feat(levm): implement Cache and refactor Db (#991)

**Motivation**

<!-- Why does this pull request exist? What are its goals? -->
We currently have an in-memory database that should soon be replaced by
the node's in-disk database. For that purpose we want to allow us to
switch between both kinds of databases. The need to implement this PR's
features and refactor the `Db` arose while working on #904.

**Description**

<!-- A clear and concise general description of the changes this PR
introduces -->
This PR includes:
- Adding a `Cache` to store warm accounts. This removes the need of
having `accessed_accounts` and `accessed_storage_slots` sets in
`Substate` because we know that if they are cached then they are warm.
- Making our `Db` implement the `Database` trait and interact with it
only using methods and not it's attributes, so in the future we can
implement that trait for the actual node's database.
- Fix call opcodes and remove delegate attribute from `CallFrame`.

<!-- Link to issues: Resolves #111, Resolves #222 -->

Part of #814.

---------

Co-authored-by: Juani Medone <[email protected]>
Co-authored-by: maximopalopoli <[email protected]>
Co-authored-by: Javier Chatruc <[email protected]>

crates/vm/levm/Cargo.toml

@@ -19,7 +19,3 @@ hex = "0.4.3"
 
 [features]
 ethereum_foundation_tests = []
-
-[profile.test]
-opt-level = 3
-debug-assertions = true

crates/vm/levm/docs/substate.md

@@ -0,0 +1,3 @@
+## Substate
+
+`accessed_addresses` and `accessed_storage_keys` follow the structure defined in [EIP 2929](https://eips.ethereum.org/EIPS/eip-2929#specification)

crates/vm/levm/src/call_frame.rs

@@ -62,10 +62,13 @@ pub struct CallFrame {
     pub gas_limit: U256,
     pub gas_used: U256,
     pub pc: usize,
-    pub msg_sender: Address, // Origin address?
+    /// Address of the account that sent the message
+    pub msg_sender: Address,
+    /// Address of the recipient of the message
     pub to: Address,
+    /// Address of the code to execute. Usually the same as `to`, but can be different
     pub code_address: Address,
-    pub delegate: Option<Address>,
+    /// Bytecode to execute
     pub bytecode: Bytes,
     pub msg_value: U256,
     pub stack: Stack, // max 1024 in the future
@@ -98,7 +101,6 @@ impl CallFrame {
         msg_sender: Address,
         to: Address,
         code_address: Address,
-        delegate: Option<Address>,
         bytecode: Bytes,
         msg_value: U256,
         calldata: Bytes,
@@ -112,7 +114,6 @@ impl CallFrame {
             msg_sender,
             to,
             code_address,
-            delegate,
             bytecode,
             msg_value,
             calldata,

crates/vm/levm/src/constants.rs

@@ -47,7 +47,6 @@ pub mod gas_cost {
     pub const RETURNDATACOPY_STATIC: U256 = U256([3, 0, 0, 0]);
     pub const RETURNDATACOPY_DYNAMIC_BASE: U256 = U256([3, 0, 0, 0]);
     pub const ADDRESS: U256 = U256([2, 0, 0, 0]);
-    pub const BALANCE: U256 = U256([100, 0, 0, 0]);
     pub const ORIGIN: U256 = U256([2, 0, 0, 0]);
     pub const CALLER: U256 = U256([2, 0, 0, 0]);
     pub const BLOCKHASH: U256 = U256([20, 0, 0, 0]);

crates/vm/levm/src/db.rs

@@ -0,0 +1,126 @@
+use crate::vm::{Account, AccountInfo, StorageSlot};
+use ethereum_types::{Address, U256};
+use keccak_hash::H256;
+use std::collections::HashMap;
+
+pub trait Database {
+    fn get_account_info(&self, address: Address) -> AccountInfo;
+    fn get_storage_slot(&self, address: Address, key: H256) -> U256;
+    fn get_block_hash(&self, block_number: u64) -> Option<H256>;
+}
+
+#[derive(Debug, Default)]
+pub struct Db {
+    pub accounts: HashMap<Address, Account>,
+    pub block_hashes: HashMap<u64, H256>,
+}
+
+// Methods here are for testing purposes only, for initializing the Db with some values
+impl Db {
+    pub fn new() -> Self {
+        Self {
+            accounts: HashMap::new(),
+            block_hashes: HashMap::new(),
+        }
+    }
+
+    /// Add accounts to database
+    pub fn add_accounts(&mut self, accounts: Vec<(Address, Account)>) {
+        self.accounts.extend(accounts);
+    }
+
+    /// Add block hashes to database
+    pub fn add_block_hashes(&mut self, block_hashes: Vec<(u64, H256)>) {
+        self.block_hashes.extend(block_hashes);
+    }
+
+    /// Builder method with accounts [for testing only]
+    pub fn with_accounts(mut self, accounts: HashMap<Address, Account>) -> Self {
+        self.accounts = accounts;
+        self
+    }
+
+    /// Builder method with block hashes [for testing only]
+    pub fn with_block_hashes(mut self, block_hashes: HashMap<u64, H256>) -> Self {
+        self.block_hashes = block_hashes;
+        self
+    }
+}
+
+impl Database for Db {
+    fn get_account_info(&self, address: Address) -> AccountInfo {
+        self.accounts
+            .get(&address)
+            .unwrap_or(&Account::default())
+            .info
+            .clone()
+    }
+
+    fn get_storage_slot(&self, address: Address, key: H256) -> U256 {
+        // both `original_value` and `current_value` should work here because they have the same values on Db
+        self.accounts
+            .get(&address)
+            .unwrap_or(&Account::default())
+            .storage
+            .get(&key)
+            .unwrap_or(&StorageSlot::default())
+            .original_value
+    }
+
+    fn get_block_hash(&self, block_number: u64) -> Option<H256> {
+        self.block_hashes.get(&block_number).cloned()
+    }
+}
+
+#[derive(Debug, Default, Clone)]
+pub struct Cache {
+    pub accounts: HashMap<Address, Account>,
+}
+
+impl Cache {
+    pub fn get_account(&self, address: Address) -> Option<&Account> {
+        self.accounts.get(&address)
+    }
+
+    pub fn get_mut_account(&mut self, address: Address) -> Option<&mut Account> {
+        self.accounts.get_mut(&address)
+    }
+
+    pub fn get_storage_slot(&self, address: Address, key: H256) -> Option<StorageSlot> {
+        self.get_account(address)
+            .expect("Account should have been cached")
+            .storage
+            .get(&key)
+            .cloned()
+    }
+
+    pub fn add_account(&mut self, address: &Address, account: &Account) {
+        self.accounts.insert(*address, account.clone());
+    }
+
+    pub fn write_account_storage(&mut self, address: &Address, key: H256, slot: StorageSlot) {
+        self.accounts
+            .get_mut(address)
+            .expect("Account should have been cached")
+            .storage
+            .insert(key, slot);
+    }
+
+    pub fn increment_account_nonce(&mut self, address: &Address) {
+        if let Some(account) = self.accounts.get_mut(address) {
+            account.info.nonce += 1;
+        }
+    }
+
+    pub fn is_account_cached(&self, address: &Address) -> bool {
+        self.accounts.contains_key(address)
+    }
+
+    pub fn is_slot_cached(&self, address: &Address, key: H256) -> bool {
+        self.is_account_cached(address)
+            && self
+                .get_account(*address)
+                .map(|account| account.storage.contains_key(&key))
+                .unwrap_or(false)
+    }
+}

crates/vm/levm/src/lib.rs

@@ -1,6 +1,7 @@
 pub mod block;
 pub mod call_frame;
 pub mod constants;
+pub mod db;
 pub mod errors;
 pub mod memory;
 pub mod opcode_handlers;

crates/vm/levm/src/opcode_handlers/block.rs

@@ -30,10 +30,12 @@ impl VM {
             return Ok(OpcodeSuccess::Continue);
         }
 
-        if let Some(block_hash) = self.db.block_hashes.get(&block_number) {
+        let block_number = block_number.as_u64();
+
+        if let Some(block_hash) = self.db.get_block_hash(block_number) {
             current_call_frame
                 .stack
-                .push(U256::from_big_endian(&block_hash.0))?;
+                .push(U256::from_big_endian(block_hash.as_bytes()))?;
         } else {
             current_call_frame.stack.push(U256::zero())?;
         }
@@ -125,9 +127,15 @@ impl VM {
     ) -> Result<OpcodeSuccess, VMError> {
         self.increase_consumed_gas(current_call_frame, gas_cost::SELFBALANCE)?;
 
-        let balance = self.db.balance(&current_call_frame.code_address);
-        current_call_frame.stack.push(balance)?;
+        // the current account should have been cached when the contract was called
+        let balance = self
+            .cache
+            .get_account(current_call_frame.code_address)
+            .expect("The current account should always be cached")
+            .info
+            .balance;
 
+        current_call_frame.stack.push(balance)?;
         Ok(OpcodeSuccess::Continue)
     }
 

crates/vm/levm/src/opcode_handlers/environment.rs

@@ -1,6 +1,9 @@
 use super::*;
 use crate::{
-    constants::{call_opcode, WORD_SIZE},
+    constants::{
+        call_opcode::{COLD_ADDRESS_ACCESS_COST, WARM_ADDRESS_ACCESS_COST},
+        WORD_SIZE,
+    },
     vm::word_to_address,
 };
 use sha3::{Digest, Keccak256};
@@ -16,11 +19,7 @@ impl VM {
     ) -> Result<OpcodeSuccess, VMError> {
         self.increase_consumed_gas(current_call_frame, gas_cost::ADDRESS)?;
 
-        let addr = if current_call_frame.delegate.is_some() {
-            current_call_frame.msg_sender
-        } else {
-            current_call_frame.code_address
-        };
+        let addr = current_call_frame.to; // The recipient of the current call.
 
         current_call_frame.stack.push(U256::from(addr.as_bytes()))?;
 
@@ -32,13 +31,18 @@ impl VM {
         &mut self,
         current_call_frame: &mut CallFrame,
     ) -> Result<OpcodeSuccess, VMError> {
-        self.increase_consumed_gas(current_call_frame, gas_cost::BALANCE)?;
+        let address = &word_to_address(current_call_frame.stack.pop()?);
 
-        let addr = current_call_frame.stack.pop()?;
+        if self.cache.is_account_cached(address) {
+            self.increase_consumed_gas(current_call_frame, WARM_ADDRESS_ACCESS_COST)?;
+        } else {
+            self.increase_consumed_gas(current_call_frame, COLD_ADDRESS_ACCESS_COST)?;
+            self.cache_from_db(address);
+        };
 
-        let balance = self.db.balance(&word_to_address(addr));
-        current_call_frame.stack.push(balance)?;
+        let balance = self.cache.get_account(*address).unwrap().info.balance;
 
+        current_call_frame.stack.push(balance)?;
         Ok(OpcodeSuccess::Continue)
     }
 
@@ -237,17 +241,23 @@ impl VM {
         current_call_frame: &mut CallFrame,
     ) -> Result<OpcodeSuccess, VMError> {
         let address = word_to_address(current_call_frame.stack.pop()?);
-        let gas_cost = if self.accrued_substate.warm_addresses.contains(&address) {
-            call_opcode::WARM_ADDRESS_ACCESS_COST
+
+        if self.cache.is_account_cached(&address) {
+            self.increase_consumed_gas(current_call_frame, WARM_ADDRESS_ACCESS_COST)?;
         } else {
-            call_opcode::COLD_ADDRESS_ACCESS_COST
+            self.increase_consumed_gas(current_call_frame, COLD_ADDRESS_ACCESS_COST)?;
+            self.cache_from_db(&address);
         };
 
-        self.increase_consumed_gas(current_call_frame, gas_cost)?;
-
-        let code_size = self.db.get_account_bytecode(&address).len();
-        current_call_frame.stack.push(code_size.into())?;
+        let bytecode = self
+            .cache
+            .get_account(address)
+            .unwrap()
+            .info
+            .bytecode
+            .clone();
 
+        current_call_frame.stack.push(bytecode.len().into())?;
         Ok(OpcodeSuccess::Continue)
     }
 
@@ -277,26 +287,32 @@ impl VM {
         let memory_expansion_cost = current_call_frame
             .memory
             .expansion_cost(dest_offset + size)?;
-        let address_access_cost = if self.accrued_substate.warm_addresses.contains(&address) {
-            call_opcode::WARM_ADDRESS_ACCESS_COST
+        let gas_cost =
+            gas_cost::EXTCODECOPY_DYNAMIC_BASE * minimum_word_size + memory_expansion_cost;
+
+        if self.cache.is_account_cached(&address) {
+            self.increase_consumed_gas(current_call_frame, gas_cost + WARM_ADDRESS_ACCESS_COST)?;
         } else {
-            call_opcode::COLD_ADDRESS_ACCESS_COST
+            self.increase_consumed_gas(current_call_frame, gas_cost + COLD_ADDRESS_ACCESS_COST)?;
+            self.cache_from_db(&address);
         };
-        let gas_cost = gas_cost::EXTCODECOPY_DYNAMIC_BASE * minimum_word_size
-            + memory_expansion_cost
-            + address_access_cost;
 
-        self.increase_consumed_gas(current_call_frame, gas_cost)?;
+        let mut bytecode = self
+            .cache
+            .get_account(address)
+            .unwrap()
+            .info
+            .bytecode
+            .clone();
 
-        let mut code = self.db.get_account_bytecode(&address);
-        if code.len() < offset + size {
-            let mut extended_code = code.to_vec();
+        if bytecode.len() < offset + size {
+            let mut extended_code = bytecode.to_vec();
             extended_code.resize(offset + size, 0);
-            code = Bytes::from(extended_code);
+            bytecode = Bytes::from(extended_code);
         }
         current_call_frame
             .memory
-            .store_bytes(dest_offset, &code[offset..offset + size]);
+            .store_bytes(dest_offset, &bytecode[offset..offset + size]);
 
         Ok(OpcodeSuccess::Continue)
     }
@@ -364,17 +380,24 @@ impl VM {
         current_call_frame: &mut CallFrame,
     ) -> Result<OpcodeSuccess, VMError> {
         let address = word_to_address(current_call_frame.stack.pop()?);
-        let gas_cost = if self.accrued_substate.warm_addresses.contains(&address) {
-            call_opcode::WARM_ADDRESS_ACCESS_COST
+
+        if self.cache.is_account_cached(&address) {
+            self.increase_consumed_gas(current_call_frame, WARM_ADDRESS_ACCESS_COST)?;
         } else {
-            call_opcode::COLD_ADDRESS_ACCESS_COST
+            self.increase_consumed_gas(current_call_frame, COLD_ADDRESS_ACCESS_COST)?;
+            self.cache_from_db(&address);
         };
 
-        self.increase_consumed_gas(current_call_frame, gas_cost)?;
+        let bytecode = self
+            .cache
+            .get_account(address)
+            .unwrap()
+            .info
+            .bytecode
+            .clone();
 
-        let code = self.db.get_account_bytecode(&address);
         let mut hasher = Keccak256::new();
-        hasher.update(code);
+        hasher.update(bytecode);
         let result = hasher.finalize();
         current_call_frame
             .stack