core/rawdb: refactor db inspector for extending multiple ancient store (#25896)

This PR ports a few changes from PBSS:

- Fix the snapshot generator waiter in case the generation is not even initialized
- Refactor db inspector for ancient store

core/state/snapshot/snapshot.go

@@ -187,8 +187,9 @@ type Tree struct {
 // If the memory layers in the journal do not match the disk layer (e.g. there is
 // a gap) or the journal is missing, there are two repair cases:
 //
-//   - if the 'recovery' parameter is true, all memory diff-layers will be discarded.
-//     This case happens when the snapshot is 'ahead' of the state trie.
+//   - if the 'recovery' parameter is true, memory diff-layers and the disk-layer
+//     will all be kept. This case happens when the snapshot is 'ahead' of the
+//     state trie.
 //   - otherwise, the entire snapshot is considered invalid and will be recreated on
 //     a background thread.
 func New(config Config, diskdb ethdb.KeyValueStore, triedb *trie.Database, root common.Hash) (*Tree, error) {
@@ -199,16 +200,16 @@ func New(config Config, diskdb ethdb.KeyValueStore, triedb *trie.Database, root
 		triedb: triedb,
 		layers: make(map[common.Hash]snapshot),
 	}
-	// Create the building waiter iff the background generation is allowed
-	if !config.NoBuild && !config.AsyncBuild {
-		defer snap.waitBuild()
-	}
 	// Attempt to load a previously persisted snapshot and rebuild one if failed
 	head, disabled, err := loadSnapshot(diskdb, triedb, root, config.CacheSize, config.Recovery, config.NoBuild)
 	if disabled {
 		log.Warn("Snapshot maintenance disabled (syncing)")
 		return snap, nil
 	}
+	// Create the building waiter iff the background generation is allowed
+	if !config.NoBuild && !config.AsyncBuild {
+		defer snap.waitBuild()
+	}
 	if err != nil {
 		log.Warn("Failed to load snapshot", "err", err)
 		if !config.NoBuild {

core/state/state_object.go

@@ -63,7 +63,7 @@ func (s Storage) Copy() Storage {
 // The usage pattern is as follows:
 // First you need to obtain a state object.
 // Account values can be accessed and modified through the object.
-// Finally, call CommitTrie to write the modified storage trie into a database.
+// Finally, call commitTrie to write the modified storage trie into a database.
 type stateObject struct {
 	address  common.Address
 	addrHash common.Hash // hash of ethereum address of the account
@@ -374,9 +374,9 @@ func (s *stateObject) updateRoot(db Database) {
 	s.data.Root = s.trie.Hash()
 }
 
-// CommitTrie the storage trie of the object to db.
-// This updates the trie root.
-func (s *stateObject) CommitTrie(db Database) (*trie.NodeSet, error) {
+// commitTrie submits the storage changes into the storage trie and re-computes
+// the root. Besides, all trie changes will be collected in a nodeset and returned.
+func (s *stateObject) commitTrie(db Database) (*trie.NodeSet, error) {
 	// If nothing changed, don't bother with hashing anything
 	if s.updateTrie(db) == nil {
 		return nil, nil

core/state/statedb.go

@@ -920,7 +920,7 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 				obj.dirtyCode = false
 			}
 			// Write any storage changes in the state object to its storage trie
-			set, err := obj.CommitTrie(s.db)
+			set, err := obj.commitTrie(s.db)
 			if err != nil {
 				return common.Hash{}, err
 			}
@@ -934,6 +934,12 @@ func (s *StateDB) Commit(deleteEmptyObjects bool) (common.Hash, error) {
 				storageTrieNodesDeleted += deleted
 			}
 		}
+		// If the contract is destructed, the storage is still left in the
+		// database as dangling data. Theoretically it's should be wiped from
+		// database as well, but in hash-based-scheme it's extremely hard to
+		// determine that if the trie nodes are also referenced by other storage,
+		// and in path-based-scheme some technical challenges are still unsolved.
+		// Although it won't affect the correctness but please fix it TODO(rjl493456442).
 	}
 	if len(s.stateObjectsDirty) > 0 {
 		s.stateObjectsDirty = make(map[common.Address]struct{})