core,les: headerchain import in batches (ethereum#21471)

* core: add test for headerchain inserts

* core, light: write headerchains in batches

* core: change to one callback per batch of inserted headers + review concerns

* core: error-check on batch write

* core: unexport writeHeaders

* core: remove callback parameter in InsertHeaderChain

The semantics of InsertHeaderChain are now much simpler: it is now an
all-or-nothing operation. The new WriteStatus return value allows
callers to check for the canonicality of the insertion. This change
simplifies use of HeaderChain in package les, where the callback was
previously used to post chain events.

* core: skip some hashing when writing headers

* core: less hashing in header validation

* core: fix headerchain flaw regarding blacklisted hashes

Co-authored-by: Felix Lange <fjl@twurst.com>

Author: holiman

core/blockchain.go

@@ -2438,12 +2438,8 @@ func (bc *BlockChain) InsertHeaderChain(chain []*types.Header, checkFreq int) (i
 
 	bc.wg.Add(1)
 	defer bc.wg.Done()
-
-	whFunc := func(header *types.Header) error {
-		_, err := bc.hc.WriteHeader(header)
-		return err
-	}
-	return bc.hc.InsertHeaderChain(chain, whFunc, start)
+	_, err := bc.hc.InsertHeaderChain(chain, start)
+	return 0, err
 }
 
 // CurrentHeader retrieves the current head header of the canonical chain. The

core/headerchain.go

@@ -129,118 +129,192 @@ func (hc *HeaderChain) GetBlockNumber(hash common.Hash) *uint64 {
 	return number
 }
 
-// WriteHeader writes a header into the local chain, given that its parent is
-// already known. If the total difficulty of the newly inserted header becomes
-// greater than the current known TD, the canonical chain is re-routed.
+type headerWriteResult struct {
+	status     WriteStatus
+	ignored    int
+	imported   int
+	lastHash   common.Hash
+	lastHeader *types.Header
+}
+
+// WriteHeaders writes a chain of headers into the local chain, given that the parents
+// are already known. If the total difficulty of the newly inserted chain becomes
+// greater than the current known TD, the canonical chain is reorged.
 //
 // Note: This method is not concurrent-safe with inserting blocks simultaneously
 // into the chain, as side effects caused by reorganisations cannot be emulated
 // without the real blocks. Hence, writing headers directly should only be done
 // in two scenarios: pure-header mode of operation (light clients), or properly
 // separated header/block phases (non-archive clients).
-func (hc *HeaderChain) WriteHeader(header *types.Header) (status WriteStatus, err error) {
-	// Cache some values to prevent constant recalculation
+func (hc *HeaderChain) writeHeaders(headers []*types.Header) (result *headerWriteResult, err error) {
+	if len(headers) == 0 {
+		return &headerWriteResult{}, nil
+	}
+	ptd := hc.GetTd(headers[0].ParentHash, headers[0].Number.Uint64()-1)
+	if ptd == nil {
+		return &headerWriteResult{}, consensus.ErrUnknownAncestor
+	}
 	var (
-		hash   = header.Hash()
-		number = header.Number.Uint64()
+		lastNumber = headers[0].Number.Uint64() - 1 // Last successfully imported number
+		lastHash   = headers[0].ParentHash          // Last imported header hash
+		newTD      = new(big.Int).Set(ptd)          // Total difficulty of inserted chain
+
+		lastHeader    *types.Header
+		inserted      []numberHash // Ephemeral lookup of number/hash for the chain
+		firstInserted = -1         // Index of the first non-ignored header
 	)
-	// Calculate the total difficulty of the header
-	ptd := hc.GetTd(header.ParentHash, number-1)
-	if ptd == nil {
-		return NonStatTy, consensus.ErrUnknownAncestor
-	}
-	head := hc.CurrentHeader().Number.Uint64()
-	localTd := hc.GetTd(hc.currentHeaderHash, head)
-	externTd := new(big.Int).Add(header.Difficulty, ptd)
-
-	// Irrelevant of the canonical status, write the td and header to the database
-	//
-	// Note all the components of header(td, hash->number index and header) should
-	// be written atomically.
-	headerBatch := hc.chainDb.NewBatch()
-	rawdb.WriteTd(headerBatch, hash, number, externTd)
-	rawdb.WriteHeader(headerBatch, header)
-	if err := headerBatch.Write(); err != nil {
-		log.Crit("Failed to write header into disk", "err", err)
+
+	batch := hc.chainDb.NewBatch()
+	for i, header := range headers {
+		var hash common.Hash
+		// The headers have already been validated at this point, so we already
+		// know that it's a contiguous chain, where
+		// headers[i].Hash() == headers[i+1].ParentHash
+		if i < len(headers)-1 {
+			hash = headers[i+1].ParentHash
+		} else {
+			hash = header.Hash()
+		}
+		number := header.Number.Uint64()
+		newTD.Add(newTD, header.Difficulty)
+
+		// If the header is already known, skip it, otherwise store
+		if !hc.HasHeader(hash, number) {
+			// Irrelevant of the canonical status, write the TD and header to the database.
+			rawdb.WriteTd(batch, hash, number, newTD)
+			hc.tdCache.Add(hash, new(big.Int).Set(newTD))
+
+			rawdb.WriteHeader(batch, header)
+			inserted = append(inserted, numberHash{number, hash})
+			hc.headerCache.Add(hash, header)
+			hc.numberCache.Add(hash, number)
+			if firstInserted < 0 {
+				firstInserted = i
+			}
+		}
+		lastHeader, lastHash, lastNumber = header, hash, number
+	}
+
+	// Skip the slow disk write of all headers if interrupted.
+	if hc.procInterrupt() {
+		log.Debug("Premature abort during headers import")
+		return &headerWriteResult{}, errors.New("aborted")
 	}
+	// Commit to disk!
+	if err := batch.Write(); err != nil {
+		log.Crit("Failed to write headers", "error", err)
+	}
+	batch.Reset()
+
+	var (
+		head    = hc.CurrentHeader().Number.Uint64()
+		localTD = hc.GetTd(hc.currentHeaderHash, head)
+		status  = SideStatTy
+	)
 	// If the total difficulty is higher than our known, add it to the canonical chain
 	// Second clause in the if statement reduces the vulnerability to selfish mining.
 	// Please refer to http://www.cs.cornell.edu/~ie53/publications/btcProcFC.pdf
-	reorg := externTd.Cmp(localTd) > 0
-	if !reorg && externTd.Cmp(localTd) == 0 {
-		if header.Number.Uint64() < head {
+	reorg := newTD.Cmp(localTD) > 0
+	if !reorg && newTD.Cmp(localTD) == 0 {
+		if lastNumber < head {
 			reorg = true
-		} else if header.Number.Uint64() == head {
+		} else if lastNumber == head {
 			reorg = mrand.Float64() < 0.5
 		}
 	}
+	// If the parent of the (first) block is already the canon header,
+	// we don't have to go backwards to delete canon blocks, but
+	// simply pile them onto the existing chain
+	chainAlreadyCanon := headers[0].ParentHash == hc.currentHeaderHash
 	if reorg {
 		// If the header can be added into canonical chain, adjust the
 		// header chain markers(canonical indexes and head header flag).
 		//
 		// Note all markers should be written atomically.
-
-		// Delete any canonical number assignments above the new head
-		markerBatch := hc.chainDb.NewBatch()
-		for i := number + 1; ; i++ {
-			hash := rawdb.ReadCanonicalHash(hc.chainDb, i)
-			if hash == (common.Hash{}) {
-				break
+		markerBatch := batch // we can reuse the batch to keep allocs down
+		if !chainAlreadyCanon {
+			// Delete any canonical number assignments above the new head
+			for i := lastNumber + 1; ; i++ {
+				hash := rawdb.ReadCanonicalHash(hc.chainDb, i)
+				if hash == (common.Hash{}) {
+					break
+				}
+				rawdb.DeleteCanonicalHash(markerBatch, i)
+			}
+			// Overwrite any stale canonical number assignments, going
+			// backwards from the first header in this import
+			var (
+				headHash   = headers[0].ParentHash          // inserted[0].parent?
+				headNumber = headers[0].Number.Uint64() - 1 // inserted[0].num-1 ?
+				headHeader = hc.GetHeader(headHash, headNumber)
+			)
+			for rawdb.ReadCanonicalHash(hc.chainDb, headNumber) != headHash {
+				rawdb.WriteCanonicalHash(markerBatch, headHash, headNumber)
+				headHash = headHeader.ParentHash
+				headNumber = headHeader.Number.Uint64() - 1
+				headHeader = hc.GetHeader(headHash, headNumber)
+			}
+			// If some of the older headers were already known, but obtained canon-status
+			// during this import batch, then we need to write that now
+			// Further down, we continue writing the staus for the ones that
+			// were not already known
+			for i := 0; i < firstInserted; i++ {
+				hash := headers[i].Hash()
+				num := headers[i].Number.Uint64()
+				rawdb.WriteCanonicalHash(markerBatch, hash, num)
+				rawdb.WriteHeadHeaderHash(markerBatch, hash)
 			}
-			rawdb.DeleteCanonicalHash(markerBatch, i)
 		}
-
-		// Overwrite any stale canonical number assignments
-		var (
-			headHash   = header.ParentHash
-			headNumber = header.Number.Uint64() - 1
-			headHeader = hc.GetHeader(headHash, headNumber)
-		)
-		for rawdb.ReadCanonicalHash(hc.chainDb, headNumber) != headHash {
-			rawdb.WriteCanonicalHash(markerBatch, headHash, headNumber)
-
-			headHash = headHeader.ParentHash
-			headNumber = headHeader.Number.Uint64() - 1
-			headHeader = hc.GetHeader(headHash, headNumber)
+		// Extend the canonical chain with the new headers
+		for _, hn := range inserted {
+			rawdb.WriteCanonicalHash(markerBatch, hn.hash, hn.number)
+			rawdb.WriteHeadHeaderHash(markerBatch, hn.hash)
 		}
-		// Extend the canonical chain with the new header
-		rawdb.WriteCanonicalHash(markerBatch, hash, number)
-		rawdb.WriteHeadHeaderHash(markerBatch, hash)
 		if err := markerBatch.Write(); err != nil {
 			log.Crit("Failed to write header markers into disk", "err", err)
 		}
+		markerBatch.Reset()
 		// Last step update all in-memory head header markers
-		hc.currentHeaderHash = hash
-		hc.currentHeader.Store(types.CopyHeader(header))
-		headHeaderGauge.Update(header.Number.Int64())
+		hc.currentHeaderHash = lastHash
+		hc.currentHeader.Store(types.CopyHeader(lastHeader))
+		headHeaderGauge.Update(lastHeader.Number.Int64())
 
+		// Chain status is canonical since this insert was a reorg.
+		// Note that all inserts which have higher TD than existing are 'reorg'.
 		status = CanonStatTy
-	} else {
-		status = SideStatTy
 	}
-	hc.tdCache.Add(hash, externTd)
-	hc.headerCache.Add(hash, header)
-	hc.numberCache.Add(hash, number)
-	return
-}
 
-// WhCallback is a callback function for inserting individual headers.
-// A callback is used for two reasons: first, in a LightChain, status should be
-// processed and light chain events sent, while in a BlockChain this is not
-// necessary since chain events are sent after inserting blocks. Second, the
-// header writes should be protected by the parent chain mutex individually.
-type WhCallback func(*types.Header) error
+	if len(inserted) == 0 {
+		status = NonStatTy
+	}
+	return &headerWriteResult{
+		status:     status,
+		ignored:    len(headers) - len(inserted),
+		imported:   len(inserted),
+		lastHash:   lastHash,
+		lastHeader: lastHeader,
+	}, nil
+}
 
 func (hc *HeaderChain) ValidateHeaderChain(chain []*types.Header, checkFreq int) (int, error) {
 	// Do a sanity check that the provided chain is actually ordered and linked
 	for i := 1; i < len(chain); i++ {
-		if chain[i].Number.Uint64() != chain[i-1].Number.Uint64()+1 || chain[i].ParentHash != chain[i-1].Hash() {
+		parentHash := chain[i-1].Hash()
+		if chain[i].Number.Uint64() != chain[i-1].Number.Uint64()+1 || chain[i].ParentHash != parentHash {
 			// Chain broke ancestry, log a message (programming error) and skip insertion
 			log.Error("Non contiguous header insert", "number", chain[i].Number, "hash", chain[i].Hash(),
-				"parent", chain[i].ParentHash, "prevnumber", chain[i-1].Number, "prevhash", chain[i-1].Hash())
+				"parent", chain[i].ParentHash, "prevnumber", chain[i-1].Number, "prevhash", parentHash)
 
 			return 0, fmt.Errorf("non contiguous insert: item %d is #%d [%x…], item %d is #%d [%x…] (parent [%x…])", i-1, chain[i-1].Number,
-				chain[i-1].Hash().Bytes()[:4], i, chain[i].Number, chain[i].Hash().Bytes()[:4], chain[i].ParentHash[:4])
+				parentHash.Bytes()[:4], i, chain[i].Number, chain[i].Hash().Bytes()[:4], chain[i].ParentHash[:4])
+		}
+		// If the header is a banned one, straight out abort
+		if BadHashes[parentHash] {
+			return i - 1, ErrBlacklistedHash
+		}
+		// If it's the last header in the cunk, we need to check it too
+		if i == len(chain)-1 && BadHashes[chain[i].Hash()] {
+			return i, ErrBlacklistedHash
 		}
 	}
 
@@ -263,16 +337,12 @@ func (hc *HeaderChain) ValidateHeaderChain(chain []*types.Header, checkFreq int)
 	defer close(abort)
 
 	// Iterate over the headers and ensure they all check out
-	for i, header := range chain {
+	for i := range chain {
 		// If the chain is terminating, stop processing blocks
 		if hc.procInterrupt() {
 			log.Debug("Premature abort during headers verification")
 			return 0, errors.New("aborted")
 		}
-		// If the header is a banned one, straight out abort
-		if BadHashes[header.Hash()] {
-			return i, ErrBlacklistedHash
-		}
 		// Otherwise wait for headers checks and ensure they pass
 		if err := <-results; err != nil {
 			return i, err
@@ -282,55 +352,41 @@ func (hc *HeaderChain) ValidateHeaderChain(chain []*types.Header, checkFreq int)
 	return 0, nil
 }
 
-// InsertHeaderChain attempts to insert the given header chain in to the local
-// chain, possibly creating a reorg. If an error is returned, it will return the
-// index number of the failing header as well an error describing what went wrong.
+// InsertHeaderChain inserts the given headers.
 //
-// The verify parameter can be used to fine tune whether nonce verification
-// should be done or not. The reason behind the optional check is because some
-// of the header retrieval mechanisms already need to verfy nonces, as well as
-// because nonces can be verified sparsely, not needing to check each.
-func (hc *HeaderChain) InsertHeaderChain(chain []*types.Header, writeHeader WhCallback, start time.Time) (int, error) {
-	// Collect some import statistics to report on
-	stats := struct{ processed, ignored int }{}
-	// All headers passed verification, import them into the database
-	for i, header := range chain {
-		// Short circuit insertion if shutting down
-		if hc.procInterrupt() {
-			log.Debug("Premature abort during headers import")
-			return i, errors.New("aborted")
-		}
-		// If the header's already known, skip it, otherwise store
-		hash := header.Hash()
-		if hc.HasHeader(hash, header.Number.Uint64()) {
-			externTd := hc.GetTd(hash, header.Number.Uint64())
-			localTd := hc.GetTd(hc.currentHeaderHash, hc.CurrentHeader().Number.Uint64())
-			if externTd == nil || externTd.Cmp(localTd) <= 0 {
-				stats.ignored++
-				continue
-			}
-		}
-		if err := writeHeader(header); err != nil {
-			return i, err
-		}
-		stats.processed++
+// The validity of the headers is NOT CHECKED by this method, i.e. they need to be
+// validated by ValidateHeaderChain before calling InsertHeaderChain.
+//
+// This insert is all-or-nothing. If this returns an error, no headers were written,
+// otherwise they were all processed successfully.
+//
+// The returned 'write status' says if the inserted headers are part of the canonical chain
+// or a side chain.
+func (hc *HeaderChain) InsertHeaderChain(chain []*types.Header, start time.Time) (WriteStatus, error) {
+	if hc.procInterrupt() {
+		return 0, errors.New("aborted")
 	}
-	// Report some public statistics so the user has a clue what's going on
-	last := chain[len(chain)-1]
+	res, err := hc.writeHeaders(chain)
 
+	// Report some public statistics so the user has a clue what's going on
 	context := []interface{}{
-		"count", stats.processed, "elapsed", common.PrettyDuration(time.Since(start)),
-		"number", last.Number, "hash", last.Hash(),
+		"count", res.imported,
+		"elapsed", common.PrettyDuration(time.Since(start)),
+	}
+	if err != nil {
+		context = append(context, "err", err)
 	}
-	if timestamp := time.Unix(int64(last.Time), 0); time.Since(timestamp) > time.Minute {
-		context = append(context, []interface{}{"age", common.PrettyAge(timestamp)}...)
+	if last := res.lastHeader; last != nil {
+		context = append(context, "number", last.Number, "hash", res.lastHash)
+		if timestamp := time.Unix(int64(last.Time), 0); time.Since(timestamp) > time.Minute {
+			context = append(context, []interface{}{"age", common.PrettyAge(timestamp)}...)
+		}
 	}
-	if stats.ignored > 0 {
-		context = append(context, []interface{}{"ignored", stats.ignored}...)
+	if res.ignored > 0 {
+		context = append(context, []interface{}{"ignored", res.ignored}...)
 	}
 	log.Info("Imported new block headers", context...)
-
-	return 0, nil
+	return res.status, err
 }
 
 // GetBlockHashesFromHash retrieves a number of block hashes starting at a given