Official Go implementation of the Ethereum protocol
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
go-ethereum/beacon/light/sync/test_helpers.go

259 lines
7.7 KiB

// 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 <http://www.gnu.org/licenses/>.
package sync
import (
"reflect"
"testing"
"github.com/ethereum/go-ethereum/beacon/light"
"github.com/ethereum/go-ethereum/beacon/light/request"
"github.com/ethereum/go-ethereum/beacon/types"
)
type requestWithID struct {
sid request.ServerAndID
request request.Request
}
type TestScheduler struct {
t *testing.T
module request.Module
events []request.Event
servers []request.Server
allowance map[request.Server]int
sent map[int][]requestWithID
testIndex int
expFail map[request.Server]int // expected Server.Fail calls during next Run
lastId request.ID
}
func NewTestScheduler(t *testing.T, module request.Module) *TestScheduler {
return &TestScheduler{
t: t,
module: module,
allowance: make(map[request.Server]int),
expFail: make(map[request.Server]int),
sent: make(map[int][]requestWithID),
}
}
func (ts *TestScheduler) Run(testIndex int, exp ...any) {
expReqs := make([]requestWithID, len(exp)/2)
id := ts.lastId
for i := range expReqs {
id++
expReqs[i] = requestWithID{
sid: request.ServerAndID{Server: exp[i*2].(request.Server), ID: id},
request: exp[i*2+1].(request.Request),
}
}
if len(expReqs) == 0 {
expReqs = nil
}
ts.testIndex = testIndex
ts.module.Process(ts, ts.events)
ts.events = nil
for server, count := range ts.expFail {
delete(ts.expFail, server)
if count == 0 {
continue
}
ts.t.Errorf("Missing %d Server.Fail(s) from server %s in test case #%d", count, server.Name(), testIndex)
}
if !reflect.DeepEqual(ts.sent[testIndex], expReqs) {
ts.t.Errorf("Wrong sent requests in test case #%d (expected %v, got %v)", testIndex, expReqs, ts.sent[testIndex])
}
}
func (ts *TestScheduler) CanSendTo() (cs []request.Server) {
for _, server := range ts.servers {
if ts.allowance[server] > 0 {
cs = append(cs, server)
}
}
return
}
func (ts *TestScheduler) Send(server request.Server, req request.Request) request.ID {
ts.lastId++
ts.sent[ts.testIndex] = append(ts.sent[ts.testIndex], requestWithID{
sid: request.ServerAndID{Server: server, ID: ts.lastId},
request: req,
})
ts.allowance[server]--
return ts.lastId
}
func (ts *TestScheduler) Fail(server request.Server, desc string) {
if ts.expFail[server] == 0 {
ts.t.Errorf("Unexpected Fail from server %s in test case #%d: %s", server.Name(), ts.testIndex, desc)
return
}
ts.expFail[server]--
}
func (ts *TestScheduler) Request(testIndex, reqIndex int) requestWithID {
if len(ts.sent[testIndex]) < reqIndex {
ts.t.Errorf("Missing request from test case %d index %d", testIndex, reqIndex)
return requestWithID{}
}
return ts.sent[testIndex][reqIndex-1]
}
func (ts *TestScheduler) ServerEvent(evType *request.EventType, server request.Server, data any) {
ts.events = append(ts.events, request.Event{
Type: evType,
Server: server,
Data: data,
})
}
func (ts *TestScheduler) RequestEvent(evType *request.EventType, req requestWithID, resp request.Response) {
if req.request == nil {
return
}
ts.events = append(ts.events, request.Event{
Type: evType,
Server: req.sid.Server,
Data: request.RequestResponse{
ID: req.sid.ID,
Request: req.request,
Response: resp,
},
})
}
func (ts *TestScheduler) AddServer(server request.Server, allowance int) {
ts.servers = append(ts.servers, server)
ts.allowance[server] = allowance
ts.ServerEvent(request.EvRegistered, server, nil)
}
func (ts *TestScheduler) RemoveServer(server request.Server) {
ts.servers = append(ts.servers, server)
for i, s := range ts.servers {
if s == server {
copy(ts.servers[i:len(ts.servers)-1], ts.servers[i+1:])
ts.servers = ts.servers[:len(ts.servers)-1]
break
}
}
delete(ts.allowance, server)
ts.ServerEvent(request.EvUnregistered, server, nil)
}
func (ts *TestScheduler) AddAllowance(server request.Server, allowance int) {
ts.allowance[server] += allowance
}
func (ts *TestScheduler) ExpFail(server request.Server) {
ts.expFail[server]++
}
type TestCommitteeChain struct {
fsp, nsp uint64
init bool
}
func (tc *TestCommitteeChain) CheckpointInit(bootstrap types.BootstrapData) error {
tc.fsp, tc.nsp, tc.init = bootstrap.Header.SyncPeriod(), bootstrap.Header.SyncPeriod()+2, true
return nil
}
func (tc *TestCommitteeChain) InsertUpdate(update *types.LightClientUpdate, nextCommittee *types.SerializedSyncCommittee) error {
period := update.AttestedHeader.Header.SyncPeriod()
if period < tc.fsp || period > tc.nsp || !tc.init {
return light.ErrInvalidPeriod
}
if period == tc.nsp {
tc.nsp++
}
return nil
}
func (tc *TestCommitteeChain) NextSyncPeriod() (uint64, bool) {
return tc.nsp, tc.init
}
func (tc *TestCommitteeChain) ExpInit(t *testing.T, ExpInit bool) {
if tc.init != ExpInit {
t.Errorf("Incorrect init flag (expected %v, got %v)", ExpInit, tc.init)
}
}
func (tc *TestCommitteeChain) SetNextSyncPeriod(nsp uint64) {
tc.init, tc.nsp = true, nsp
}
func (tc *TestCommitteeChain) ExpNextSyncPeriod(t *testing.T, expNsp uint64) {
tc.ExpInit(t, true)
if tc.nsp != expNsp {
t.Errorf("Incorrect NextSyncPeriod (expected %d, got %d)", expNsp, tc.nsp)
}
}
type TestHeadTracker struct {
phead types.HeadInfo
validated []types.OptimisticUpdate
finality types.FinalityUpdate
}
func (ht *TestHeadTracker) ValidateOptimistic(update types.OptimisticUpdate) (bool, error) {
ht.validated = append(ht.validated, update)
return true, nil
}
func (ht *TestHeadTracker) ValidateFinality(update types.FinalityUpdate) (bool, error) {
ht.finality = update
return true, nil
}
func (ht *TestHeadTracker) ValidatedFinality() (types.FinalityUpdate, bool) {
return ht.finality, ht.finality.Attested.Header != (types.Header{})
}
func (ht *TestHeadTracker) ExpValidated(t *testing.T, tci int, expHeads []types.OptimisticUpdate) {
for i, expHead := range expHeads {
if i >= len(ht.validated) {
t.Errorf("Missing validated head in test case #%d index #%d (expected {slot %d blockRoot %x}, got none)", tci, i, expHead.Attested.Header.Slot, expHead.Attested.Header.Hash())
continue
}
if !reflect.DeepEqual(ht.validated[i], expHead) {
vhead := ht.validated[i].Attested.Header
t.Errorf("Wrong validated head in test case #%d index #%d (expected {slot %d blockRoot %x}, got {slot %d blockRoot %x})", tci, i, expHead.Attested.Header.Slot, expHead.Attested.Header.Hash(), vhead.Slot, vhead.Hash())
}
}
for i := len(expHeads); i < len(ht.validated); i++ {
vhead := ht.validated[i].Attested.Header
t.Errorf("Unexpected validated head in test case #%d index #%d (expected none, got {slot %d blockRoot %x})", tci, i, vhead.Slot, vhead.Hash())
}
ht.validated = nil
}
func (ht *TestHeadTracker) SetPrefetchHead(head types.HeadInfo) {
ht.phead = head
}
func (ht *TestHeadTracker) ExpPrefetch(t *testing.T, tci int, exp types.HeadInfo) {
if ht.phead != exp {
t.Errorf("Wrong prefetch head in test case #%d (expected {slot %d blockRoot %x}, got {slot %d blockRoot %x})", tci, exp.Slot, exp.BlockRoot, ht.phead.Slot, ht.phead.BlockRoot)
}
}