EVM: return just the bytecode hash (#1015)

* EVM: return just the bytecode hash

Previously, we returned a multihash. However, this got serialized as a map.

Now, we just return the raw digest encoded as 32 bytes.

* chore: state size changes

We've shaved quite a few bytes off by correctly serializing this.

* test_vm: fix hash method

We need to return the digest, not the encoded multihash. This only
affects the tests.

actors/evm/src/interpreter/instructions/ext.rs

@@ -1,29 +1,19 @@
 use crate::interpreter::instructions::memory::copy_to_memory;
 use crate::interpreter::{address::EthAddress, precompiles::Precompiles};
-use crate::U256;
+use crate::{BytecodeHash, U256};
 use cid::Cid;
 use fil_actors_runtime::deserialize_block;
 use fil_actors_runtime::runtime::builtins::Type;
 use fil_actors_runtime::ActorError;
 use fvm_ipld_blockstore::Blockstore;
-use fvm_shared::crypto::hash::SupportedHashes;
 use fvm_shared::sys::SendFlags;
 use fvm_shared::{address::Address, econ::TokenAmount};
-use multihash::Multihash;
 use num_traits::Zero;
 use {
     crate::interpreter::{ExecutionState, StatusCode, System},
     fil_actors_runtime::runtime::Runtime,
 };
 
-/// Keccak256 hash of `[0xfe]`, "native bytecode"
-pub const NATIVE_BYTECODE_HASH: [u8; 32] =
-    hex_literal::hex!("bcc90f2d6dada5b18e155c17a1c0a55920aae94f39857d39d0d8ed07ae8f228b");
-
-/// Keccak256 hash of `[]`, empty bytecode
-pub const EMPTY_EVM_HASH: [u8; 32] =
-    hex_literal::hex!("c5d2460186f7233c927e7db2dcc703c0e500b653ca82273b7bfad8045d85a470");
-
 pub fn extcodesize(
     _state: &mut ExecutionState,
     system: &mut System<impl Runtime>,
@@ -52,34 +42,28 @@ pub fn extcodehash(
     let addr = match get_contract_type(system.rt, addr) {
         ContractType::EVM(a) => a,
         // _Technically_ since we have native "bytecode" set as 0xfe this is valid, though we cant differentiate between different native actors.
-        ContractType::Native(_) => return Ok(NATIVE_BYTECODE_HASH.into()),
+        ContractType::Native(_) => return Ok(BytecodeHash::NATIVE_ACTOR.into()),
         // Precompiles "exist" and therefore aren't empty (although spec-wise they can be either 0 or keccak("") ).
-        ContractType::Precompile => return Ok(EMPTY_EVM_HASH.into()),
+        ContractType::Precompile => return Ok(BytecodeHash::EMPTY.into()),
         // NOTE: There may be accounts that in EVM would be considered "empty" (as defined in EIP-161) and give 0, but we will instead return keccak("").
         //      The FVM does not have chain state cleanup so contracts will never end up "empty" and be removed, they will either exist (in any state in the contract lifecycle)
         //      and return keccak(""), or not exist (where nothing has ever been deployed at that address) and return 0.
         // TODO: With account abstraction, this may be something other than an empty hash!
-        ContractType::Account => return Ok(EMPTY_EVM_HASH.into()),
+        ContractType::Account => return Ok(BytecodeHash::EMPTY.into()),
         // Not found
         ContractType::NotFound => return Ok(U256::zero()),
     };
 
     // multihash { keccak256(bytecode) }
-    let bytecode_hash: Multihash = deserialize_block(system.send(
+    let bytecode_hash: BytecodeHash = deserialize_block(system.send(
         &addr,
         crate::Method::GetBytecodeHash as u64,
         Default::default(),
         TokenAmount::zero(),
         None,
         SendFlags::READ_ONLY,
     )?)?;
-
-    let digest = bytecode_hash.digest();
-    debug_assert_eq!(SupportedHashes::Keccak256 as u64, bytecode_hash.code());
-
-    // Take the first 32 bytes of the Multihash
-    let digest_len = digest.len().min(32);
-    Ok(digest[..digest_len].into())
+    Ok(bytecode_hash.into())
 }
 
 pub fn extcodecopy(

actors/evm/src/interpreter/system.rs

@@ -15,10 +15,11 @@ use fvm_shared::{
     sys::SendFlags,
     MethodNum, IPLD_RAW,
 };
-use multihash::{Code, Multihash};
+use multihash::Code;
 use once_cell::unsync::OnceCell;
 
 use crate::state::{State, Tombstone};
+use crate::BytecodeHash;
 
 use super::{address::EthAddress, Bytecode};
 
@@ -84,11 +85,11 @@ pub struct EvmBytecode {
     /// CID of the contract
     pub cid: Cid,
     /// Keccak256 hash of the contract
-    pub evm_hash: Multihash,
+    pub evm_hash: BytecodeHash,
 }
 
 impl EvmBytecode {
-    fn new(cid: Cid, evm_hash: Multihash) -> Self {
+    fn new(cid: Cid, evm_hash: BytecodeHash) -> Self {
         Self { cid, evm_hash }
     }
 }
@@ -318,11 +319,9 @@ impl<'r, RT: Runtime> System<'r, RT> {
             ));
         }
 
-        let code_hash = multihash::Multihash::wrap(
-            SupportedHashes::Keccak256 as u64,
-            &self.rt.hash(SupportedHashes::Keccak256, bytecode),
-        )
-        .context_code(ExitCode::USR_ILLEGAL_STATE, "failed to hash bytecode with keccak")?;
+        let code_hash = self.rt.hash(SupportedHashes::Keccak256, bytecode)[..]
+            .try_into()
+            .context_code(ExitCode::USR_ASSERTION_FAILED, "expected a 32byte digest")?;
 
         let cid = self
             .rt

actors/evm/src/interpreter/uints.rs

@@ -6,6 +6,8 @@
 use serde::{Deserialize, Serialize};
 use substrate_bn::arith;
 
+use crate::BytecodeHash;
+
 use {
     fvm_shared::bigint::BigInt, fvm_shared::econ::TokenAmount, std::cmp::Ordering, std::fmt,
     uint::construct_uint,
@@ -109,6 +111,13 @@ impl From<&U256> for TokenAmount {
     }
 }
 
+impl From<BytecodeHash> for U256 {
+    fn from(bytecode: BytecodeHash) -> Self {
+        let bytes: [u8; 32] = bytecode.into();
+        Self::from(bytes)
+    }
+}
+
 impl Serialize for U256 {
     fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
     where

actors/evm/src/lib.rs

@@ -2,11 +2,8 @@ use fil_actors_runtime::{actor_error, AsActorError, EAM_ACTOR_ADDR, INIT_ACTOR_A
 use fvm_ipld_encoding::ipld_block::IpldBlock;
 use fvm_ipld_encoding::{strict_bytes, BytesDe, BytesSer, DAG_CBOR};
 use fvm_shared::address::Address;
-use fvm_shared::crypto::hash::SupportedHashes;
 use fvm_shared::error::ExitCode;
-use interpreter::instructions::ext::EMPTY_EVM_HASH;
 use interpreter::{address::EthAddress, system::load_bytecode};
-use multihash::Multihash;
 
 use crate::interpreter::output::Outcome;
 
@@ -240,14 +237,14 @@ impl EvmContractActor {
         }
     }
 
-    pub fn bytecode_hash(rt: &mut impl Runtime) -> Result<multihash::Multihash, ActorError> {
+    pub fn bytecode_hash(rt: &mut impl Runtime) -> Result<BytecodeHash, ActorError> {
         // Any caller can fetch the bytecode hash of a contract; this is where EXTCODEHASH gets it's value for EVM contracts.
         rt.validate_immediate_caller_accept_any()?;
 
         // return value must be either keccak("") or keccak(bytecode)
         let state: State = rt.state()?;
         if is_dead(rt, &state) {
-            Ok(Multihash::wrap(SupportedHashes::Keccak256 as u64, &EMPTY_EVM_HASH).unwrap())
+            Ok(BytecodeHash::EMPTY)
         } else {
             Ok(state.bytecode_hash)
         }
@@ -340,8 +337,8 @@ impl ActorCode for EvmContractActor {
                 Ok(IpldBlock::serialize_cbor(&ret)?)
             }
             Some(Method::GetBytecodeHash) => {
-                let multihash = Self::bytecode_hash(rt)?;
-                Ok(IpldBlock::serialize_cbor(&multihash)?)
+                let hash = Self::bytecode_hash(rt)?;
+                Ok(IpldBlock::serialize_cbor(&hash)?)
             }
             Some(Method::GetStorageAt) => {
                 let value = Self::storage_at(

actors/evm/src/state.rs

@@ -1,8 +1,12 @@
+use std::array::TryFromSliceError;
+
 use fvm_shared::ActorID;
 
 use {
     cid::Cid,
+    fvm_ipld_encoding::strict_bytes,
     fvm_ipld_encoding::tuple::*,
+    serde::{Deserialize, Serialize},
     serde_tuple::{Deserialize_tuple, Serialize_tuple},
 };
 
@@ -15,6 +19,71 @@ pub struct Tombstone {
     pub nonce: u64,
 }
 
+/// A Keccak256 digest of EVM bytecode.
+#[derive(Deserialize, Serialize, Clone, Copy, Eq, PartialEq)]
+#[serde(transparent)]
+pub struct BytecodeHash(#[serde(with = "strict_bytes")] [u8; 32]);
+
+impl std::fmt::Debug for BytecodeHash {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_tuple("BytecodeHash").field(&format_args!("{}", self)).finish()
+    }
+}
+
+impl std::fmt::Display for BytecodeHash {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        if f.alternate() {
+            write!(f, "0x")?;
+        }
+        for b in self.0 {
+            write!(f, "{b:02X}")?;
+        }
+        Ok(())
+    }
+}
+
+impl BytecodeHash {
+    pub const ZERO: Self = Self([0; 32]);
+
+    /// Keccak256 hash of `[0xfe]`, "native bytecode"
+    pub const NATIVE_ACTOR: Self =
+        Self(hex_literal::hex!("bcc90f2d6dada5b18e155c17a1c0a55920aae94f39857d39d0d8ed07ae8f228b"));
+
+    /// Keccak256 hash of `[]`, empty bytecode
+    pub const EMPTY: Self =
+        Self(hex_literal::hex!("c5d2460186f7233c927e7db2dcc703c0e500b653ca82273b7bfad8045d85a470"));
+
+    pub fn as_slice(&self) -> &[u8] {
+        &self.0
+    }
+}
+
+impl From<[u8; 32]> for BytecodeHash {
+    fn from(digest: [u8; 32]) -> Self {
+        BytecodeHash(digest)
+    }
+}
+
+impl From<BytecodeHash> for [u8; 32] {
+    fn from(digest: BytecodeHash) -> Self {
+        digest.0
+    }
+}
+
+impl From<BytecodeHash> for Vec<u8> {
+    fn from(digest: BytecodeHash) -> Self {
+        digest.0.into()
+    }
+}
+
+impl TryFrom<&[u8]> for BytecodeHash {
+    type Error = TryFromSliceError;
+
+    fn try_from(value: &[u8]) -> Result<Self, TryFromSliceError> {
+        Ok(Self(value.try_into()?))
+    }
+}
+
 /// Data stored by an EVM contract.
 /// This runs on the fvm-evm-runtime actor code cid.
 #[derive(Debug, Serialize_tuple, Deserialize_tuple)]
@@ -24,7 +93,7 @@ pub struct State {
     pub bytecode: Cid,
 
     /// The EVM contract bytecode hash keccak256(bytecode)
-    pub bytecode_hash: multihash::Multihash,
+    pub bytecode_hash: BytecodeHash,
 
     /// The EVM contract state dictionary.
     /// All eth contract state is a map of U256 -> U256 values.
@@ -54,3 +123,33 @@ pub struct State {
     /// See https://github.com/filecoin-project/ref-fvm/issues/1174 for some context.
     pub tombstone: Option<Tombstone>,
 }
+
+#[cfg(test)]
+mod test {
+    use fvm_ipld_encoding::{from_slice, to_vec, BytesDe};
+
+    use crate::BytecodeHash;
+    #[test]
+    fn test_bytecode_hash_serde() {
+        let encoded = to_vec(&BytecodeHash::EMPTY).unwrap();
+        let BytesDe(decoded) = from_slice(&encoded).unwrap();
+        assert_eq!(BytecodeHash::try_from(&decoded[..]).unwrap(), BytecodeHash::EMPTY);
+    }
+
+    #[test]
+    fn test_bytecode_hash_format() {
+        assert_eq!(
+            BytecodeHash::ZERO.to_string(),
+            "0000000000000000000000000000000000000000000000000000000000000000"
+        );
+        assert_eq!(
+            format!("{:#}", BytecodeHash::ZERO),
+            "0x0000000000000000000000000000000000000000000000000000000000000000"
+        );
+
+        assert_eq!(
+            format!("{:?}", BytecodeHash::ZERO),
+            "BytecodeHash(0000000000000000000000000000000000000000000000000000000000000000)"
+        );
+    }
+}