core/state, trie: port changes from PBSS (#26763)

Author: rjl493456442

core/state/statedb.go

@@ -970,8 +970,8 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 		storageTrieNodesUpdated int
 		storageTrieNodesDeleted int
 		nodes                   = trie.NewMergedNodeSet()
+		codeWriter              = s.db.DiskDB().NewBatch()
 	)
-	codeWriter := s.db.DiskDB().NewBatch()
 	for addr := range s.stateObjectsDirty {
 		if obj := s.stateObjects[addr]; !obj.deleted {
 			// Write any contract code associated with the state object

trie/committer.go

@@ -33,29 +33,20 @@ type leaf struct {
 // insertion order.
 type committer struct {
 	nodes       *NodeSet
-	tracer      *tracer
 	collectLeaf bool
 }
 
 // newCommitter creates a new committer or picks one from the pool.
-func newCommitter(owner common.Hash, tracer *tracer, collectLeaf bool) *committer {
+func newCommitter(nodeset *NodeSet, collectLeaf bool) *committer {
 	return &committer{
-		nodes:       NewNodeSet(owner),
-		tracer:      tracer,
+		nodes:       nodeset,
 		collectLeaf: collectLeaf,
 	}
 }
 
-// Commit collapses a node down into a hash node and returns it along with
-// the modified nodeset.
-func (c *committer) Commit(n node) (hashNode, *NodeSet) {
-	h := c.commit(nil, n)
-	// Some nodes can be deleted from trie which can't be captured
-	// by committer itself. Iterate all deleted nodes tracked by
-	// tracer and marked them as deleted only if they are present
-	// in database previously.
-	c.tracer.markDeletions(c.nodes)
-	return h.(hashNode), c.nodes
+// Commit collapses a node down into a hash node.
+func (c *committer) Commit(n node) hashNode {
+	return c.commit(nil, n).(hashNode)
 }
 
 // commit collapses a node down into a hash node and returns it.
@@ -74,9 +65,7 @@ func (c *committer) commit(path []byte, n node) node {
 		// If the child is fullNode, recursively commit,
 		// otherwise it can only be hashNode or valueNode.
 		if _, ok := cn.Val.(*fullNode); ok {
-			childV := c.commit(append(path, cn.Key...), cn.Val)
-
-			collapsed.Val = childV
+			collapsed.Val = c.commit(append(path, cn.Key...), cn.Val)
 		}
 		// The key needs to be copied, since we're adding it to the
 		// modified nodeset.
@@ -85,12 +74,6 @@ func (c *committer) commit(path []byte, n node) node {
 		if hn, ok := hashedNode.(hashNode); ok {
 			return hn
 		}
-		// The short node now is embedded in its parent. Mark the node as
-		// deleted if it's present in database previously. It's equivalent
-		// as deletion from database's perspective.
-		if prev := c.tracer.getPrev(path); len(prev) != 0 {
-			c.nodes.markDeleted(path, prev)
-		}
 		return collapsed
 	case *fullNode:
 		hashedKids := c.commitChildren(path, cn)
@@ -101,12 +84,6 @@ func (c *committer) commit(path []byte, n node) node {
 		if hn, ok := hashedNode.(hashNode); ok {
 			return hn
 		}
-		// The full node now is embedded in its parent. Mark the node as
-		// deleted if it's present in database previously. It's equivalent
-		// as deletion from database's perspective.
-		if prev := c.tracer.getPrev(path); len(prev) != 0 {
-			c.nodes.markDeleted(path, prev)
-		}
 		return collapsed
 	case hashNode:
 		return cn
@@ -134,8 +111,7 @@ func (c *committer) commitChildren(path []byte, n *fullNode) [17]node {
 		// Commit the child recursively and store the "hashed" value.
 		// Note the returned node can be some embedded nodes, so it's
 		// possible the type is not hashNode.
-		hashed := c.commit(append(path, byte(i)), child)
-		children[i] = hashed
+		children[i] = c.commit(append(path, byte(i)), child)
 	}
 	// For the 17th child, it's possible the type is valuenode.
 	if n.Children[16] != nil {
@@ -155,6 +131,12 @@ func (c *committer) store(path []byte, n node) node {
 	// usually is leaf node). But small value (less than 32bytes) is not
 	// our target (leaves in account trie only).
 	if hash == nil {
+		// The node is embedded in its parent, in other words, this node
+		// will not be stored in the database independently, mark it as
+		// deleted only if the node was existent in database before.
+		if _, ok := c.nodes.accessList[string(path)]; ok {
+			c.nodes.markDeleted(path)
+		}
 		return n
 	}
 	// We have the hash already, estimate the RLP encoding-size of the node.
@@ -169,7 +151,7 @@ func (c *committer) store(path []byte, n node) node {
 		}
 	)
 	// Collect the dirty node to nodeset for return.
-	c.nodes.markUpdated(path, mnode, c.tracer.getPrev(path))
+	c.nodes.markUpdated(path, mnode)
 
 	// Collect the corresponding leaf node if it's required. We don't check
 	// full node since it's impossible to store value in fullNode. The key

trie/database.go

@@ -792,13 +792,12 @@ func (db *Database) Update(nodes *MergedNodeSet) error {
 	}
 	for _, owner := range order {
 		subset := nodes.sets[owner]
-		for _, path := range subset.updates.order {
-			n, ok := subset.updates.nodes[path]
-			if !ok {
-				return fmt.Errorf("missing node %x %v", owner, path)
+		subset.forEachWithOrder(func(path string, n *memoryNode) {
+			if n.isDeleted() {
+				return // ignore deletion
 			}
 			db.insert(n.hash, int(n.size), n.node)
-		}
+		})
 	}
 	// Link up the account trie and storage trie if the node points
 	// to an account trie leaf.

trie/nodeset.go

@@ -19,6 +19,7 @@ package trie
 import (
 	"fmt"
 	"reflect"
+	"sort"
 	"strings"
 
 	"github.com/ethereum/go-ethereum/common"
@@ -40,8 +41,8 @@ var memoryNodeSize = int(reflect.TypeOf(memoryNode{}).Size())
 
 // memorySize returns the total memory size used by this node.
 // nolint:unused
-func (n *memoryNode) memorySize(key int) int {
-	return int(n.size) + memoryNodeSize + key
+func (n *memoryNode) memorySize(pathlen int) int {
+	return int(n.size) + memoryNodeSize + pathlen
 }
 
 // rlp returns the raw rlp encoded blob of the cached trie node, either directly
@@ -64,7 +65,13 @@ func (n *memoryNode) obj() node {
 	return expandNode(n.hash[:], n.node)
 }
 
+// isDeleted returns the indicator if the node is marked as deleted.
+func (n *memoryNode) isDeleted() bool {
+	return n.hash == (common.Hash{})
+}
+
 // nodeWithPrev wraps the memoryNode with the previous node value.
+// nolint: unused
 type nodeWithPrev struct {
 	*memoryNode
 	prev []byte // RLP-encoded previous value, nil means it's non-existent
@@ -79,81 +86,77 @@ func (n *nodeWithPrev) unwrap() *memoryNode {
 // memorySize returns the total memory size used by this node. It overloads
 // the function in memoryNode by counting the size of previous value as well.
 // nolint: unused
-func (n *nodeWithPrev) memorySize(key int) int {
-	return n.memoryNode.memorySize(key) + len(n.prev)
-}
-
-// nodesWithOrder represents a collection of dirty nodes which includes
-// newly-inserted and updated nodes. The modification order of all nodes
-// is represented by order list.
-type nodesWithOrder struct {
-	order []string                 // the path list of dirty nodes, sort by insertion order
-	nodes map[string]*nodeWithPrev // the map of dirty nodes, keyed by node path
+func (n *nodeWithPrev) memorySize(pathlen int) int {
+	return n.memoryNode.memorySize(pathlen) + len(n.prev)
 }
 
 // NodeSet contains all dirty nodes collected during the commit operation.
 // Each node is keyed by path. It's not thread-safe to use.
 type NodeSet struct {
-	owner   common.Hash       // the identifier of the trie
-	updates *nodesWithOrder   // the set of updated nodes(newly inserted, updated)
-	deletes map[string][]byte // the map of deleted nodes, keyed by node
-	leaves  []*leaf           // the list of dirty leaves
+	owner   common.Hash            // the identifier of the trie
+	nodes   map[string]*memoryNode // the set of dirty nodes(inserted, updated, deleted)
+	leaves  []*leaf                // the list of dirty leaves
+	updates int                    // the count of updated and inserted nodes
+	deletes int                    // the count of deleted nodes
+
+	// The list of accessed nodes, which records the original node value.
+	// The origin value is expected to be nil for newly inserted node
+	// and is expected to be non-nil for other types(updated, deleted).
+	accessList map[string][]byte
 }
 
 // NewNodeSet initializes an empty node set to be used for tracking dirty nodes
 // from a specific account or storage trie. The owner is zero for the account
 // trie and the owning account address hash for storage tries.
-func NewNodeSet(owner common.Hash) *NodeSet {
+func NewNodeSet(owner common.Hash, accessList map[string][]byte) *NodeSet {
 	return &NodeSet{
-		owner: owner,
-		updates: &nodesWithOrder{
-			nodes: make(map[string]*nodeWithPrev),
-		},
-		deletes: make(map[string][]byte),
+		owner:      owner,
+		nodes:      make(map[string]*memoryNode),
+		accessList: accessList,
 	}
 }
 
-/*
-// NewNodeSetWithDeletion initializes the nodeset with provided deletion set.
-func NewNodeSetWithDeletion(owner common.Hash, paths [][]byte, prev [][]byte) *NodeSet {
-	set := NewNodeSet(owner)
-	for i, path := range paths {
-		set.markDeleted(path, prev[i])
+// forEachWithOrder iterates the dirty nodes with the order from bottom to top,
+// right to left, nodes with the longest path will be iterated first.
+func (set *NodeSet) forEachWithOrder(callback func(path string, n *memoryNode)) {
+	var paths sort.StringSlice
+	for path := range set.nodes {
+		paths = append(paths, path)
+	}
+	// Bottom-up, longest path first
+	sort.Sort(sort.Reverse(paths))
+	for _, path := range paths {
+		callback(path, set.nodes[path])
 	}
-	return set
 }
-*/
 
-// markUpdated marks the node as dirty(newly-inserted or updated) with provided
-// node path, node object along with its previous value.
-func (set *NodeSet) markUpdated(path []byte, node *memoryNode, prev []byte) {
-	set.updates.order = append(set.updates.order, string(path))
-	set.updates.nodes[string(path)] = &nodeWithPrev{
-		memoryNode: node,
-		prev:       prev,
-	}
+// markUpdated marks the node as dirty(newly-inserted or updated).
+func (set *NodeSet) markUpdated(path []byte, node *memoryNode) {
+	set.nodes[string(path)] = node
+	set.updates += 1
 }
 
-// markDeleted marks the node as deleted with provided path and previous value.
-func (set *NodeSet) markDeleted(path []byte, prev []byte) {
-	set.deletes[string(path)] = prev
+// markDeleted marks the node as deleted.
+func (set *NodeSet) markDeleted(path []byte) {
+	set.nodes[string(path)] = &memoryNode{}
+	set.deletes += 1
 }
 
 // addLeaf collects the provided leaf node into set.
 func (set *NodeSet) addLeaf(node *leaf) {
 	set.leaves = append(set.leaves, node)
 }
 
-// Size returns the number of updated and deleted nodes contained in the set.
+// Size returns the number of dirty nodes in set.
 func (set *NodeSet) Size() (int, int) {
-	return len(set.updates.order), len(set.deletes)
+	return set.updates, set.deletes
 }
 
 // Hashes returns the hashes of all updated nodes. TODO(rjl493456442) how can
 // we get rid of it?
 func (set *NodeSet) Hashes() []common.Hash {
 	var ret []common.Hash
-	for _, node := range set.updates.nodes {
+	for _, node := range set.nodes {
 		ret = append(ret, node.hash)
 	}
 	return ret
@@ -163,19 +166,23 @@ func (set *NodeSet) Hashes() []common.Hash {
 func (set *NodeSet) Summary() string {
 	var out = new(strings.Builder)
 	fmt.Fprintf(out, "nodeset owner: %v\n", set.owner)
-	if set.updates != nil {
-		for _, key := range set.updates.order {
-			updated := set.updates.nodes[key]
-			if updated.prev != nil {
-				fmt.Fprintf(out, "  [*]: %x -> %v prev: %x\n", key, updated.hash, updated.prev)
-			} else {
-				fmt.Fprintf(out, "  [+]: %x -> %v\n", key, updated.hash)
+	if set.nodes != nil {
+		for path, n := range set.nodes {
+			// Deletion
+			if n.isDeleted() {
+				fmt.Fprintf(out, "  [-]: %x prev: %x\n", path, set.accessList[path])
+				continue
 			}
+			// Insertion
+			origin, ok := set.accessList[path]
+			if !ok {
+				fmt.Fprintf(out, "  [+]: %x -> %v\n", path, n.hash)
+				continue
+			}
+			// Update
+			fmt.Fprintf(out, "  [*]: %x -> %v prev: %x\n", path, n.hash, origin)
 		}
 	}
-	for k, n := range set.deletes {
-		fmt.Fprintf(out, "  [-]: %x -> %x\n", k, n)
-	}
 	for _, n := range set.leaves {
 		fmt.Fprintf(out, "[leaf]: %v\n", n)
 	}

trie/proof.go

@@ -563,7 +563,7 @@ func VerifyRangeProof(rootHash common.Hash, firstKey []byte, lastKey []byte, key
 	}
 	// Rebuild the trie with the leaf stream, the shape of trie
 	// should be same with the original one.
-	tr := &Trie{root: root, reader: newEmptyReader()}
+	tr := &Trie{root: root, reader: newEmptyReader(), tracer: newTracer()}
 	if empty {
 		tr.root = nil
 	}