aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorgary rong <garyrong0905@gmail.com>2018-08-28 21:59:05 +0800
committerPéter Szilágyi <peterke@gmail.com>2018-08-28 21:59:05 +0800
commitc1c003e4ff36c22d67662ca661fc78cde850d401 (patch)
treeb8bea54350fb6894cfd63ebc87a164acc3fba7e6
parent63352bf4247f05d8ef255ff8c63290225c3bc671 (diff)
downloaddexon-c1c003e4ff36c22d67662ca661fc78cde850d401.tar
dexon-c1c003e4ff36c22d67662ca661fc78cde850d401.tar.gz
dexon-c1c003e4ff36c22d67662ca661fc78cde850d401.tar.bz2
dexon-c1c003e4ff36c22d67662ca661fc78cde850d401.tar.lz
dexon-c1c003e4ff36c22d67662ca661fc78cde850d401.tar.xz
dexon-c1c003e4ff36c22d67662ca661fc78cde850d401.tar.zst
dexon-c1c003e4ff36c22d67662ca661fc78cde850d401.zip
consensus, miner: stale block mining support (#17506)
* consensus, miner: stale block supporting * consensus, miner: refactor seal signature * cmd, consensus, eth: add miner noverify flag * cmd, consensus, miner: polish
-rw-r--r--cmd/geth/main.go1
-rw-r--r--cmd/geth/usage.go1
-rw-r--r--cmd/utils/flags.go11
-rw-r--r--consensus/clique/clique.go39
-rw-r--r--consensus/consensus.go9
-rw-r--r--consensus/ethash/algorithm_test.go2
-rw-r--r--consensus/ethash/ethash.go33
-rw-r--r--consensus/ethash/ethash_test.go40
-rw-r--r--consensus/ethash/sealer.go134
-rw-r--r--consensus/ethash/sealer_test.go94
-rw-r--r--eth/api_tracer.go2
-rw-r--r--eth/backend.go8
-rw-r--r--eth/config.go1
-rw-r--r--eth/gen_config.go7
-rw-r--r--les/backend.go2
-rw-r--r--miner/worker.go116
16 files changed, 317 insertions, 183 deletions
diff --git a/cmd/geth/main.go b/cmd/geth/main.go
index 7b8a8244e..0ed67d0d5 100644
--- a/cmd/geth/main.go
+++ b/cmd/geth/main.go
@@ -108,6 +108,7 @@ var (
utils.MinerExtraDataFlag,
utils.MinerLegacyExtraDataFlag,
utils.MinerRecommitIntervalFlag,
+ utils.MinerNoVerfiyFlag,
utils.NATFlag,
utils.NoDiscoverFlag,
utils.DiscoveryV5Flag,
diff --git a/cmd/geth/usage.go b/cmd/geth/usage.go
index 1e27d0ae8..2f8260e0f 100644
--- a/cmd/geth/usage.go
+++ b/cmd/geth/usage.go
@@ -192,6 +192,7 @@ var AppHelpFlagGroups = []flagGroup{
utils.MinerEtherbaseFlag,
utils.MinerExtraDataFlag,
utils.MinerRecommitIntervalFlag,
+ utils.MinerNoVerfiyFlag,
},
},
{
diff --git a/cmd/utils/flags.go b/cmd/utils/flags.go
index b9a33ffe7..13430ad56 100644
--- a/cmd/utils/flags.go
+++ b/cmd/utils/flags.go
@@ -366,9 +366,13 @@ var (
}
MinerRecommitIntervalFlag = cli.DurationFlag{
Name: "miner.recommit",
- Usage: "Time interval to recreate the block being mined.",
+ Usage: "Time interval to recreate the block being mined",
Value: eth.DefaultConfig.MinerRecommit,
}
+ MinerNoVerfiyFlag = cli.BoolFlag{
+ Name: "miner.noverify",
+ Usage: "Disable remote sealing verification",
+ }
// Account settings
UnlockedAccountFlag = cli.StringFlag{
Name: "unlock",
@@ -1151,6 +1155,9 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *eth.Config) {
if ctx.GlobalIsSet(MinerRecommitIntervalFlag.Name) {
cfg.MinerRecommit = ctx.Duration(MinerRecommitIntervalFlag.Name)
}
+ if ctx.GlobalIsSet(MinerNoVerfiyFlag.Name) {
+ cfg.MinerNoverify = ctx.Bool(MinerNoVerfiyFlag.Name)
+ }
if ctx.GlobalIsSet(VMEnableDebugFlag.Name) {
// TODO(fjl): force-enable this in --dev mode
cfg.EnablePreimageRecording = ctx.GlobalBool(VMEnableDebugFlag.Name)
@@ -1345,7 +1352,7 @@ func MakeChain(ctx *cli.Context, stack *node.Node) (chain *core.BlockChain, chai
DatasetDir: stack.ResolvePath(eth.DefaultConfig.Ethash.DatasetDir),
DatasetsInMem: eth.DefaultConfig.Ethash.DatasetsInMem,
DatasetsOnDisk: eth.DefaultConfig.Ethash.DatasetsOnDisk,
- }, nil)
+ }, nil, false)
}
}
if gcmode := ctx.GlobalString(GCModeFlag.Name); gcmode != "full" && gcmode != "archive" {
diff --git a/consensus/clique/clique.go b/consensus/clique/clique.go
index 3730c91f6..547290984 100644
--- a/consensus/clique/clique.go
+++ b/consensus/clique/clique.go
@@ -590,17 +590,17 @@ func (c *Clique) Authorize(signer common.Address, signFn SignerFn) {
// Seal implements consensus.Engine, attempting to create a sealed block using
// the local signing credentials.
-func (c *Clique) Seal(chain consensus.ChainReader, block *types.Block, stop <-chan struct{}) (*types.Block, error) {
+func (c *Clique) Seal(chain consensus.ChainReader, block *types.Block, results chan<- *types.Block, stop <-chan struct{}) error {
header := block.Header()
// Sealing the genesis block is not supported
number := header.Number.Uint64()
if number == 0 {
- return nil, errUnknownBlock
+ return errUnknownBlock
}
// For 0-period chains, refuse to seal empty blocks (no reward but would spin sealing)
if c.config.Period == 0 && len(block.Transactions()) == 0 {
- return nil, errWaitTransactions
+ return errWaitTransactions
}
// Don't hold the signer fields for the entire sealing procedure
c.lock.RLock()
@@ -610,10 +610,10 @@ func (c *Clique) Seal(chain consensus.ChainReader, block *types.Block, stop <-ch
// Bail out if we're unauthorized to sign a block
snap, err := c.snapshot(chain, number-1, header.ParentHash, nil)
if err != nil {
- return nil, err
+ return err
}
if _, authorized := snap.Signers[signer]; !authorized {
- return nil, errUnauthorized
+ return errUnauthorized
}
// If we're amongst the recent signers, wait for the next block
for seen, recent := range snap.Recents {
@@ -621,8 +621,7 @@ func (c *Clique) Seal(chain consensus.ChainReader, block *types.Block, stop <-ch
// Signer is among recents, only wait if the current block doesn't shift it out
if limit := uint64(len(snap.Signers)/2 + 1); number < limit || seen > number-limit {
log.Info("Signed recently, must wait for others")
- <-stop
- return nil, nil
+ return nil
}
}
}
@@ -635,21 +634,29 @@ func (c *Clique) Seal(chain consensus.ChainReader, block *types.Block, stop <-ch
log.Trace("Out-of-turn signing requested", "wiggle", common.PrettyDuration(wiggle))
}
- log.Trace("Waiting for slot to sign and propagate", "delay", common.PrettyDuration(delay))
-
- select {
- case <-stop:
- return nil, nil
- case <-time.After(delay):
- }
// Sign all the things!
sighash, err := signFn(accounts.Account{Address: signer}, sigHash(header).Bytes())
if err != nil {
- return nil, err
+ return err
}
copy(header.Extra[len(header.Extra)-extraSeal:], sighash)
+ // Wait until sealing is terminated or delay timeout.
+ log.Trace("Waiting for slot to sign and propagate", "delay", common.PrettyDuration(delay))
+ go func() {
+ select {
+ case <-stop:
+ return
+ case <-time.After(delay):
+ }
- return block.WithSeal(header), nil
+ select {
+ case results <- block.WithSeal(header):
+ default:
+ log.Warn("Sealing result is not read by miner", "sealhash", c.SealHash(header))
+ }
+ }()
+
+ return nil
}
// CalcDifficulty is the difficulty adjustment algorithm. It returns the difficulty
diff --git a/consensus/consensus.go b/consensus/consensus.go
index 27799f13c..12ede7ff4 100644
--- a/consensus/consensus.go
+++ b/consensus/consensus.go
@@ -86,9 +86,12 @@ type Engine interface {
Finalize(chain ChainReader, header *types.Header, state *state.StateDB, txs []*types.Transaction,
uncles []*types.Header, receipts []*types.Receipt) (*types.Block, error)
- // Seal generates a new block for the given input block with the local miner's
- // seal place on top.
- Seal(chain ChainReader, block *types.Block, stop <-chan struct{}) (*types.Block, error)
+ // Seal generates a new sealing request for the given input block and pushes
+ // the result into the given channel.
+ //
+ // Note, the method returns immediately and will send the result async. More
+ // than one result may also be returned depending on the consensus algorothm.
+ Seal(chain ChainReader, block *types.Block, results chan<- *types.Block, stop <-chan struct{}) error
// SealHash returns the hash of a block prior to it being sealed.
SealHash(header *types.Header) common.Hash
diff --git a/consensus/ethash/algorithm_test.go b/consensus/ethash/algorithm_test.go
index db22cccd0..c58479e28 100644
--- a/consensus/ethash/algorithm_test.go
+++ b/consensus/ethash/algorithm_test.go
@@ -729,7 +729,7 @@ func TestConcurrentDiskCacheGeneration(t *testing.T) {
go func(idx int) {
defer pend.Done()
- ethash := New(Config{cachedir, 0, 1, "", 0, 0, ModeNormal}, nil)
+ ethash := New(Config{cachedir, 0, 1, "", 0, 0, ModeNormal}, nil, false)
defer ethash.Close()
if err := ethash.VerifySeal(nil, block.Header()); err != nil {
t.Errorf("proc %d: block verification failed: %v", idx, err)
diff --git a/consensus/ethash/ethash.go b/consensus/ethash/ethash.go
index d98c3371c..b4819ca38 100644
--- a/consensus/ethash/ethash.go
+++ b/consensus/ethash/ethash.go
@@ -50,7 +50,7 @@ var (
two256 = new(big.Int).Exp(big.NewInt(2), big.NewInt(256), big.NewInt(0))
// sharedEthash is a full instance that can be shared between multiple users.
- sharedEthash = New(Config{"", 3, 0, "", 1, 0, ModeNormal}, nil)
+ sharedEthash = New(Config{"", 3, 0, "", 1, 0, ModeNormal}, nil, false)
// algorithmRevision is the data structure version used for file naming.
algorithmRevision = 23
@@ -405,6 +405,12 @@ type Config struct {
PowMode Mode
}
+// sealTask wraps a seal block with relative result channel for remote sealer thread.
+type sealTask struct {
+ block *types.Block
+ results chan<- *types.Block
+}
+
// mineResult wraps the pow solution parameters for the specified block.
type mineResult struct {
nonce types.BlockNonce
@@ -444,12 +450,11 @@ type Ethash struct {
hashrate metrics.Meter // Meter tracking the average hashrate
// Remote sealer related fields
- workCh chan *types.Block // Notification channel to push new work to remote sealer
- resultCh chan *types.Block // Channel used by mining threads to return result
- fetchWorkCh chan *sealWork // Channel used for remote sealer to fetch mining work
- submitWorkCh chan *mineResult // Channel used for remote sealer to submit their mining result
- fetchRateCh chan chan uint64 // Channel used to gather submitted hash rate for local or remote sealer.
- submitRateCh chan *hashrate // Channel used for remote sealer to submit their mining hashrate
+ workCh chan *sealTask // Notification channel to push new work and relative result channel to remote sealer
+ fetchWorkCh chan *sealWork // Channel used for remote sealer to fetch mining work
+ submitWorkCh chan *mineResult // Channel used for remote sealer to submit their mining result
+ fetchRateCh chan chan uint64 // Channel used to gather submitted hash rate for local or remote sealer.
+ submitRateCh chan *hashrate // Channel used for remote sealer to submit their mining hashrate
// The fields below are hooks for testing
shared *Ethash // Shared PoW verifier to avoid cache regeneration
@@ -464,7 +469,7 @@ type Ethash struct {
// New creates a full sized ethash PoW scheme and starts a background thread for
// remote mining, also optionally notifying a batch of remote services of new work
// packages.
-func New(config Config, notify []string) *Ethash {
+func New(config Config, notify []string, noverify bool) *Ethash {
if config.CachesInMem <= 0 {
log.Warn("One ethash cache must always be in memory", "requested", config.CachesInMem)
config.CachesInMem = 1
@@ -481,36 +486,34 @@ func New(config Config, notify []string) *Ethash {
datasets: newlru("dataset", config.DatasetsInMem, newDataset),
update: make(chan struct{}),
hashrate: metrics.NewMeter(),
- workCh: make(chan *types.Block),
- resultCh: make(chan *types.Block),
+ workCh: make(chan *sealTask),
fetchWorkCh: make(chan *sealWork),
submitWorkCh: make(chan *mineResult),
fetchRateCh: make(chan chan uint64),
submitRateCh: make(chan *hashrate),
exitCh: make(chan chan error),
}
- go ethash.remote(notify)
+ go ethash.remote(notify, noverify)
return ethash
}
// NewTester creates a small sized ethash PoW scheme useful only for testing
// purposes.
-func NewTester(notify []string) *Ethash {
+func NewTester(notify []string, noverify bool) *Ethash {
ethash := &Ethash{
config: Config{PowMode: ModeTest},
caches: newlru("cache", 1, newCache),
datasets: newlru("dataset", 1, newDataset),
update: make(chan struct{}),
hashrate: metrics.NewMeter(),
- workCh: make(chan *types.Block),
- resultCh: make(chan *types.Block),
+ workCh: make(chan *sealTask),
fetchWorkCh: make(chan *sealWork),
submitWorkCh: make(chan *mineResult),
fetchRateCh: make(chan chan uint64),
submitRateCh: make(chan *hashrate),
exitCh: make(chan chan error),
}
- go ethash.remote(notify)
+ go ethash.remote(notify, noverify)
return ethash
}
diff --git a/consensus/ethash/ethash_test.go b/consensus/ethash/ethash_test.go
index b190d63d6..8eded2ca8 100644
--- a/consensus/ethash/ethash_test.go
+++ b/consensus/ethash/ethash_test.go
@@ -34,17 +34,23 @@ import (
func TestTestMode(t *testing.T) {
header := &types.Header{Number: big.NewInt(1), Difficulty: big.NewInt(100)}
- ethash := NewTester(nil)
+ ethash := NewTester(nil, false)
defer ethash.Close()
- block, err := ethash.Seal(nil, types.NewBlockWithHeader(header), nil)
+ results := make(chan *types.Block)
+ err := ethash.Seal(nil, types.NewBlockWithHeader(header), results, nil)
if err != nil {
t.Fatalf("failed to seal block: %v", err)
}
- header.Nonce = types.EncodeNonce(block.Nonce())
- header.MixDigest = block.MixDigest()
- if err := ethash.VerifySeal(nil, header); err != nil {
- t.Fatalf("unexpected verification error: %v", err)
+ select {
+ case block := <-results:
+ header.Nonce = types.EncodeNonce(block.Nonce())
+ header.MixDigest = block.MixDigest()
+ if err := ethash.VerifySeal(nil, header); err != nil {
+ t.Fatalf("unexpected verification error: %v", err)
+ }
+ case <-time.NewTimer(time.Second).C:
+ t.Error("sealing result timeout")
}
}
@@ -56,7 +62,7 @@ func TestCacheFileEvict(t *testing.T) {
t.Fatal(err)
}
defer os.RemoveAll(tmpdir)
- e := New(Config{CachesInMem: 3, CachesOnDisk: 10, CacheDir: tmpdir, PowMode: ModeTest}, nil)
+ e := New(Config{CachesInMem: 3, CachesOnDisk: 10, CacheDir: tmpdir, PowMode: ModeTest}, nil, false)
defer e.Close()
workers := 8
@@ -85,7 +91,7 @@ func verifyTest(wg *sync.WaitGroup, e *Ethash, workerIndex, epochs int) {
}
func TestRemoteSealer(t *testing.T) {
- ethash := NewTester(nil)
+ ethash := NewTester(nil, false)
defer ethash.Close()
api := &API{ethash}
@@ -97,7 +103,8 @@ func TestRemoteSealer(t *testing.T) {
sealhash := ethash.SealHash(header)
// Push new work.
- ethash.Seal(nil, block, nil)
+ results := make(chan *types.Block)
+ ethash.Seal(nil, block, results, nil)
var (
work [3]string
@@ -114,20 +121,11 @@ func TestRemoteSealer(t *testing.T) {
header = &types.Header{Number: big.NewInt(1), Difficulty: big.NewInt(1000)}
block = types.NewBlockWithHeader(header)
sealhash = ethash.SealHash(header)
- ethash.Seal(nil, block, nil)
+ ethash.Seal(nil, block, results, nil)
if work, err = api.GetWork(); err != nil || work[0] != sealhash.Hex() {
t.Error("expect to return the latest pushed work")
}
- // Push block with higher block number.
- newHead := &types.Header{Number: big.NewInt(2), Difficulty: big.NewInt(100)}
- newBlock := types.NewBlockWithHeader(newHead)
- newSealhash := ethash.SealHash(newHead)
- ethash.Seal(nil, newBlock, nil)
-
- if res := api.SubmitWork(types.BlockNonce{}, newSealhash, common.Hash{}); res {
- t.Error("expect to return false when submit a stale solution")
- }
}
func TestHashRate(t *testing.T) {
@@ -136,7 +134,7 @@ func TestHashRate(t *testing.T) {
expect uint64
ids = []common.Hash{common.HexToHash("a"), common.HexToHash("b"), common.HexToHash("c")}
)
- ethash := NewTester(nil)
+ ethash := NewTester(nil, false)
defer ethash.Close()
if tot := ethash.Hashrate(); tot != 0 {
@@ -156,7 +154,7 @@ func TestHashRate(t *testing.T) {
}
func TestClosedRemoteSealer(t *testing.T) {
- ethash := NewTester(nil)
+ ethash := NewTester(nil, false)
time.Sleep(1 * time.Second) // ensure exit channel is listening
ethash.Close()
diff --git a/consensus/ethash/sealer.go b/consensus/ethash/sealer.go
index a458c60f6..06c98a781 100644
--- a/consensus/ethash/sealer.go
+++ b/consensus/ethash/sealer.go
@@ -35,6 +35,11 @@ import (
"github.com/ethereum/go-ethereum/log"
)
+const (
+ // staleThreshold is the maximum depth of the acceptable stale but valid ethash solution.
+ staleThreshold = 7
+)
+
var (
errNoMiningWork = errors.New("no mining work available yet")
errInvalidSealResult = errors.New("invalid or stale proof-of-work solution")
@@ -42,16 +47,21 @@ var (
// Seal implements consensus.Engine, attempting to find a nonce that satisfies
// the block's difficulty requirements.
-func (ethash *Ethash) Seal(chain consensus.ChainReader, block *types.Block, stop <-chan struct{}) (*types.Block, error) {
+func (ethash *Ethash) Seal(chain consensus.ChainReader, block *types.Block, results chan<- *types.Block, stop <-chan struct{}) error {
// If we're running a fake PoW, simply return a 0 nonce immediately
if ethash.config.PowMode == ModeFake || ethash.config.PowMode == ModeFullFake {
header := block.Header()
header.Nonce, header.MixDigest = types.BlockNonce{}, common.Hash{}
- return block.WithSeal(header), nil
+ select {
+ case results <- block.WithSeal(header):
+ default:
+ log.Warn("Sealing result is not read by miner", "mode", "fake", "sealhash", ethash.SealHash(block.Header()))
+ }
+ return nil
}
// If we're running a shared PoW, delegate sealing to it
if ethash.shared != nil {
- return ethash.shared.Seal(chain, block, stop)
+ return ethash.shared.Seal(chain, block, results, stop)
}
// Create a runner and the multiple search threads it directs
abort := make(chan struct{})
@@ -62,7 +72,7 @@ func (ethash *Ethash) Seal(chain consensus.ChainReader, block *types.Block, stop
seed, err := crand.Int(crand.Reader, big.NewInt(math.MaxInt64))
if err != nil {
ethash.lock.Unlock()
- return nil, err
+ return err
}
ethash.rand = rand.New(rand.NewSource(seed.Int64()))
}
@@ -75,34 +85,45 @@ func (ethash *Ethash) Seal(chain consensus.ChainReader, block *types.Block, stop
}
// Push new work to remote sealer
if ethash.workCh != nil {
- ethash.workCh <- block
+ ethash.workCh <- &sealTask{block: block, results: results}
}
- var pend sync.WaitGroup
+ var (
+ pend sync.WaitGroup
+ locals = make(chan *types.Block)
+ )
for i := 0; i < threads; i++ {
pend.Add(1)
go func(id int, nonce uint64) {
defer pend.Done()
- ethash.mine(block, id, nonce, abort, ethash.resultCh)
+ ethash.mine(block, id, nonce, abort, locals)
}(i, uint64(ethash.rand.Int63()))
}
// Wait until sealing is terminated or a nonce is found
- var result *types.Block
- select {
- case <-stop:
- // Outside abort, stop all miner threads
- close(abort)
- case result = <-ethash.resultCh:
- // One of the threads found a block, abort all others
- close(abort)
- case <-ethash.update:
- // Thread count was changed on user request, restart
- close(abort)
+ go func() {
+ var result *types.Block
+ select {
+ case <-stop:
+ // Outside abort, stop all miner threads
+ close(abort)
+ case result = <-locals:
+ // One of the threads found a block, abort all others
+ select {
+ case results <- result:
+ default:
+ log.Warn("Sealing result is not read by miner", "mode", "local", "sealhash", ethash.SealHash(block.Header()))
+ }
+ close(abort)
+ case <-ethash.update:
+ // Thread count was changed on user request, restart
+ close(abort)
+ if err := ethash.Seal(chain, block, results, stop); err != nil {
+ log.Error("Failed to restart sealing after update", "err", err)
+ }
+ }
+ // Wait for all miners to terminate and return the block
pend.Wait()
- return ethash.Seal(chain, block, stop)
- }
- // Wait for all miners to terminate and return the block
- pend.Wait()
- return result, nil
+ }()
+ return nil
}
// mine is the actual proof-of-work miner that searches for a nonce starting from
@@ -165,11 +186,12 @@ search:
}
// remote is a standalone goroutine to handle remote mining related stuff.
-func (ethash *Ethash) remote(notify []string) {
+func (ethash *Ethash) remote(notify []string, noverify bool) {
var (
works = make(map[common.Hash]*types.Block)
rates = make(map[common.Hash]hashrate)
+ results chan<- *types.Block
currentBlock *types.Block
currentWork [3]string
@@ -226,11 +248,15 @@ func (ethash *Ethash) remote(notify []string) {
// submitWork verifies the submitted pow solution, returning
// whether the solution was accepted or not (not can be both a bad pow as well as
// any other error, like no pending work or stale mining result).
- submitWork := func(nonce types.BlockNonce, mixDigest common.Hash, hash common.Hash) bool {
+ submitWork := func(nonce types.BlockNonce, mixDigest common.Hash, sealhash common.Hash) bool {
+ if currentBlock == nil {
+ log.Error("Pending work without block", "sealhash", sealhash)
+ return false
+ }
// Make sure the work submitted is present
- block := works[hash]
+ block := works[sealhash]
if block == nil {
- log.Info("Work submitted but none pending", "hash", hash)
+ log.Warn("Work submitted but none pending", "sealhash", sealhash, "curnumber", currentBlock.NumberU64())
return false
}
// Verify the correctness of submitted result.
@@ -239,26 +265,36 @@ func (ethash *Ethash) remote(notify []string) {
header.MixDigest = mixDigest
start := time.Now()
- if err := ethash.verifySeal(nil, header, true); err != nil {
- log.Warn("Invalid proof-of-work submitted", "hash", hash, "elapsed", time.Since(start), "err", err)
- return false
+ if !noverify {
+ if err := ethash.verifySeal(nil, header, true); err != nil {
+ log.Warn("Invalid proof-of-work submitted", "sealhash", sealhash, "elapsed", time.Since(start), "err", err)
+ return false
+ }
}
- // Make sure the result channel is created.
- if ethash.resultCh == nil {
+ // Make sure the result channel is assigned.
+ if results == nil {
log.Warn("Ethash result channel is empty, submitted mining result is rejected")
return false
}
- log.Trace("Verified correct proof-of-work", "hash", hash, "elapsed", time.Since(start))
+ log.Trace("Verified correct proof-of-work", "sealhash", sealhash, "elapsed", time.Since(start))
// Solutions seems to be valid, return to the miner and notify acceptance.
- select {
- case ethash.resultCh <- block.WithSeal(header):
- delete(works, hash)
- return true
- default:
- log.Info("Work submitted is stale", "hash", hash)
- return false
+ solution := block.WithSeal(header)
+
+ // The submitted solution is within the scope of acceptance.
+ if solution.NumberU64()+staleThreshold > currentBlock.NumberU64() {
+ select {
+ case results <- solution:
+ log.Debug("Work submitted is acceptable", "number", solution.NumberU64(), "sealhash", sealhash, "hash", solution.Hash())
+ return true
+ default:
+ log.Warn("Sealing result is not read by miner", "mode", "remote", "sealhash", sealhash)
+ return false
+ }
}
+ // The submitted block is too old to accept, drop it.
+ log.Warn("Work submitted is too old", "number", solution.NumberU64(), "sealhash", sealhash, "hash", solution.Hash())
+ return false
}
ticker := time.NewTicker(5 * time.Second)
@@ -266,14 +302,12 @@ func (ethash *Ethash) remote(notify []string) {
for {
select {
- case block := <-ethash.workCh:
- if currentBlock != nil && block.ParentHash() != currentBlock.ParentHash() {
- // Start new round mining, throw out all previous work.
- works = make(map[common.Hash]*types.Block)
- }
+ case work := <-ethash.workCh:
// Update current work with new received block.
// Note same work can be past twice, happens when changing CPU threads.
- makeWork(block)
+ results = work.results
+
+ makeWork(work.block)
// Notify and requested URLs of the new work availability
notifyWork()
@@ -315,6 +349,14 @@ func (ethash *Ethash) remote(notify []string) {
delete(rates, id)
}
}
+ // Clear stale pending blocks
+ if currentBlock != nil {
+ for hash, block := range works {
+ if block.NumberU64()+staleThreshold <= currentBlock.NumberU64() {
+ delete(works, hash)
+ }
+ }
+ }
case errc := <-ethash.exitCh:
// Exit remote loop if ethash is closed and return relevant error.
diff --git a/consensus/ethash/sealer_test.go b/consensus/ethash/sealer_test.go
index d1b66f9cf..31d18b67c 100644
--- a/consensus/ethash/sealer_test.go
+++ b/consensus/ethash/sealer_test.go
@@ -41,14 +41,14 @@ func TestRemoteNotify(t *testing.T) {
go server.Serve(listener)
// Create the custom ethash engine
- ethash := NewTester([]string{"http://" + listener.Addr().String()})
+ ethash := NewTester([]string{"http://" + listener.Addr().String()}, false)
defer ethash.Close()
// Stream a work task and ensure the notification bubbles out
header := &types.Header{Number: big.NewInt(1), Difficulty: big.NewInt(100)}
block := types.NewBlockWithHeader(header)
- ethash.Seal(nil, block, nil)
+ ethash.Seal(nil, block, nil, nil)
select {
case work := <-sink:
if want := ethash.SealHash(header).Hex(); work[0] != want {
@@ -66,7 +66,7 @@ func TestRemoteNotify(t *testing.T) {
}
}
-// Tests that pushing work packages fast to the miner doesn't cause any daa race
+// Tests that pushing work packages fast to the miner doesn't cause any data race
// issues in the notifications.
func TestRemoteMultiNotify(t *testing.T) {
// Start a simple webserver to capture notifications
@@ -95,7 +95,7 @@ func TestRemoteMultiNotify(t *testing.T) {
go server.Serve(listener)
// Create the custom ethash engine
- ethash := NewTester([]string{"http://" + listener.Addr().String()})
+ ethash := NewTester([]string{"http://" + listener.Addr().String()}, false)
defer ethash.Close()
// Stream a lot of work task and ensure all the notifications bubble out
@@ -103,7 +103,7 @@ func TestRemoteMultiNotify(t *testing.T) {
header := &types.Header{Number: big.NewInt(int64(i)), Difficulty: big.NewInt(100)}
block := types.NewBlockWithHeader(header)
- ethash.Seal(nil, block, nil)
+ ethash.Seal(nil, block, nil, nil)
}
for i := 0; i < cap(sink); i++ {
select {
@@ -113,3 +113,87 @@ func TestRemoteMultiNotify(t *testing.T) {
}
}
}
+
+// Tests whether stale solutions are correctly processed.
+func TestStaleSubmission(t *testing.T) {
+ ethash := NewTester(nil, true)
+ defer ethash.Close()
+ api := &API{ethash}
+
+ fakeNonce, fakeDigest := types.BlockNonce{0x01, 0x02, 0x03}, common.HexToHash("deadbeef")
+
+ testcases := []struct {
+ headers []*types.Header
+ submitIndex int
+ submitRes bool
+ }{
+ // Case1: submit solution for the latest mining package
+ {
+ []*types.Header{
+ {ParentHash: common.BytesToHash([]byte{0xa}), Number: big.NewInt(1), Difficulty: big.NewInt(100000000)},
+ },
+ 0,
+ true,
+ },
+ // Case2: submit solution for the previous package but have same parent.
+ {
+ []*types.Header{
+ {ParentHash: common.BytesToHash([]byte{0xb}), Number: big.NewInt(2), Difficulty: big.NewInt(100000000)},
+ {ParentHash: common.BytesToHash([]byte{0xb}), Number: big.NewInt(2), Difficulty: big.NewInt(100000001)},
+ },
+ 0,
+ true,
+ },
+ // Case3: submit stale but acceptable solution
+ {
+ []*types.Header{
+ {ParentHash: common.BytesToHash([]byte{0xc}), Number: big.NewInt(3), Difficulty: big.NewInt(100000000)},
+ {ParentHash: common.BytesToHash([]byte{0xd}), Number: big.NewInt(9), Difficulty: big.NewInt(100000000)},
+ },
+ 0,
+ true,
+ },
+ // Case4: submit very old solution
+ {
+ []*types.Header{
+ {ParentHash: common.BytesToHash([]byte{0xe}), Number: big.NewInt(10), Difficulty: big.NewInt(100000000)},
+ {ParentHash: common.BytesToHash([]byte{0xf}), Number: big.NewInt(17), Difficulty: big.NewInt(100000000)},
+ },
+ 0,
+ false,
+ },
+ }
+ results := make(chan *types.Block, 16)
+
+ for id, c := range testcases {
+ for _, h := range c.headers {
+ ethash.Seal(nil, types.NewBlockWithHeader(h), results, nil)
+ }
+ if res := api.SubmitWork(fakeNonce, ethash.SealHash(c.headers[c.submitIndex]), fakeDigest); res != c.submitRes {
+ t.Errorf("case %d submit result mismatch, want %t, get %t", id+1, c.submitRes, res)
+ }
+ if !c.submitRes {
+ continue
+ }
+ select {
+ case res := <-results:
+ if res.Header().Nonce != fakeNonce {
+ t.Errorf("case %d block nonce mismatch, want %s, get %s", id+1, fakeNonce, res.Header().Nonce)
+ }
+ if res.Header().MixDigest != fakeDigest {
+ t.Errorf("case %d block digest mismatch, want %s, get %s", id+1, fakeDigest, res.Header().MixDigest)
+ }
+ if res.Header().Difficulty.Uint64() != c.headers[c.submitIndex].Difficulty.Uint64() {
+ t.Errorf("case %d block difficulty mismatch, want %d, get %d", id+1, c.headers[c.submitIndex].Difficulty, res.Header().Difficulty)
+ }
+ if res.Header().Number.Uint64() != c.headers[c.submitIndex].Number.Uint64() {
+ t.Errorf("case %d block number mismatch, want %d, get %d", id+1, c.headers[c.submitIndex].Number.Uint64(), res.Header().Number.Uint64())
+ }
+ if res.Header().ParentHash != c.headers[c.submitIndex].ParentHash {
+ t.Errorf("case %d block parent hash mismatch, want %s, get %s", id+1, c.headers[c.submitIndex].ParentHash.Hex(), res.Header().ParentHash.Hex())
+ }
+ case <-time.NewTimer(time.Second).C:
+ t.Errorf("case %d fetch ethash result timeout", id+1)
+ }
+ }
+}
diff --git a/eth/api_tracer.go b/eth/api_tracer.go
index 704a6cdba..0a8b9a994 100644
--- a/eth/api_tracer.go
+++ b/eth/api_tracer.go
@@ -294,7 +294,7 @@ func (api *PrivateDebugAPI) traceChain(ctx context.Context, start, end *types.Bl
failed = err
break
}
- // Reference the trie twice, once for us, once for the trancer
+ // Reference the trie twice, once for us, once for the tracer
database.TrieDB().Reference(root, common.Hash{})
if number >= origin {
database.TrieDB().Reference(root, common.Hash{})
diff --git a/eth/backend.go b/eth/backend.go
index da7e0b2cd..3032e1a6d 100644
--- a/eth/backend.go
+++ b/eth/backend.go
@@ -130,7 +130,7 @@ func New(ctx *node.ServiceContext, config *Config) (*Ethereum, error) {
chainConfig: chainConfig,
eventMux: ctx.EventMux,
accountManager: ctx.AccountManager,
- engine: CreateConsensusEngine(ctx, chainConfig, &config.Ethash, config.MinerNotify, chainDb),
+ engine: CreateConsensusEngine(ctx, chainConfig, &config.Ethash, config.MinerNotify, config.MinerNoverify, chainDb),
shutdownChan: make(chan bool),
networkID: config.NetworkId,
gasPrice: config.MinerGasPrice,
@@ -216,7 +216,7 @@ func CreateDB(ctx *node.ServiceContext, config *Config, name string) (ethdb.Data
}
// CreateConsensusEngine creates the required type of consensus engine instance for an Ethereum service
-func CreateConsensusEngine(ctx *node.ServiceContext, chainConfig *params.ChainConfig, config *ethash.Config, notify []string, db ethdb.Database) consensus.Engine {
+func CreateConsensusEngine(ctx *node.ServiceContext, chainConfig *params.ChainConfig, config *ethash.Config, notify []string, noverify bool, db ethdb.Database) consensus.Engine {
// If proof-of-authority is requested, set it up
if chainConfig.Clique != nil {
return clique.New(chainConfig.Clique, db)
@@ -228,7 +228,7 @@ func CreateConsensusEngine(ctx *node.ServiceContext, chainConfig *params.ChainCo
return ethash.NewFaker()
case ethash.ModeTest:
log.Warn("Ethash used in test mode")
- return ethash.NewTester(nil)
+ return ethash.NewTester(nil, noverify)
case ethash.ModeShared:
log.Warn("Ethash used in shared mode")
return ethash.NewShared()
@@ -240,7 +240,7 @@ func CreateConsensusEngine(ctx *node.ServiceContext, chainConfig *params.ChainCo
DatasetDir: config.DatasetDir,
DatasetsInMem: config.DatasetsInMem,
DatasetsOnDisk: config.DatasetsOnDisk,
- }, notify)
+ }, notify, noverify)
engine.SetThreads(-1) // Disable CPU mining
return engine
}
diff --git a/eth/config.go b/eth/config.go
index 5b2eca585..517d7d8f3 100644
--- a/eth/config.go
+++ b/eth/config.go
@@ -101,6 +101,7 @@ type Config struct {
MinerExtraData []byte `toml:",omitempty"`
MinerGasPrice *big.Int
MinerRecommit time.Duration
+ MinerNoverify bool
// Ethash options
Ethash ethash.Config
diff --git a/eth/gen_config.go b/eth/gen_config.go
index a72e35bcd..df4ffeb11 100644
--- a/eth/gen_config.go
+++ b/eth/gen_config.go
@@ -35,6 +35,7 @@ func (c Config) MarshalTOML() (interface{}, error) {
MinerExtraData hexutil.Bytes `toml:",omitempty"`
MinerGasPrice *big.Int
MinerRecommit time.Duration
+ MinerNoverify bool
Ethash ethash.Config
TxPool core.TxPoolConfig
GPO gasprice.Config
@@ -58,6 +59,7 @@ func (c Config) MarshalTOML() (interface{}, error) {
enc.MinerExtraData = c.MinerExtraData
enc.MinerGasPrice = c.MinerGasPrice
enc.MinerRecommit = c.MinerRecommit
+ enc.MinerNoverify = c.MinerNoverify
enc.Ethash = c.Ethash
enc.TxPool = c.TxPool
enc.GPO = c.GPO
@@ -81,11 +83,11 @@ func (c *Config) UnmarshalTOML(unmarshal func(interface{}) error) error {
TrieCache *int
TrieTimeout *time.Duration
Etherbase *common.Address `toml:",omitempty"`
- MinerThreads *int `toml:",omitempty"`
MinerNotify []string `toml:",omitempty"`
MinerExtraData *hexutil.Bytes `toml:",omitempty"`
MinerGasPrice *big.Int
MinerRecommit *time.Duration
+ MinerNoverify *bool
Ethash *ethash.Config
TxPool *core.TxPoolConfig
GPO *gasprice.Config
@@ -144,6 +146,9 @@ func (c *Config) UnmarshalTOML(unmarshal func(interface{}) error) error {
if dec.MinerRecommit != nil {
c.MinerRecommit = *dec.MinerRecommit
}
+ if dec.MinerNoverify != nil {
+ c.MinerNoverify = *dec.MinerNoverify
+ }
if dec.Ethash != nil {
c.Ethash = *dec.Ethash
}
diff --git a/les/backend.go b/les/backend.go
index 75049da08..a3474a683 100644
--- a/les/backend.go
+++ b/les/backend.go
@@ -102,7 +102,7 @@ func New(ctx *node.ServiceContext, config *eth.Config) (*LightEthereum, error) {
peers: peers,
reqDist: newRequestDistributor(peers, quitSync),
accountManager: ctx.AccountManager,
- engine: eth.CreateConsensusEngine(ctx, chainConfig, &config.Ethash, nil, chainDb),
+ engine: eth.CreateConsensusEngine(ctx, chainConfig, &config.Ethash, nil, false, chainDb),
shutdownChan: make(chan bool),
networkId: config.NetworkId,
bloomRequests: make(chan chan *bloombits.Retrieval),
diff --git a/miner/worker.go b/miner/worker.go
index 1a881799d..ca68da6e9 100644
--- a/miner/worker.go
+++ b/miner/worker.go
@@ -73,7 +73,7 @@ const (
// increasing upper limit or decreasing lower limit so that the limit can be reachable.
intervalAdjustBias = 200 * 1000.0 * 1000.0
- // staleThreshold is the maximum distance of the acceptable stale block.
+ // staleThreshold is the maximum depth of the acceptable stale block.
staleThreshold = 7
)
@@ -139,7 +139,7 @@ type worker struct {
// Channels
newWorkCh chan *newWorkReq
taskCh chan *task
- resultCh chan *task
+ resultCh chan *types.Block
startCh chan struct{}
exitCh chan struct{}
resubmitIntervalCh chan time.Duration
@@ -186,7 +186,7 @@ func newWorker(config *params.ChainConfig, engine consensus.Engine, eth Backend,
chainSideCh: make(chan core.ChainSideEvent, chainSideChanSize),
newWorkCh: make(chan *newWorkReq),
taskCh: make(chan *task),
- resultCh: make(chan *task, resultQueueSize),
+ resultCh: make(chan *types.Block, resultQueueSize),
exitCh: make(chan struct{}),
startCh: make(chan struct{}, 1),
resubmitIntervalCh: make(chan time.Duration),
@@ -269,18 +269,10 @@ func (w *worker) isRunning() bool {
return atomic.LoadInt32(&w.running) == 1
}
-// close terminates all background threads maintained by the worker and cleans up buffered channels.
+// close terminates all background threads maintained by the worker.
// Note the worker does not support being closed multiple times.
func (w *worker) close() {
close(w.exitCh)
- // Clean up buffered channels
- for empty := false; !empty; {
- select {
- case <-w.resultCh:
- default:
- empty = true
- }
- }
}
// newWorkLoop is a standalone goroutine to submit new mining work upon received events.
@@ -471,42 +463,6 @@ func (w *worker) mainLoop() {
}
}
-// seal pushes a sealing task to consensus engine and submits the result.
-func (w *worker) seal(t *task, stop <-chan struct{}) {
- if w.skipSealHook != nil && w.skipSealHook(t) {
- return
- }
- // The reason for caching task first is:
- // A previous sealing action will be canceled by subsequent actions,
- // however, remote miner may submit a result based on the cancelled task.
- // So we should only submit the pending state corresponding to the seal result.
- // TODO(rjl493456442) Replace the seal-wait logic structure
- w.pendingMu.Lock()
- w.pendingTasks[w.engine.SealHash(t.block.Header())] = t
- w.pendingMu.Unlock()
-
- if block, err := w.engine.Seal(w.chain, t.block, stop); block != nil {
- sealhash := w.engine.SealHash(block.Header())
- w.pendingMu.RLock()
- task, exist := w.pendingTasks[sealhash]
- w.pendingMu.RUnlock()
- if !exist {
- log.Error("Block found but no relative pending task", "number", block.Number(), "sealhash", sealhash, "hash", block.Hash())
- return
- }
- // Assemble sealing result
- task.block = block
- log.Info("Successfully sealed new block", "number", block.Number(), "sealhash", sealhash, "hash", block.Hash(),
- "elapsed", common.PrettyDuration(time.Since(task.createdAt)))
- select {
- case w.resultCh <- task:
- case <-w.exitCh:
- }
- } else if err != nil {
- log.Warn("Block sealing failed", "err", err)
- }
-}
-
// taskLoop is a standalone goroutine to fetch sealing task from the generator and
// push them to consensus engine.
func (w *worker) taskLoop() {
@@ -533,10 +489,20 @@ func (w *worker) taskLoop() {
if sealHash == prev {
continue
}
+ // Interrupt previous sealing operation
interrupt()
- stopCh = make(chan struct{})
- prev = sealHash
- go w.seal(task, stopCh)
+ stopCh, prev = make(chan struct{}), sealHash
+
+ if w.skipSealHook != nil && w.skipSealHook(task) {
+ continue
+ }
+ w.pendingMu.Lock()
+ w.pendingTasks[w.engine.SealHash(task.block.Header())] = task
+ w.pendingMu.Unlock()
+
+ if err := w.engine.Seal(w.chain, task.block, w.resultCh, stopCh); err != nil {
+ log.Warn("Block sealing failed", "err", err)
+ }
case <-w.exitCh:
interrupt()
return
@@ -549,38 +515,54 @@ func (w *worker) taskLoop() {
func (w *worker) resultLoop() {
for {
select {
- case result := <-w.resultCh:
+ case block := <-w.resultCh:
// Short circuit when receiving empty result.
- if result == nil {
+ if block == nil {
continue
}
// Short circuit when receiving duplicate result caused by resubmitting.
- block := result.block
if w.chain.HasBlock(block.Hash(), block.NumberU64()) {
continue
}
- // Update the block hash in all logs since it is now available and not when the
- // receipt/log of individual transactions were created.
- for _, r := range result.receipts {
- for _, l := range r.Logs {
- l.BlockHash = block.Hash()
- }
+ var (
+ sealhash = w.engine.SealHash(block.Header())
+ hash = block.Hash()
+ )
+ w.pendingMu.RLock()
+ task, exist := w.pendingTasks[sealhash]
+ w.pendingMu.RUnlock()
+ if !exist {
+ log.Error("Block found but no relative pending task", "number", block.Number(), "sealhash", sealhash, "hash", hash)
+ continue
}
- for _, log := range result.state.Logs() {
- log.BlockHash = block.Hash()
+ // Different block could share same sealhash, deep copy here to prevent write-write conflict.
+ var (
+ receipts = make([]*types.Receipt, len(task.receipts))
+ logs []*types.Log
+ )
+ for i, receipt := range task.receipts {
+ receipts[i] = new(types.Receipt)
+ *receipts[i] = *receipt
+ // Update the block hash in all logs since it is now available and not when the
+ // receipt/log of individual transactions were created.
+ for _, log := range receipt.Logs {
+ log.BlockHash = hash
+ }
+ logs = append(logs, receipt.Logs...)
}
// Commit block and state to database.
- stat, err := w.chain.WriteBlockWithState(block, result.receipts, result.state)
+ stat, err := w.chain.WriteBlockWithState(block, receipts, task.state)
if err != nil {
log.Error("Failed writing block to chain", "err", err)
continue
}
+ log.Info("Successfully sealed new block", "number", block.Number(), "sealhash", sealhash, "hash", hash,
+ "elapsed", common.PrettyDuration(time.Since(task.createdAt)))
+
// Broadcast the block and announce chain insertion event
w.mux.Post(core.NewMinedBlockEvent{Block: block})
- var (
- events []interface{}
- logs = result.state.Logs()
- )
+
+ var events []interface{}
switch stat {
case core.CanonStatTy:
events = append(events, core.ChainEvent{Block: block, Hash: block.Hash(), Logs: logs})