build: use new revm with analysis cache (#2527)

* build: use new revm with analysis cache

* refactor: use checked bytecode

See bluealloy/revm#121 (comment)

* build: use git revm

* build: use revm 1.8

* test: fix test

* fix: correct bytecode getters/setters

Whenever we output the bytecode of an account, whether to
a file or in a response, we need to return the *original*
bytecode using `Bytecode::len`, otherwise the bytecode
will differ depending on whether the bytecode has been
checked, analyzed or not.

* refactor: use `Bytecode::hash`

* fix: get original account code for traces

* refactor: remove unsafe code

anvil/src/eth/backend/db.rs

@@ -9,7 +9,7 @@ use ethers::{
 use forge::revm::KECCAK_EMPTY;
 use foundry_evm::{
     executor::DatabaseRef,
-    revm::{db::CacheDB, Database, DatabaseCommit, InMemoryDB},
+    revm::{db::CacheDB, Bytecode, Database, DatabaseCommit, InMemoryDB},
     HashMap,
 };
 
@@ -43,7 +43,7 @@ pub trait Db: DatabaseRef + Database + DatabaseCommit + Send + Sync {
             H256::from_slice(&keccak256(code.as_ref())[..])
         };
         info.code_hash = code_hash;
-        info.code = Some(code.to_vec().into());
+        info.code = Some(Bytecode::new_raw(code.0).to_checked());
         self.insert_account(address, info);
     }
 
@@ -123,7 +123,7 @@ impl DatabaseRef for StateDb {
         self.0.basic(address)
     }
 
-    fn code_by_hash(&self, code_hash: H256) -> bytes::Bytes {
+    fn code_by_hash(&self, code_hash: H256) -> Bytecode {
         self.0.code_by_hash(code_hash)
     }
 

anvil/src/eth/backend/mem/in_memory_db.rs

@@ -11,7 +11,7 @@ use tracing::{trace, warn};
 // reexport for convenience
 pub use foundry_evm::executor::{backend::MemDb, DatabaseRef};
 
-use forge::revm::KECCAK_EMPTY;
+use forge::revm::{Bytecode, KECCAK_EMPTY};
 
 impl Db for MemDb {
     fn insert_account(&mut self, address: Address, account: AccountInfo) {
@@ -29,16 +29,17 @@ impl Db for MemDb {
             .clone()
             .into_iter()
             .map(|(k, v)| {
+                let code = v
+                    .info
+                    .code
+                    .unwrap_or_else(|| self.inner.code_by_hash(v.info.code_hash))
+                    .to_checked();
                 (
                     k,
                     SerializableAccountRecord {
                         nonce: v.info.nonce,
                         balance: v.info.balance,
-                        code: v
-                            .info
-                            .code
-                            .unwrap_or_else(|| self.inner.code_by_hash(v.info.code_hash))
-                            .into(),
+                        code: code.bytes()[..code.len()].to_vec().into(),
                         storage: v.storage.into_iter().collect(),
                     },
                 )
@@ -57,7 +58,11 @@ impl Db for MemDb {
                 AccountInfo {
                     balance: account.balance,
                     code_hash: KECCAK_EMPTY, // will be set automatically
-                    code: if account.code.0.is_empty() { None } else { Some(account.code.0) },
+                    code: if account.code.0.is_empty() {
+                        None
+                    } else {
+                        Some(Bytecode::new_raw(account.code.0).to_checked())
+                    },
                     // use max nonce in case account is imported multiple times with difference
                     // nonces to prevent collisions
                     nonce: std::cmp::max(
@@ -110,7 +115,7 @@ mod tests {
         Address,
     };
     use bytes::Bytes;
-    use forge::revm::KECCAK_EMPTY;
+    use forge::revm::{Bytecode, KECCAK_EMPTY};
     use foundry_evm::{
         executor::{backend::MemDb, DatabaseRef},
         HashMap,
@@ -126,7 +131,8 @@ mod tests {
 
         let mut dump_db = MemDb::default();
 
-        let contract_code: Bytes = Bytes::from("fake contract code");
+        let contract_code: Bytecode =
+            Bytecode::new_raw(Bytes::from("fake contract code")).to_checked();
 
         dump_db.insert_account(
             test_addr,
@@ -163,7 +169,8 @@ mod tests {
         let test_addr2: Address =
             Address::from_str("0x70997970c51812dc3a010c7d01b50e0d17dc79c8").unwrap();
 
-        let contract_code: Bytes = Bytes::from("fake contract code");
+        let contract_code: Bytecode =
+            Bytecode::new_raw(Bytes::from("fake contract code")).to_checked();
 
         let mut db = MemDb::default();
 
@@ -199,7 +206,7 @@ mod tests {
             test_addr,
             SerializableAccountRecord {
                 balance: 100100.into(),
-                code: contract_code.clone().into(),
+                code: contract_code.bytes()[..contract_code.len()].to_vec().into(),
                 nonce: 100,
                 storage: new_storage,
             },

anvil/src/eth/backend/mem/mod.rs

@@ -1,5 +1,4 @@
 //! In memory blockchain backend
-
 use crate::{
     eth::{
         backend::{
@@ -1096,11 +1095,11 @@ impl Backend {
             trace!(target: "backend", "get code for {:?}", address);
             let account = db.basic(address);
             let code = if let Some(code) = account.code {
-                code.into()
+                code
             } else {
-                db.code_by_hash(account.code_hash).into()
+                db.code_by_hash(account.code_hash)
             };
-            Ok(code)
+            Ok(code.bytes()[..code.len()].to_vec().into())
         })
     }
 

evm/Cargo.toml

@@ -36,7 +36,7 @@ once_cell = "1.13"
 # EVM
 bytes = "1.1.0"
 hashbrown = { version = "0.12", features = ["serde"] }
-revm = { version = "1.7", default-features = false, features = ["std", "k256", "with-serde", "memory_limit"] }
+revm = { version = "1.8", default-features = false, features = ["std", "k256", "with-serde", "memory_limit"] }
 
 # Fuzzer
 proptest = "1.0.0"

evm/src/coverage/visitor.rs

@@ -481,7 +481,7 @@ impl Visitor {
 
             // We found a push, so we do some PC -> IC translation accounting, but we also check if
             // this push is coupled with the JUMPI we are interested in.
-            if opcode_infos[op as usize].is_push {
+            if opcode_infos[op as usize].is_push() {
                 let element = if let Some(element) = source_map.get(pc - cumulative_push_size) {
                     element
                 } else {

evm/src/executor/backend/fuzz.rs

@@ -5,12 +5,11 @@ use crate::{
     },
     Address,
 };
-use bytes::Bytes;
 use ethers::prelude::{H160, H256, U256};
 use hashbrown::HashMap as Map;
 use revm::{
-    db::DatabaseRef, Account, AccountInfo, Database, Env, Inspector, Log, Return, SubRoutine,
-    TransactOut,
+    db::DatabaseRef, Account, AccountInfo, Bytecode, Database, Env, Inspector, Log, Return,
+    SubRoutine, TransactOut,
 };
 use std::borrow::Cow;
 use tracing::trace;
@@ -142,7 +141,7 @@ impl<'a> DatabaseRef for FuzzBackendWrapper<'a> {
         DatabaseRef::basic(self.backend.as_ref(), address)
     }
 
-    fn code_by_hash(&self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&self, code_hash: H256) -> Bytecode {
         DatabaseRef::code_by_hash(self.backend.as_ref(), code_hash)
     }
 
@@ -159,7 +158,7 @@ impl<'a> Database for FuzzBackendWrapper<'a> {
     fn basic(&mut self, address: H160) -> AccountInfo {
         DatabaseRef::basic(self, address)
     }
-    fn code_by_hash(&mut self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&mut self, code_hash: H256) -> Bytecode {
         DatabaseRef::code_by_hash(self, code_hash)
     }
     fn storage(&mut self, address: H160, index: U256) -> U256 {

evm/src/executor/backend/in_memory_db.rs

@@ -1,9 +1,7 @@
 //! The in memory DB
-
-use bytes::Bytes;
 use ethers::prelude::{H160, H256, U256};
 use hashbrown::HashMap as Map;
-use revm::{db::DatabaseRef, Account, AccountInfo, Database, DatabaseCommit, InMemoryDB};
+use revm::{db::DatabaseRef, Account, AccountInfo, Bytecode, Database, DatabaseCommit, InMemoryDB};
 
 use crate::executor::snapshot::Snapshots;
 
@@ -27,7 +25,7 @@ impl DatabaseRef for MemDb {
         DatabaseRef::basic(&self.inner, address)
     }
 
-    fn code_by_hash(&self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&self, code_hash: H256) -> Bytecode {
         DatabaseRef::code_by_hash(&self.inner, code_hash)
     }
 
@@ -45,7 +43,7 @@ impl Database for MemDb {
         Database::basic(&mut self.inner, address)
     }
 
-    fn code_by_hash(&mut self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&mut self, code_hash: H256) -> Bytecode {
         Database::code_by_hash(&mut self.inner, code_hash)
     }
 

evm/src/executor/backend/mod.rs

@@ -2,16 +2,15 @@ use crate::executor::{
     fork::{CreateFork, ForkId, MultiFork, SharedBackend},
     snapshot::Snapshots,
 };
-use bytes::Bytes;
 use ethers::{
     prelude::{H160, H256, U256},
     types::Address,
 };
 use hashbrown::HashMap as Map;
 use revm::{
     db::{CacheDB, DatabaseRef},
-    Account, AccountInfo, Database, DatabaseCommit, Env, InMemoryDB, Inspector, Log, Return,
-    SubRoutine, TransactOut, TransactTo, KECCAK_EMPTY,
+    Account, AccountInfo, Bytecode, Database, DatabaseCommit, Env, InMemoryDB, Inspector, Log,
+    Return, SubRoutine, TransactOut, TransactTo, KECCAK_EMPTY,
 };
 use std::collections::{HashMap, HashSet};
 use tracing::{trace, warn};
@@ -702,7 +701,7 @@ impl DatabaseRef for Backend {
         }
     }
 
-    fn code_by_hash(&self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&self, code_hash: H256) -> Bytecode {
         if let Some(db) = self.active_fork_db() {
             db.code_by_hash(code_hash)
         } else {
@@ -736,7 +735,7 @@ impl<'a> DatabaseRef for &'a mut Backend {
         }
     }
 
-    fn code_by_hash(&self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&self, code_hash: H256) -> Bytecode {
         if let Some(db) = self.active_fork_db() {
             DatabaseRef::code_by_hash(db, code_hash)
         } else {
@@ -780,7 +779,7 @@ impl Database for Backend {
         }
     }
 
-    fn code_by_hash(&mut self, code_hash: H256) -> Bytes {
+    fn code_by_hash(&mut self, code_hash: H256) -> Bytecode {
         if let Some(db) = self.active_fork_db_mut() {
             db.code_by_hash(code_hash)
         } else {

evm/src/executor/fork/backend.rs

@@ -1,6 +1,4 @@
 //! Smart caching and deduplication of requests when using a forking provider
-use revm::{db::DatabaseRef, AccountInfo, KECCAK_EMPTY};
-
 use crate::executor::fork::{cache::FlushJsonBlockCacheDB, BlockchainDb};
 use ethers::{
     core::abi::ethereum_types::BigEndianHash,
@@ -14,6 +12,7 @@ use futures::{
     task::{Context, Poll},
     Future, FutureExt,
 };
+use revm::{db::DatabaseRef, AccountInfo, Bytecode, KECCAK_EMPTY};
 use std::{
     collections::{hash_map::Entry, HashMap, VecDeque},
     pin::Pin,
@@ -280,8 +279,12 @@ where
                             };
 
                             // update the cache
-                            let acc =
-                                AccountInfo { nonce: nonce.as_u64(), balance, code, code_hash };
+                            let acc = AccountInfo {
+                                nonce: nonce.as_u64(),
+                                balance,
+                                code: code.map(|bytes| Bytecode::new_raw(bytes).to_checked()),
+                                code_hash,
+                            };
                             pin.db.accounts().write().insert(addr, acc.clone());
 
                             // notify all listeners
@@ -495,7 +498,7 @@ impl DatabaseRef for SharedBackend {
         })
     }
 
-    fn code_by_hash(&self, _address: H256) -> bytes::Bytes {
+    fn code_by_hash(&self, _address: H256) -> Bytecode {
         panic!("Should not be called. Code is already loaded.")
     }
 

evm/src/executor/fork/cache.rs

@@ -1,8 +1,5 @@
 //! Cache related abstraction
-use ethers::{
-    types::{Address, H256, U256},
-    utils::keccak256,
-};
+use ethers::types::{Address, H256, U256};
 use parking_lot::RwLock;
 use revm::{Account, AccountInfo, DatabaseCommit, Filth, KECCAK_EMPTY};
 use serde::{ser::SerializeMap, Deserialize, Deserializer, Serialize, Serializer};
@@ -197,12 +194,8 @@ impl MemDb {
                 storage.remove(&add);
             } else {
                 // insert account
-                if let Some(code_hash) = acc
-                    .info
-                    .code
-                    .as_ref()
-                    .filter(|code| !code.is_empty())
-                    .map(|code| H256::from_slice(&keccak256(code)))
+                if let Some(code_hash) =
+                    acc.info.code.as_ref().filter(|code| !code.is_empty()).map(|code| code.hash())
                 {
                     acc.info.code_hash = code_hash;
                 } else if acc.info.code_hash.is_zero() {