feat(EOF): Bytecode::new_raw supports EOF, new_raw_checked added (#1607)

* feat(EOF): Bytecode::new_raw supports EOF, new_raw_checked added

* Use from abbreviation

bins/revme/src/cmd/evmrunner.rs

@@ -2,7 +2,7 @@ use revm::{
     db::BenchmarkDB,
     inspector_handle_register,
     inspectors::TracerEip3155,
-    primitives::{Address, Bytecode, TxKind},
+    primitives::{eof::EofDecodeError, Address, Bytecode, TxKind},
     Evm,
 };
 use std::io::Error as IoError;
@@ -24,13 +24,9 @@ pub enum Errors {
     #[error("EVM Error")]
     EVMError,
     #[error(transparent)]
-    Io(IoError),
-}
-
-impl From<IoError> for Errors {
-    fn from(e: IoError) -> Self {
-        Errors::Io(e)
-    }
+    Io(#[from] IoError),
+    #[error(transparent)]
+    EofError(#[from] EofDecodeError),
 }
 
 /// Evm runner command allows running arbitrary evm bytecode.
@@ -59,7 +55,7 @@ pub struct Cmd {
 }
 
 impl Cmd {
-    /// Run statetest command.
+    /// Run evm runner command.
     pub fn run(&self) -> Result<(), Errors> {
         let bytecode_str: Cow<'_, str> = if let Some(path) = &self.path {
             // check if path exists.
@@ -78,9 +74,9 @@ impl Cmd {
         // BenchmarkDB is dummy state that implements Database trait.
         // the bytecode is deployed at zero address.
         let mut evm = Evm::builder()
-            .with_db(BenchmarkDB::new_bytecode(Bytecode::new_raw(
+            .with_db(BenchmarkDB::new_bytecode(Bytecode::new_raw_checked(
                 bytecode.into(),
-            )))
+            )?))
             .modify_tx_env(|tx| {
                 // execution globals block hash/gas_limit/coinbase/timestamp..
                 tx.caller = "0x0000000000000000000000000000000000000001"

bins/revme/src/cmd/statetest/runner.rs

@@ -9,8 +9,8 @@ use revm::{
     inspector_handle_register,
     inspectors::TracerEip3155,
     primitives::{
-        calc_excess_blob_gas, keccak256, Bytecode, Bytes, EVMResultGeneric, Env, Eof,
-        ExecutionResult, SpecId, TxKind, B256, EOF_MAGIC_BYTES,
+        calc_excess_blob_gas, keccak256, Bytecode, Bytes, EVMResultGeneric, Env, ExecutionResult,
+        SpecId, TxKind, B256,
     },
     Evm, State,
 };
@@ -258,12 +258,7 @@ pub fn execute_test_suite(
         let mut cache_state = revm::CacheState::new(false);
         for (address, info) in unit.pre {
             let code_hash = keccak256(&info.code);
-            let bytecode = match info.code.get(..2) {
-                Some(magic) if magic == &EOF_MAGIC_BYTES => {
-                    Bytecode::Eof(Eof::decode(info.code.clone()).unwrap().into())
-                }
-                _ => Bytecode::new_raw(info.code),
-            };
+            let bytecode = Bytecode::new_raw(info.code);
             let acc_info = revm::primitives::AccountInfo {
                 balance: info.balance,
                 code_hash,

crates/primitives/src/bytecode.rs

@@ -1,6 +1,7 @@
 pub mod eof;
 pub mod legacy;
 
+use eof::EofDecodeError;
 pub use eof::{Eof, EOF_MAGIC, EOF_MAGIC_BYTES, EOF_MAGIC_HASH};
 pub use legacy::{JumpTable, LegacyAnalyzedBytecode};
 use std::sync::Arc;
@@ -68,9 +69,25 @@ impl Bytecode {
     }
 
     /// Creates a new raw [`Bytecode`].
+    ///
+    /// # Panics
+    ///
+    /// Panics if bytecode is EOF and has incorrect format.
     #[inline]
     pub fn new_raw(bytecode: Bytes) -> Self {
-        Self::LegacyRaw(bytecode)
+        Self::new_raw_checked(bytecode).expect("Expect correct EOF bytecode")
+    }
+
+    /// Creates a new raw [`Bytecode`].
+    ///
+    /// Returns an error on incorrect EOF format.
+    #[inline]
+    pub fn new_raw_checked(bytecode: Bytes) -> Result<Self, EofDecodeError> {
+        if bytecode.get(..2) == Some(&[0xEF, 00]) {
+            Ok(Self::Eof(Arc::new(Eof::decode(bytecode)?)))
+        } else {
+            Ok(Self::LegacyRaw(bytecode))
+        }
     }
 
     /// Create new checked bytecode.

crates/primitives/src/bytecode/eof.rs

@@ -9,7 +9,7 @@ pub use types_section::TypesSection;
 
 use crate::{b256, bytes, Bytes, B256};
 use core::cmp::min;
-use std::{vec, vec::Vec};
+use std::{fmt, vec, vec::Vec};
 
 /// Hash of EF00 bytes that is used for EXTCODEHASH when called from legacy bytecode.
 pub const EOF_MAGIC_HASH: B256 =
@@ -157,6 +157,36 @@ pub enum EofDecodeError {
     TooManyContainerSections,
 }
 
+impl fmt::Display for EofDecodeError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let s = match self {
+            EofDecodeError::MissingInput => "Short input while processing EOF",
+            EofDecodeError::MissingBodyWithoutData => "Short body while processing EOF",
+            EofDecodeError::DanglingData => "Body size is more than specified in the header",
+            EofDecodeError::InvalidTypesSection => "Invalid types section data",
+            EofDecodeError::InvalidTypesSectionSize => "Invalid types section size",
+            EofDecodeError::InvalidEOFMagicNumber => "Invalid EOF magic number",
+            EofDecodeError::InvalidEOFVersion => "Invalid EOF version",
+            EofDecodeError::InvalidTypesKind => "Invalid number for types kind",
+            EofDecodeError::InvalidCodeKind => "Invalid number for code kind",
+            EofDecodeError::InvalidTerminalByte => "Invalid terminal code",
+            EofDecodeError::InvalidDataKind => "Invalid data kind",
+            EofDecodeError::InvalidKindAfterCode => "Invalid kind after code",
+            EofDecodeError::MismatchCodeAndTypesSize => "Mismatch of code and types sizes",
+            EofDecodeError::NonSizes => "There should be at least one size",
+            EofDecodeError::ShortInputForSizes => "Missing size",
+            EofDecodeError::ZeroSize => "Size cant be zero",
+            EofDecodeError::TooManyCodeSections => "Invalid code number",
+            EofDecodeError::ZeroCodeSections => "Invalid number of code sections",
+            EofDecodeError::TooManyContainerSections => "Invalid container number",
+        };
+        f.write_str(s)
+    }
+}
+
+#[cfg(feature = "std")]
+impl std::error::Error for EofDecodeError {}
+
 #[cfg(test)]
 mod test {
 

crates/revm/src/builder.rs

@@ -460,7 +460,7 @@ mod test {
 
     #[test]
     fn simple_add_stateful_instruction() {
-        let code = Bytecode::new_raw([0xEF, 0x00].into());
+        let code = Bytecode::new_raw([0xED, 0x00].into());
         let code_hash = code.hash_slow();
         let to_addr = address!("ffffffffffffffffffffffffffffffffffffffff");
 
@@ -493,7 +493,7 @@ mod test {
                 // can insert the custom instruction as a boxed instruction
                 handler
                     .instruction_table
-                    .insert_boxed(0xEF, custom_instruction);
+                    .insert_boxed(0xED, custom_instruction);
             }))
             .build();
 
@@ -514,7 +514,7 @@ mod test {
             gas!(interp, CUSTOM_INSTRUCTION_COST);
         }
 
-        let code = Bytecode::new_raw([0xEF, 0x00].into());
+        let code = Bytecode::new_raw([0xED, 0x00].into());
         let code_hash = code.hash_slow();
         let to_addr = address!("ffffffffffffffffffffffffffffffffffffffff");
 
@@ -525,7 +525,7 @@ mod test {
             })
             .modify_tx_env(|tx| tx.transact_to = TxKind::Call(to_addr))
             .append_handler_register(|handler| {
-                handler.instruction_table.insert(0xEF, custom_instruction)
+                handler.instruction_table.insert(0xED, custom_instruction)
             })
             .build();