perf(core): make TrieDb use NodeHash as key (#2517)

**Motivation**

Followup on #2516 using the fact that NodeHash is Copy to use it as the
key for the trie db instead of a Vec

**Description**

Changes the trait TrieDb to use a NodeHash as key instead of a generic
vec, allowing less expensive clones when passing around keys since
NodeHash is copy and doesn't do any allocation.

Author: edg-l

CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## Perf
 
+### 2025-04-28
+
+- Make TrieDb trait use NodeHash as key [2517](https://github.com/lambdaclass/ethrex/pull/2517)
+
 ### 2025-04-22
 
 - Avoid calculating state transitions after every block in bulk mode [2519](https://github.com/lambdaclass/ethrex/pull/2519)

crates/common/trie/db.rs

@@ -1,23 +1,23 @@
-use crate::error::TrieError;
+use crate::{error::TrieError, NodeHash};
 use std::{
     collections::HashMap,
     sync::{Arc, Mutex},
 };
 
 pub trait TrieDB: Send + Sync {
-    fn get(&self, key: Vec<u8>) -> Result<Option<Vec<u8>>, TrieError>;
-    fn put(&self, key: Vec<u8>, value: Vec<u8>) -> Result<(), TrieError>;
+    fn get(&self, key: NodeHash) -> Result<Option<Vec<u8>>, TrieError>;
+    fn put(&self, key: NodeHash, value: Vec<u8>) -> Result<(), TrieError>;
     // fn put_batch(&self, key: Vec<u8>, value: Vec<u8>) -> Result<(), TrieError>;
-    fn put_batch(&self, key_values: Vec<(Vec<u8>, Vec<u8>)>) -> Result<(), TrieError>;
+    fn put_batch(&self, key_values: Vec<(NodeHash, Vec<u8>)>) -> Result<(), TrieError>;
 }
 
 /// InMemory implementation for the TrieDB trait, with get and put operations.
 pub struct InMemoryTrieDB {
-    inner: Arc<Mutex<HashMap<Vec<u8>, Vec<u8>>>>,
+    inner: Arc<Mutex<HashMap<NodeHash, Vec<u8>>>>,
 }
 
 impl InMemoryTrieDB {
-    pub const fn new(map: Arc<Mutex<HashMap<Vec<u8>, Vec<u8>>>>) -> Self {
+    pub const fn new(map: Arc<Mutex<HashMap<NodeHash, Vec<u8>>>>) -> Self {
         Self { inner: map }
     }
     pub fn new_empty() -> Self {
@@ -28,7 +28,7 @@ impl InMemoryTrieDB {
 }
 
 impl TrieDB for InMemoryTrieDB {
-    fn get(&self, key: Vec<u8>) -> Result<Option<Vec<u8>>, TrieError> {
+    fn get(&self, key: NodeHash) -> Result<Option<Vec<u8>>, TrieError> {
         Ok(self
             .inner
             .lock()
@@ -37,15 +37,15 @@ impl TrieDB for InMemoryTrieDB {
             .cloned())
     }
 
-    fn put(&self, key: Vec<u8>, value: Vec<u8>) -> Result<(), TrieError> {
+    fn put(&self, key: NodeHash, value: Vec<u8>) -> Result<(), TrieError> {
         self.inner
             .lock()
             .map_err(|_| TrieError::LockError)?
             .insert(key, value);
         Ok(())
     }
 
-    fn put_batch(&self, key_values: Vec<(Vec<u8>, Vec<u8>)>) -> Result<(), TrieError> {
+    fn put_batch(&self, key_values: Vec<(NodeHash, Vec<u8>)>) -> Result<(), TrieError> {
         let mut db = self.inner.lock().map_err(|_| TrieError::LockError)?;
 
         for (key, value) in key_values {

crates/common/trie/node_hash.rs

@@ -87,6 +87,20 @@ impl NodeHash {
         }
         encoder
     }
+
+    pub fn len(&self) -> usize {
+        match self {
+            NodeHash::Hashed(h256) => h256.as_bytes().len(),
+            NodeHash::Inline(value) => value.1 as usize,
+        }
+    }
+
+    pub fn is_empty(&self) -> bool {
+        match self {
+            NodeHash::Hashed(h256) => h256.as_bytes().is_empty(),
+            NodeHash::Inline(value) => value.1 == 0,
+        }
+    }
 }
 
 impl From<Vec<u8>> for NodeHash {

crates/common/trie/state.rs

@@ -33,7 +33,7 @@ impl TrieState {
             return Ok(Some(node.clone()));
         };
         self.db
-            .get(hash.into())?
+            .get(hash)?
             .map(|rlp| Node::decode(&rlp).map_err(TrieError::RLPDecode))
             .transpose()
     }
@@ -68,7 +68,7 @@ impl TrieState {
     fn commit_node_tail_recursive(
         &mut self,
         node_hash: &NodeHash,
-        acc: &mut Vec<(Vec<u8>, Vec<u8>)>,
+        acc: &mut Vec<(NodeHash, Vec<u8>)>,
     ) -> Result<(), TrieError> {
         let Some(node) = self.cache.remove(node_hash) else {
             // If the node is not in the cache then it means it is already stored in the DB
@@ -87,7 +87,7 @@ impl TrieState {
             Node::Leaf(_) => {}
         }
         // Commit self
-        acc.push((node_hash.into(), node.encode_to_vec()));
+        acc.push((*node_hash, node.encode_to_vec()));
 
         Ok(())
     }
@@ -96,7 +96,7 @@ impl TrieState {
     pub fn write_node(&mut self, node: Node, hash: NodeHash) -> Result<(), TrieError> {
         // Don't insert the node if it is already inlined on the parent
         if matches!(hash, NodeHash::Hashed(_)) {
-            self.db.put(hash.into(), node.encode_to_vec())?;
+            self.db.put(hash, node.encode_to_vec())?;
         }
         Ok(())
     }
@@ -108,7 +108,7 @@ impl TrieState {
             .iter()
             .filter_map(|node| {
                 let hash = node.compute_hash();
-                matches!(hash, NodeHash::Hashed(_)).then(|| (hash.into(), node.encode_to_vec()))
+                matches!(hash, NodeHash::Hashed(_)).then(|| (hash, node.encode_to_vec()))
             })
             .collect();
         self.db.put_batch(key_values)?;

crates/common/trie/trie.rs

@@ -11,14 +11,13 @@ mod trie_iter;
 mod verify_range;
 use ethereum_types::H256;
 use ethrex_rlp::constants::RLP_NULL;
-use node_hash::NodeHash;
 use sha3::{Digest, Keccak256};
 use std::collections::HashSet;
 
 pub use self::db::{InMemoryTrieDB, TrieDB};
 pub use self::nibbles::Nibbles;
 pub use self::verify_range::verify_range;
-pub use self::{node::Node, state::TrieState};
+pub use self::{node::Node, node_hash::NodeHash, state::TrieState};
 
 pub use self::error::TrieError;
 use self::{node::LeafNode, trie_iter::TrieIterator};
@@ -241,15 +240,15 @@ impl Trie {
         struct NullTrieDB;
 
         impl TrieDB for NullTrieDB {
-            fn get(&self, _key: Vec<u8>) -> Result<Option<Vec<u8>>, TrieError> {
+            fn get(&self, _key: NodeHash) -> Result<Option<Vec<u8>>, TrieError> {
                 Ok(None)
             }
 
-            fn put(&self, _key: Vec<u8>, _value: Vec<u8>) -> Result<(), TrieError> {
+            fn put(&self, _key: NodeHash, _value: Vec<u8>) -> Result<(), TrieError> {
                 Ok(())
             }
 
-            fn put_batch(&self, _key_values: Vec<(Vec<u8>, Vec<u8>)>) -> Result<(), TrieError> {
+            fn put_batch(&self, _key_values: Vec<(NodeHash, Vec<u8>)>) -> Result<(), TrieError> {
                 Ok(())
             }
         }
@@ -340,7 +339,7 @@ impl Trie {
         use std::sync::Arc;
         use std::sync::Mutex;
 
-        let hmap: HashMap<Vec<u8>, Vec<u8>> = HashMap::new();
+        let hmap: HashMap<NodeHash, Vec<u8>> = HashMap::new();
         let map = Arc::new(Mutex::new(hmap));
         let db = InMemoryTrieDB::new(map);
         Trie::new(Box::new(db))

crates/storage/store_db/in_memory.rs

@@ -9,14 +9,14 @@ use ethrex_common::types::{
     payload::PayloadBundle, AccountState, Block, BlockBody, BlockHash, BlockHeader, BlockNumber,
     ChainConfig, Index, Receipt,
 };
-use ethrex_trie::{InMemoryTrieDB, Nibbles, Trie};
+use ethrex_trie::{InMemoryTrieDB, Nibbles, NodeHash, Trie};
 use std::{
     collections::{BTreeMap, HashMap},
     fmt::Debug,
     sync::{Arc, Mutex, MutexGuard},
 };
 
-pub type NodeMap = Arc<Mutex<HashMap<Vec<u8>, Vec<u8>>>>;
+pub type NodeMap = Arc<Mutex<HashMap<NodeHash, Vec<u8>>>>;
 
 #[derive(Default, Clone)]
 pub struct Store(Arc<Mutex<StoreInner>>);

crates/storage/store_db/libmdbx.rs

@@ -19,7 +19,7 @@ use ethrex_common::types::{
 use ethrex_rlp::decode::RLPDecode;
 use ethrex_rlp::encode::RLPEncode;
 use ethrex_rlp::error::RLPDecodeError;
-use ethrex_trie::{Nibbles, Trie};
+use ethrex_trie::{Nibbles, NodeHash, Trie};
 use libmdbx::orm::{Decodable, DupSort, Encodable, Table};
 use libmdbx::{
     dupsort,
@@ -1150,7 +1150,7 @@ table!(
 
 table!(
     /// state trie nodes
-    ( StateTrieNodes ) Vec<u8> => Vec<u8>
+    ( StateTrieNodes ) NodeHash => Vec<u8>
 );
 
 // Local Blocks

crates/storage/trie_db/libmdbx.rs

@@ -1,4 +1,4 @@
-use ethrex_trie::error::TrieError;
+use ethrex_trie::{error::TrieError, NodeHash};
 use libmdbx::orm::{Database, Table};
 use std::{marker::PhantomData, sync::Arc};
 /// Libmdbx implementation for the TrieDB trait, with get and put operations.
@@ -11,7 +11,7 @@ use ethrex_trie::TrieDB;
 
 impl<T> LibmdbxTrieDB<T>
 where
-    T: Table<Key = Vec<u8>, Value = Vec<u8>>,
+    T: Table<Key = NodeHash, Value = Vec<u8>>,
 {
     pub fn new(db: Arc<Database>) -> Self {
         Self {
@@ -23,20 +23,20 @@ where
 
 impl<T> TrieDB for LibmdbxTrieDB<T>
 where
-    T: Table<Key = Vec<u8>, Value = Vec<u8>>,
+    T: Table<Key = NodeHash, Value = Vec<u8>>,
 {
-    fn get(&self, key: Vec<u8>) -> Result<Option<Vec<u8>>, TrieError> {
+    fn get(&self, key: NodeHash) -> Result<Option<Vec<u8>>, TrieError> {
         let txn = self.db.begin_read().map_err(TrieError::DbError)?;
         txn.get::<T>(key).map_err(TrieError::DbError)
     }
 
-    fn put(&self, key: Vec<u8>, value: Vec<u8>) -> Result<(), TrieError> {
+    fn put(&self, key: NodeHash, value: Vec<u8>) -> Result<(), TrieError> {
         let txn = self.db.begin_readwrite().map_err(TrieError::DbError)?;
         txn.upsert::<T>(key, value).map_err(TrieError::DbError)?;
         txn.commit().map_err(TrieError::DbError)
     }
 
-    fn put_batch(&self, key_values: Vec<(Vec<u8>, Vec<u8>)>) -> Result<(), TrieError> {
+    fn put_batch(&self, key_values: Vec<(NodeHash, Vec<u8>)>) -> Result<(), TrieError> {
         let txn = self.db.begin_readwrite().map_err(TrieError::DbError)?;
         for (key, value) in key_values {
             txn.upsert::<T>(key, value).map_err(TrieError::DbError)?;
@@ -49,6 +49,7 @@ where
 mod test {
     use super::LibmdbxTrieDB;
     use crate::trie_db::test_utils::libmdbx::{new_db, TestNodes};
+    use ethrex_trie::NodeHash;
     use ethrex_trie::Trie;
     use ethrex_trie::TrieDB;
     use libmdbx::{
@@ -62,24 +63,25 @@ mod test {
     fn simple_addition() {
         table!(
             /// NodeHash to Node table
-            ( Nodes )  Vec<u8> => Vec<u8>
+            ( Nodes )  NodeHash => Vec<u8>
         );
         let inner_db = new_db::<Nodes>();
+        let key = NodeHash::from_encoded_raw(b"hello");
         let db = LibmdbxTrieDB::<Nodes>::new(inner_db);
-        assert_eq!(db.get("hello".into()).unwrap(), None);
-        db.put("hello".into(), "value".into()).unwrap();
-        assert_eq!(db.get("hello".into()).unwrap(), Some("value".into()));
+        assert_eq!(db.get(key).unwrap(), None);
+        db.put(key, "value".into()).unwrap();
+        assert_eq!(db.get(key).unwrap(), Some("value".into()));
     }
 
     #[test]
     fn different_tables() {
         table!(
             /// vec to vec
-            ( TableA ) Vec<u8> => Vec<u8>
+            ( TableA ) NodeHash => Vec<u8>
         );
         table!(
             /// vec to vec
-            ( TableB ) Vec<u8> => Vec<u8>
+            ( TableB ) NodeHash => Vec<u8>
         );
         let tables = [table_info!(TableA), table_info!(TableB)]
             .into_iter()
@@ -88,8 +90,9 @@ mod test {
         let inner_db = Arc::new(Database::create(None, &tables).unwrap());
         let db_a = LibmdbxTrieDB::<TableA>::new(inner_db.clone());
         let db_b = LibmdbxTrieDB::<TableB>::new(inner_db.clone());
-        db_a.put("hello".into(), "value".into()).unwrap();
-        assert_eq!(db_b.get("hello".into()).unwrap(), None);
+        let key = NodeHash::from_encoded_raw(b"hello");
+        db_a.put(key, "value".into()).unwrap();
+        assert_eq!(db_b.get(key).unwrap(), None);
     }
 
     #[test]

crates/storage/trie_db/libmdbx_dupsort.rs

@@ -1,8 +1,8 @@
 use std::{marker::PhantomData, sync::Arc};
 
 use super::utils::node_hash_to_fixed_size;
-use ethrex_trie::error::TrieError;
 use ethrex_trie::TrieDB;
+use ethrex_trie::{error::TrieError, NodeHash};
 use libmdbx::orm::{Database, DupSort, Encodable};
 
 /// Libmdbx implementation for the TrieDB trait for a dupsort table with a fixed primary key.
@@ -37,13 +37,13 @@ where
     T: DupSort<Key = (SK, [u8; 33]), SeekKey = SK, Value = Vec<u8>>,
     SK: Clone + Encodable,
 {
-    fn get(&self, key: Vec<u8>) -> Result<Option<Vec<u8>>, TrieError> {
+    fn get(&self, key: NodeHash) -> Result<Option<Vec<u8>>, TrieError> {
         let txn = self.db.begin_read().map_err(TrieError::DbError)?;
         txn.get::<T>((self.fixed_key.clone(), node_hash_to_fixed_size(key)))
             .map_err(TrieError::DbError)
     }
 
-    fn put(&self, key: Vec<u8>, value: Vec<u8>) -> Result<(), TrieError> {
+    fn put(&self, key: NodeHash, value: Vec<u8>) -> Result<(), TrieError> {
         let txn = self.db.begin_readwrite().map_err(TrieError::DbError)?;
         txn.upsert::<T>(
             (self.fixed_key.clone(), node_hash_to_fixed_size(key)),
@@ -53,7 +53,7 @@ where
         txn.commit().map_err(TrieError::DbError)
     }
 
-    fn put_batch(&self, key_values: Vec<(Vec<u8>, Vec<u8>)>) -> Result<(), TrieError> {
+    fn put_batch(&self, key_values: Vec<(NodeHash, Vec<u8>)>) -> Result<(), TrieError> {
         let txn = self.db.begin_readwrite().map_err(TrieError::DbError)?;
         for (key, value) in key_values {
             txn.upsert::<T>(
@@ -82,19 +82,21 @@ mod test {
     fn simple_addition() {
         let inner_db = new_db::<Nodes>();
         let db = LibmdbxDupsortTrieDB::<Nodes, [u8; 32]>::new(inner_db, [5; 32]);
-        assert_eq!(db.get("hello".into()).unwrap(), None);
-        db.put("hello".into(), "value".into()).unwrap();
-        assert_eq!(db.get("hello".into()).unwrap(), Some("value".into()));
+        let key = NodeHash::from_encoded_raw(b"hello");
+        assert_eq!(db.get(key).unwrap(), None);
+        db.put(key, "value".into()).unwrap();
+        assert_eq!(db.get(key).unwrap(), Some("value".into()));
     }
 
     #[test]
     fn different_keys() {
         let inner_db = new_db::<Nodes>();
         let db_a = LibmdbxDupsortTrieDB::<Nodes, [u8; 32]>::new(inner_db.clone(), [5; 32]);
         let db_b = LibmdbxDupsortTrieDB::<Nodes, [u8; 32]>::new(inner_db, [7; 32]);
-        db_a.put("hello".into(), "hello!".into()).unwrap();
-        db_b.put("hello".into(), "go away!".into()).unwrap();
-        assert_eq!(db_a.get("hello".into()).unwrap(), Some("hello!".into()));
-        assert_eq!(db_b.get("hello".into()).unwrap(), Some("go away!".into()));
+        let key = NodeHash::from_encoded_raw(b"hello");
+        db_a.put(key, "hello!".into()).unwrap();
+        db_b.put(key, "go away!".into()).unwrap();
+        assert_eq!(db_a.get(key).unwrap(), Some("hello!".into()));
+        assert_eq!(db_b.get(key).unwrap(), Some("go away!".into()));
     }
 }