feat(ef-tests): validate state root (#4995)

README.md

@@ -100,7 +100,7 @@ cargo test --workspace --features geth-tests
 # Note: Requires cloning https://github.com/ethereum/tests
 # 
 #   cd testing/ef-tests && git clone https://github.com/ethereum/tests ethereum-tests
-cargo test --workspace --features ef-tests
+cargo test -p ef-tests --features ef-tests
 ```
 
 We recommend using [`cargo nextest`](https://nexte.st/) to speed up testing. With nextest installed, simply substitute `cargo test` with `cargo nextest run`.

testing/ef-tests/src/cases/blockchain_test.rs

@@ -1,13 +1,13 @@
 //! Test runners for `BlockchainTests` in <https://github.com/ethereum/tests>
 
 use crate::{
-    models::{BlockchainTest, ForkSpec, RootOrState},
+    models::{BlockchainTest, ForkSpec},
     Case, Error, Suite,
 };
 use alloy_rlp::Decodable;
 use reth_db::test_utils::create_test_rw_db;
 use reth_primitives::{BlockBody, SealedBlock};
-use reth_provider::{BlockWriter, ProviderFactory};
+use reth_provider::{BlockWriter, HashingWriter, ProviderFactory};
 use reth_stages::{stages::ExecutionStage, ExecInput, Stage};
 use std::{collections::BTreeMap, fs, path::Path, sync::Arc};
 
@@ -88,8 +88,8 @@ impl Case for BlockchainTestCase {
             let mut last_block = None;
             for block in case.blocks.iter() {
                 let decoded = SealedBlock::decode(&mut block.rlp.as_ref())?;
-                last_block = Some(decoded.number);
+                provider.insert_block(decoded.clone(), None, None)?;
-                provider.insert_block(decoded, None, None)?;
+                last_block = Some(decoded);
             }
 
             // Call execution stage
@@ -98,31 +98,36 @@ impl Case for BlockchainTestCase {
                     Arc::new(case.network.clone().into()),
                 ));
 
+                let target = last_block.as_ref().map(|b| b.number);
                 tokio::runtime::Builder::new_current_thread()
                     .build()
                     .expect("Could not build tokio RT")
                     .block_on(async {
                         // ignore error
-                        let _ = stage
+                        let _ =
-                            .execute(&provider, ExecInput { target: last_block, checkpoint: None })
+                            stage.execute(&provider, ExecInput { target, checkpoint: None }).await;
-                            .await;
                     });
             }
 
             // Validate post state
-            match &case.post_state {
+            if let Some(state) = &case.post_state {
-                Some(RootOrState::Root(root)) => {
-                    // TODO: We should really check the state root here...
-                    println!("Post-state root: #{root:?}")
-                }
-                Some(RootOrState::State(state)) => {
                 for (&address, account) in state.iter() {
                     account.assert_db(address, provider.tx_ref())?;
                 }
-                }
+            } else if let Some(expected_state_root) = case.post_state_hash {
-                None => println!("No post-state"),
+                // `insert_hashes` will insert hashed data, compute the state root and match it to
+                // expected internally
+                let last_block = last_block.unwrap_or_default();
+                provider.insert_hashes(
+                    0..=last_block.number,
+                    last_block.hash,
+                    expected_state_root,
+                )?;
+            } else {
+                return Err(Error::MissingPostState)
             }
 
+            // Drop provider without committing to the database.
             drop(provider);
         }
         Ok(())

testing/ef-tests/src/models.rs

@@ -26,7 +26,9 @@ pub struct BlockchainTest {
     /// Block data.
     pub blocks: Vec<Block>,
     /// The expected post state.
-    pub post_state: Option<RootOrState>,
+    pub post_state: Option<BTreeMap<Address, Account>>,
+    /// The expected post state merkle root.
+    pub post_state_hash: Option<B256>,
     /// The test pre-state.
     pub pre: State,
     /// Hash of the best block.
@@ -153,23 +155,28 @@ impl State {
         Tx: DbTxMut<'a>,
     {
         for (&address, account) in self.0.iter() {
+            let hashed_address = keccak256(address);
             let has_code = !account.code.is_empty();
             let code_hash = has_code.then(|| keccak256(&account.code));
-            tx.put::<tables::PlainAccountState>(
+            let reth_account = RethAccount {
-                address,
-                RethAccount {
                 balance: account.balance.0,
                 nonce: account.nonce.0.to::<u64>(),
                 bytecode_hash: code_hash,
-                },
+            };
-            )?;
+            tx.put::<tables::PlainAccountState>(address, reth_account)?;
+            tx.put::<tables::HashedAccount>(hashed_address, reth_account)?;
             if let Some(code_hash) = code_hash {
                 tx.put::<tables::Bytecodes>(code_hash, Bytecode::new_raw(account.code.clone()))?;
             }
-            account.storage.iter().try_for_each(|(k, v)| {
+            account.storage.iter().filter(|(_, v)| v.0 != U256::ZERO).try_for_each(|(k, v)| {
+                let storage_key = B256::from_slice(&k.0.to_be_bytes::<32>());
                 tx.put::<tables::PlainStorageState>(
                     address,
-                    StorageEntry { key: B256::from_slice(&k.0.to_be_bytes::<32>()), value: v.0 },
+                    StorageEntry { key: storage_key, value: v.0 },
+                )?;
+                tx.put::<tables::HashedStorage>(
+                    hashed_address,
+                    StorageEntry { key: keccak256(storage_key), value: v.0 },
                 )
             })?;
         }
@@ -186,16 +193,6 @@ impl Deref for State {
     }
 }
 
-/// Merkle root hash or storage accounts.
-#[derive(Clone, Debug, PartialEq, Eq, Deserialize)]
-#[serde(untagged)]
-pub enum RootOrState {
-    /// If state is too big, only state root is present
-    Root(B256),
-    /// State
-    State(BTreeMap<Address, Account>),
-}
-
 /// An account.
 #[derive(Debug, PartialEq, Eq, Deserialize, Clone)]
 #[serde(deny_unknown_fields)]

testing/ef-tests/src/result.rs

@@ -17,6 +17,9 @@ pub enum Error {
     /// The test was skipped
     #[error("Test was skipped")]
     Skipped,
+    /// No post state found in test
+    #[error("No post state found for validation")]
+    MissingPostState,
     /// An IO error occurred
     #[error("An error occurred interacting with the file system at {path}: {error}")]
     Io {