != nil {
break
}
nodes = append(nodes, blob)
bytes += uint64(len(blob))
default:
var stRoot common.Hash
// Storage slots requested, open the storage trie and retrieve from there
if snap == nil {
// We don't have the requested state snapshotted yet (or it is stale),
// but can look up the account via the trie instead.
account, err := accTrie.GetAccountByHash(common.BytesToHash(pathset[0]))
loads += 8 // We don't know the exact cost of lookup, this is an estimate
if err != nil || account == nil {
break
}
stRoot = account.Root
} else {
account, err := snap.Account(common.BytesToHash(pathset[0]))
loads++ // always account database reads, even for failures
if err != nil || account == nil {
break
}
stRoot = common.BytesToHash(account.Root)
}
id := trie.StorageTrieID(req.Root, common.BytesToHash(pathset[0]), stRoot)
stTrie, err := trie.NewStateTrie(id, triedb)
loads++ // always account database reads, even for failures
if err != nil {
break
}
for _, path := range pathset[1:] {
blob, resolved, err := stTrie.GetNode(path)
loads += resolved // always account database reads, even for failures
if err != nil {
break
}
nodes = append(nodes, blob)
bytes += uint64(len(blob))
// Sanity check limits to avoid DoS on the store trie loads
if bytes > req.Bytes || loads > maxTrieNodeLookups || time.Since(start) > maxTrieNodeTimeSpent {
break
}
}
}
// Abort request processing if we've exceeded our limits
if bytes > req.Bytes || loads > maxTrieNodeLookups || time.Since(start) > maxTrieNodeTimeSpent {
break
}
}
return nodes, nil
}
// NodeInfo represents a short summary of the `snap` sub-protocol metadata
// known about the host peer.
type NodeInfo struct{}
// nodeInfo retrieves some `snap` protocol metadata about the running host node.
func nodeInfo(chain *core.BlockChain) *NodeInfo {
return &NodeInfo{}
}