core, trie: intermediate mempool between trie and database (#15857)

This commit reduces database I/O by not writing every state trie to disk.

7 files changed, 64 insertions, 41 deletions

diff --git a/light/lightchain.go b/light/lightchain.go
index f47957512..24529ef82 100644
--- a/light/lightchain.go
+++ b/light/lightchain.go
@@ -18,6 +18,7 @@ package light
 
 import (
 	"context"
+	"errors"
 	"math/big"
 	"sync"
 	"sync/atomic"
@@ -26,6 +27,7 @@ import (
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/consensus"
 	"github.com/ethereum/go-ethereum/core"
+	"github.com/ethereum/go-ethereum/core/state"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/event"
@@ -212,6 +214,11 @@ func (bc *LightChain) Genesis() *types.Block {
 	return bc.genesisBlock
 }
 
+// State returns a new mutable state based on the current HEAD block.
+func (bc *LightChain) State() (*state.StateDB, error) {
+	return nil, errors.New("not implemented, needs client/server interface split")
+}
+
 // GetBody retrieves a block body (transactions and uncles) from the database
 // or ODR service by hash, caching it if found.
 func (self *LightChain) GetBody(ctx context.Context, hash common.Hash) (*types.Body, error) {
diff --git a/light/nodeset.go b/light/nodeset.go
index c530a4fbe..ffdb71bb7 100644
--- a/light/nodeset.go
+++ b/light/nodeset.go
@@ -22,8 +22,8 @@ import (
 
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/crypto"
+	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/rlp"
-	"github.com/ethereum/go-ethereum/trie"
 )
 
 // NodeSet stores a set of trie nodes. It implements trie.Database and can also
@@ -99,7 +99,7 @@ func (db *NodeSet) NodeList() NodeList {
 }
 
 // Store writes the contents of the set to the given database
-func (db *NodeSet) Store(target trie.Database) {
+func (db *NodeSet) Store(target ethdb.Putter) {
 	db.lock.RLock()
 	defer db.lock.RUnlock()
 
@@ -108,11 +108,11 @@ func (db *NodeSet) Store(target trie.Database) {
 	}
 }
 
-// NodeList stores an ordered list of trie nodes. It implements trie.DatabaseWriter.
+// NodeList stores an ordered list of trie nodes. It implements ethdb.Putter.
 type NodeList []rlp.RawValue
 
 // Store writes the contents of the list to the given database
-func (n NodeList) Store(db trie.Database) {
+func (n NodeList) Store(db ethdb.Putter) {
 	for _, node := range n {
 		db.Put(crypto.Keccak256(node), node)
 	}
diff --git a/light/odr_test.go b/light/odr_test.go
index e3d07518a..d3f9374fd 100644
--- a/light/odr_test.go
+++ b/light/odr_test.go
@@ -74,7 +74,7 @@ func (odr *testOdr) Retrieve(ctx context.Context, req OdrRequest) error {
 	case *ReceiptsRequest:
 		req.Receipts = core.GetBlockReceipts(odr.sdb, req.Hash, core.GetBlockNumber(odr.sdb, req.Hash))
 	case *TrieRequest:
-		t, _ := trie.New(req.Id.Root, odr.sdb)
+		t, _ := trie.New(req.Id.Root, trie.NewDatabase(odr.sdb))
 		nodes := NewNodeSet()
 		t.Prove(req.Key, 0, nodes)
 		req.Proof = nodes
@@ -239,7 +239,7 @@ func testChainOdr(t *testing.T, protocol int, fn odrTestFn) {
 	)
 	gspec.MustCommit(ldb)
 	// Assemble the test environment
-	blockchain, _ := core.NewBlockChain(sdb, params.TestChainConfig, ethash.NewFullFaker(), vm.Config{})
+	blockchain, _ := core.NewBlockChain(sdb, nil, params.TestChainConfig, ethash.NewFullFaker(), vm.Config{})
 	gchain, _ := core.GenerateChain(params.TestChainConfig, genesis, ethash.NewFaker(), sdb, 4, testChainGen)
 	if _, err := blockchain.InsertChain(gchain); err != nil {
 		t.Fatal(err)
diff --git a/light/postprocess.go b/light/postprocess.go
index 32dbc102b..bbac58d12 100644
--- a/light/postprocess.go
+++ b/light/postprocess.go
@@ -113,7 +113,8 @@ func StoreChtRoot(db ethdb.Database, sectionIdx uint64, sectionHead, root common
 
 // ChtIndexerBackend implements core.ChainIndexerBackend
 type ChtIndexerBackend struct {
-	db, cdb              ethdb.Database
+	diskdb               ethdb.Database
+	triedb               *trie.Database
 	section, sectionSize uint64
 	lastHash             common.Hash
 	trie                 *trie.Trie
@@ -121,8 +122,6 @@ type ChtIndexerBackend struct {
 
 // NewBloomTrieIndexer creates a BloomTrie chain indexer
 func NewChtIndexer(db ethdb.Database, clientMode bool) *core.ChainIndexer {
-	cdb := ethdb.NewTable(db, ChtTablePrefix)
-	idb := ethdb.NewTable(db, "chtIndex-")
 	var sectionSize, confirmReq uint64
 	if clientMode {
 		sectionSize = ChtFrequency
@@ -131,17 +130,23 @@ func NewChtIndexer(db ethdb.Database, clientMode bool) *core.ChainIndexer {
 		sectionSize = ChtV1Frequency
 		confirmReq = HelperTrieProcessConfirmations
 	}
-	return core.NewChainIndexer(db, idb, &ChtIndexerBackend{db: db, cdb: cdb, sectionSize: sectionSize}, sectionSize, confirmReq, time.Millisecond*100, "cht")
+	idb := ethdb.NewTable(db, "chtIndex-")
+	backend := &ChtIndexerBackend{
+		diskdb:      db,
+		triedb:      trie.NewDatabase(ethdb.NewTable(db, ChtTablePrefix)),
+		sectionSize: sectionSize,
+	}
+	return core.NewChainIndexer(db, idb, backend, sectionSize, confirmReq, time.Millisecond*100, "cht")
 }
 
 // Reset implements core.ChainIndexerBackend
 func (c *ChtIndexerBackend) Reset(section uint64, lastSectionHead common.Hash) error {
 	var root common.Hash
 	if section > 0 {
-		root = GetChtRoot(c.db, section-1, lastSectionHead)
+		root = GetChtRoot(c.diskdb, section-1, lastSectionHead)
 	}
 	var err error
-	c.trie, err = trie.New(root, c.cdb)
+	c.trie, err = trie.New(root, c.triedb)
 	c.section = section
 	return err
 }
@@ -151,7 +156,7 @@ func (c *ChtIndexerBackend) Process(header *types.Header) {
 	hash, num := header.Hash(), header.Number.Uint64()
 	c.lastHash = hash
 
-	td := core.GetTd(c.db, hash, num)
+	td := core.GetTd(c.diskdb, hash, num)
 	if td == nil {
 		panic(nil)
 	}
@@ -163,17 +168,16 @@ func (c *ChtIndexerBackend) Process(header *types.Header) {
 
 // Commit implements core.ChainIndexerBackend
 func (c *ChtIndexerBackend) Commit() error {
-	batch := c.cdb.NewBatch()
-	root, err := c.trie.CommitTo(batch)
+	root, err := c.trie.Commit(nil)
 	if err != nil {
 		return err
-	} else {
-		batch.Write()
-		if ((c.section+1)*c.sectionSize)%ChtFrequency == 0 {
-			log.Info("Storing CHT", "idx", c.section*c.sectionSize/ChtFrequency, "sectionHead", fmt.Sprintf("%064x", c.lastHash), "root", fmt.Sprintf("%064x", root))
-		}
-		StoreChtRoot(c.db, c.section, c.lastHash, root)
 	}
+	c.triedb.Commit(root, false)
+
+	if ((c.section+1)*c.sectionSize)%ChtFrequency == 0 {
+		log.Info("Storing CHT", "idx", c.section*c.sectionSize/ChtFrequency, "sectionHead", fmt.Sprintf("%064x", c.lastHash), "root", fmt.Sprintf("%064x", root))
+	}
+	StoreChtRoot(c.diskdb, c.section, c.lastHash, root)
 	return nil
 }
 
@@ -205,7 +209,8 @@ func StoreBloomTrieRoot(db ethdb.Database, sectionIdx uint64, sectionHead, root
 
 // BloomTrieIndexerBackend implements core.ChainIndexerBackend
 type BloomTrieIndexerBackend struct {
-	db, cdb                                    ethdb.Database
+	diskdb                                     ethdb.Database
+	triedb                                     *trie.Database
 	section, parentSectionSize, bloomTrieRatio uint64
 	trie                                       *trie.Trie
 	sectionHeads                               []common.Hash
@@ -213,9 +218,12 @@ type BloomTrieIndexerBackend struct {
 
 // NewBloomTrieIndexer creates a BloomTrie chain indexer
 func NewBloomTrieIndexer(db ethdb.Database, clientMode bool) *core.ChainIndexer {
-	cdb := ethdb.NewTable(db, BloomTrieTablePrefix)
+	backend := &BloomTrieIndexerBackend{
+		diskdb: db,
+		triedb: trie.NewDatabase(ethdb.NewTable(db, BloomTrieTablePrefix)),
+	}
 	idb := ethdb.NewTable(db, "bltIndex-")
-	backend := &BloomTrieIndexerBackend{db: db, cdb: cdb}
+
 	var confirmReq uint64
 	if clientMode {
 		backend.parentSectionSize = BloomTrieFrequency
@@ -233,10 +241,10 @@ func NewBloomTrieIndexer(db ethdb.Database, clientMode bool) *core.ChainIndexer
 func (b *BloomTrieIndexerBackend) Reset(section uint64, lastSectionHead common.Hash) error {
 	var root common.Hash
 	if section > 0 {
-		root = GetBloomTrieRoot(b.db, section-1, lastSectionHead)
+		root = GetBloomTrieRoot(b.diskdb, section-1, lastSectionHead)
 	}
 	var err error
-	b.trie, err = trie.New(root, b.cdb)
+	b.trie, err = trie.New(root, b.triedb)
 	b.section = section
 	return err
 }
@@ -259,7 +267,7 @@ func (b *BloomTrieIndexerBackend) Commit() error {
 		binary.BigEndian.PutUint64(encKey[2:10], b.section)
 		var decomp []byte
 		for j := uint64(0); j < b.bloomTrieRatio; j++ {
-			data, err := core.GetBloomBits(b.db, i, b.section*b.bloomTrieRatio+j, b.sectionHeads[j])
+			data, err := core.GetBloomBits(b.diskdb, i, b.section*b.bloomTrieRatio+j, b.sectionHeads[j])
 			if err != nil {
 				return err
 			}
@@ -279,17 +287,15 @@ func (b *BloomTrieIndexerBackend) Commit() error {
 			b.trie.Delete(encKey[:])
 		}
 	}
-
-	batch := b.cdb.NewBatch()
-	root, err := b.trie.CommitTo(batch)
+	root, err := b.trie.Commit(nil)
 	if err != nil {
 		return err
-	} else {
-		batch.Write()
-		sectionHead := b.sectionHeads[b.bloomTrieRatio-1]
-		log.Info("Storing BloomTrie", "section", b.section, "sectionHead", fmt.Sprintf("%064x", sectionHead), "root", fmt.Sprintf("%064x", root), "compression ratio", float64(compSize)/float64(decompSize))
-		StoreBloomTrieRoot(b.db, b.section, sectionHead, root)
 	}
+	b.triedb.Commit(root, false)
+
+	sectionHead := b.sectionHeads[b.bloomTrieRatio-1]
+	log.Info("Storing BloomTrie", "section", b.section, "sectionHead", fmt.Sprintf("%064x", sectionHead), "root", fmt.Sprintf("%064x", root), "compression ratio", float64(compSize)/float64(decompSize))
+	StoreBloomTrieRoot(b.diskdb, b.section, sectionHead, root)
 
 	return nil
 }
diff --git a/light/trie.go b/light/trie.go
index 7a9c86b98..c07e99461 100644
--- a/light/trie.go
+++ b/light/trie.go
@@ -18,12 +18,14 @@ package light
 
 import (
 	"context"
+	"errors"
 	"fmt"
 
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/core/state"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/crypto"
+	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/trie"
 )
 
@@ -83,6 +85,10 @@ func (db *odrDatabase) ContractCodeSize(addrHash, codeHash common.Hash) (int, er
 	return len(code), err
 }
 
+func (db *odrDatabase) TrieDB() *trie.Database {
+	return nil
+}
+
 type odrTrie struct {
 	db   *odrDatabase
 	id   *TrieID
@@ -113,11 +119,11 @@ func (t *odrTrie) TryDelete(key []byte) error {
 	})
 }
 
-func (t *odrTrie) CommitTo(db trie.DatabaseWriter) (common.Hash, error) {
+func (t *odrTrie) Commit(onleaf trie.LeafCallback) (common.Hash, error) {
 	if t.trie == nil {
 		return t.id.Root, nil
 	}
-	return t.trie.CommitTo(db)
+	return t.trie.Commit(onleaf)
 }
 
 func (t *odrTrie) Hash() common.Hash {
@@ -135,13 +141,17 @@ func (t *odrTrie) GetKey(sha []byte) []byte {
 	return nil
 }
 
+func (t *odrTrie) Prove(key []byte, fromLevel uint, proofDb ethdb.Putter) error {
+	return errors.New("not implemented, needs client/server interface split")
+}
+
 // do tries and retries to execute a function until it returns with no error or
 // an error type other than MissingNodeError
 func (t *odrTrie) do(key []byte, fn func() error) error {
 	for {
 		var err error
 		if t.trie == nil {
-			t.trie, err = trie.New(t.id.Root, t.db.backend.Database())
+			t.trie, err = trie.New(t.id.Root, trie.NewDatabase(t.db.backend.Database()))
 		}
 		if err == nil {
 			err = fn()
@@ -167,7 +177,7 @@ func newNodeIterator(t *odrTrie, startkey []byte) trie.NodeIterator {
 	// Open the actual non-ODR trie if that hasn't happened yet.
 	if t.trie == nil {
 		it.do(func() error {
-			t, err := trie.New(t.id.Root, t.db.backend.Database())
+			t, err := trie.New(t.id.Root, trie.NewDatabase(t.db.backend.Database()))
 			if err == nil {
 				it.t.trie = t
 			}
diff --git a/light/trie_test.go b/light/trie_test.go
index d99664718..0d6b2cc1d 100644
--- a/light/trie_test.go
+++ b/light/trie_test.go
@@ -40,7 +40,7 @@ func TestNodeIterator(t *testing.T) {
 		genesis    = gspec.MustCommit(fulldb)
 	)
 	gspec.MustCommit(lightdb)
-	blockchain, _ := core.NewBlockChain(fulldb, params.TestChainConfig, ethash.NewFullFaker(), vm.Config{})
+	blockchain, _ := core.NewBlockChain(fulldb, nil, params.TestChainConfig, ethash.NewFullFaker(), vm.Config{})
 	gchain, _ := core.GenerateChain(params.TestChainConfig, genesis, ethash.NewFaker(), fulldb, 4, testChainGen)
 	if _, err := blockchain.InsertChain(gchain); err != nil {
 		panic(err)
diff --git a/light/txpool_test.go b/light/txpool_test.go
index b343f79b0..13d7d3ceb 100644
--- a/light/txpool_test.go
+++ b/light/txpool_test.go
@@ -88,7 +88,7 @@ func TestTxPool(t *testing.T) {
 	)
 	gspec.MustCommit(ldb)
 	// Assemble the test environment
-	blockchain, _ := core.NewBlockChain(sdb, params.TestChainConfig, ethash.NewFullFaker(), vm.Config{})
+	blockchain, _ := core.NewBlockChain(sdb, nil, params.TestChainConfig, ethash.NewFullFaker(), vm.Config{})
 	gchain, _ := core.GenerateChain(params.TestChainConfig, genesis, ethash.NewFaker(), sdb, poolTestBlocks, txPoolTestChainGen)
 	if _, err := blockchain.InsertChain(gchain); err != nil {
 		panic(err)