Bus-mapping for opcode calldatacopy (privacy-scaling-explorations#393)

* Support generating multiple exec-steps from one geth-step.

* Fix build error.

* Fix to use bus-mapping to generate bytecode.

* Update test cases.

* Add basic of calldatacopy bus-mappinng to just support zero call data length.

* Update bus-mapping calldatacopy.

* Push op of call_data_length and call data offset.

* Add `is_root_call` to BytecodeTestConfig.

* Replace `OpcodeId` with `ExecState` in bus-mapping ExcStep.

* Generate CopyToMemory exection step.

* Add TransactionConfig to bus-mapping handle_tx.

* Update test code.

* 1. Remove TransactionConfig and replace with `call_data`.
2. Remove gas calculation and set in `calldatacopy` bus-mapping.
3. Revert `test_ok_internal` since not want to set internal call.

* Update test cases which call `handle_tx` and `new_tx` of circuit input builder.

* Update constant max address of state circuit.

* Add unit test for calldatacopy bus-mapping.

* change api

* fix calldatacopy

* fix rebase

* Set exec_state for BeginTx and EndTx.

* Fix to return a new exec step in function `dummy_gen_associated_ops`.

* Update bus-mapping calldatacopy unit-test.

* Update for fmt and clippy.

* Fix doc test.

* Update according to code review.

* Fix a comment.

* Fix to directly use StepAuxiliaryData of bus-mapping in zkevm-circuits.

* Revert a comment and a fix.

* address comments

* fix doc

Co-authored-by: Haichen Shen <[email protected]>
Co-authored-by: Zhang Zhuo <[email protected]>

bus-mapping/src/circuit_input_builder.rs

@@ -19,7 +19,7 @@ use crate::rpc::GethClient;
 use ethers_providers::JsonRpcClient;
 
 /// Out of Gas errors by opcode
-#[derive(Debug, PartialEq)]
+#[derive(Clone, Debug, PartialEq)]
 pub enum OogError {
     /// Out of Gas for opcodes which have non-zero constant gas cost
     Constant,
@@ -65,7 +65,7 @@ pub enum OogError {
 }
 
 /// EVM Execution Error
-#[derive(Debug, PartialEq)]
+#[derive(Clone, Debug, PartialEq)]
 pub enum ExecError {
     /// Invalid Opcode
     InvalidOpcode,
@@ -101,11 +101,71 @@ pub enum ExecError {
     MaxCodeSizeExceeded,
 }
 
+/// Execution state
+#[derive(Clone, Debug, Eq, PartialEq)]
+pub enum ExecState {
+    /// EVM Opcode ID
+    Op(OpcodeId),
+    /// Virtual step Begin Tx
+    BeginTx,
+    /// Virtual step End Tx
+    EndTx,
+    /// Virtual step Copy To Memory
+    CopyToMemory,
+}
+
+impl ExecState {
+    /// Returns `true` if `ExecState` is an opcode and the opcode is a `PUSHn`.
+    pub fn is_push(&self) -> bool {
+        if let ExecState::Op(op) = self {
+            op.is_push()
+        } else {
+            false
+        }
+    }
+
+    /// Returns `true` if `ExecState` is an opcode and the opcode is a `DUPn`.
+    pub fn is_dup(&self) -> bool {
+        if let ExecState::Op(op) = self {
+            op.is_dup()
+        } else {
+            false
+        }
+    }
+
+    /// Returns `true` if `ExecState` is an opcode and the opcode is a `SWAPn`.
+    pub fn is_swap(&self) -> bool {
+        if let ExecState::Op(op) = self {
+            op.is_swap()
+        } else {
+            false
+        }
+    }
+}
+
+/// Auxiliary data of Execution step
+#[derive(Clone, Debug)]
+pub enum StepAuxiliaryData {
+    /// Auxiliary data of Copy To Memory
+    CopyToMemory {
+        /// Source start address
+        src_addr: u64,
+        /// Destination address
+        dst_addr: u64,
+        /// Bytes left
+        bytes_left: u64,
+        /// Source end address
+        src_addr_end: u64,
+        /// Indicate if copy from transaction call data
+        from_tx: bool,
+    },
+}
+
 /// An execution step of the EVM.
-#[derive(Debug)]
+#[derive(Clone, Debug)]
 pub struct ExecStep {
-    /// The opcode ID
-    pub op: OpcodeId,
+    /// Execution state
+    pub exec_state: ExecState,
     /// Program Counter
     pub pc: ProgramCounter,
     /// Stack size
@@ -129,18 +189,20 @@ pub struct ExecStep {
     pub bus_mapping_instance: Vec<OperationRef>,
     /// Error generated by this step
     pub error: Option<ExecError>,
+    /// Step auxiliary data
+    pub aux_data: Option<StepAuxiliaryData>,
 }
 
 impl ExecStep {
-    /// Create a new Self from a [`GethExecStep`].
+    /// Create a new Self from a `GethExecStep`.
     pub fn new(
         step: &GethExecStep,
         call_index: usize,
         rwc: RWCounter,
         swc: usize, // State Write Counter
     ) -> Self {
         ExecStep {
-            op: step.op,
+            exec_state: ExecState::Op(step.op),
             pc: step.pc,
             stack_size: step.stack.0.len(),
             memory_size: step.memory.0.len(),
@@ -151,14 +213,15 @@ impl ExecStep {
             swc,
             bus_mapping_instance: Vec::new(),
             error: None,
+            aux_data: None,
         }
     }
 }
 
 impl Default for ExecStep {
     fn default() -> Self {
         Self {
-            op: OpcodeId::INVALID(0),
+            exec_state: ExecState::Op(OpcodeId::INVALID(0)),
             pc: ProgramCounter(0),
             stack_size: 0,
             memory_size: 0,
@@ -169,6 +232,7 @@ impl Default for ExecStep {
             swc: 0,
             bus_mapping_instance: Vec::new(),
             error: None,
+            aux_data: None,
         }
     }
 }
@@ -368,12 +432,12 @@ impl Call {
 /// Context of a [`Call`].
 #[derive(Debug, Default)]
 pub struct CallContext {
-    // Index of call
-    index: usize,
+    /// Index of call
+    pub index: usize,
     /// State Write Counter tracks the count of state write operations in the
     /// call. When a subcall in this call succeeds, the `swc` increases by the
     /// number of successful state writes in the subcall.
-    swc: usize,
+    pub swc: usize,
 }
 
 /// A reversion group is the collection of calls and the operations which are
@@ -466,6 +530,11 @@ impl TransactionContext {
         self.is_last_tx
     }
 
+    /// Return the calls in this transaction.
+    pub fn calls(&self) -> &[CallContext] {
+        &self.calls
+    }
+
     /// Return the index of the current call (the last call in the call stack).
     fn call_index(&self) -> Result<usize, Error> {
         self.calls
@@ -541,8 +610,10 @@ pub struct Transaction {
     /// Value
     pub value: Word,
     /// Input / Call Data
-    pub input: Vec<u8>, // call_data
+    pub input: Vec<u8>,
+    /// Calls made in the transaction
     calls: Vec<Call>,
+    /// Execution steps
     steps: Vec<ExecStep>,
 }
 
@@ -654,21 +725,61 @@ pub struct CircuitInputStateRef<'a> {
     pub tx: &'a mut Transaction,
     /// Transaction Context
     pub tx_ctx: &'a mut TransactionContext,
-    /// Step
-    pub step: &'a mut ExecStep,
 }
 
 impl<'a> CircuitInputStateRef<'a> {
+    /// Create a new step from a `GethExecStep`
+    pub fn new_step(&self, geth_step: &GethExecStep) -> Result<ExecStep, Error> {
+        let call_ctx = self.tx_ctx.call_ctx()?;
+        Ok(ExecStep::new(
+            geth_step,
+            call_ctx.index,
+            self.block_ctx.rwc,
+            call_ctx.swc,
+        ))
+    }
+
+    /// Create a new BeginTx step
+    pub fn new_begin_tx_step(&self) -> ExecStep {
+        ExecStep {
+            exec_state: ExecState::BeginTx,
+            gas_left: Gas(self.tx.gas),
+            rwc: self.block_ctx.rwc,
+            ..Default::default()
+        }
+    }
+
+    /// Create a new EndTx step
+    pub fn new_end_tx_step(&self) -> ExecStep {
+        let prev_step = self
+            .tx
+            .steps()
+            .last()
+            .expect("steps should have at least one BeginTx step");
+        ExecStep {
+            exec_state: ExecState::EndTx,
+            gas_left: Gas(prev_step.gas_left.0 - prev_step.gas_cost.0),
+            rwc: self.block_ctx.rwc,
+            // For tx without code execution
+            swc: if let Some(call_ctx) = self.tx_ctx.calls().last() {
+                call_ctx.swc
+            } else {
+                0
+            },
+            ..Default::default()
+        }
+    }
+
     /// Push an [`Operation`] into the [`OperationContainer`] with the next
     /// [`RWCounter`] and then adds a reference to the stored operation
     /// ([`OperationRef`]) inside the bus-mapping instance of the current
     /// [`ExecStep`].  Then increase the block_ctx [`RWCounter`] by one.
-    pub fn push_op<T: Op>(&mut self, rw: RW, op: T) {
+    pub fn push_op<T: Op>(&mut self, step: &mut ExecStep, rw: RW, op: T) {
         let op_ref =
             self.block
                 .container
                 .insert(Operation::new(self.block_ctx.rwc.inc_pre(), rw, op));
-        self.step.bus_mapping_instance.push(op_ref);
+        step.bus_mapping_instance.push(op_ref);
     }
 
     /// Push an [`Operation`] with reversible to be true into the
@@ -679,13 +790,18 @@ impl<'a> CircuitInputStateRef<'a> {
     /// This method should be used in `Opcode::gen_associated_ops` instead of
     /// `push_op` when the operation is `RW::WRITE` and it can be reverted (for
     /// example, a write `StorageOp`).
-    pub fn push_op_reversible<T: Op>(&mut self, rw: RW, op: T) -> Result<(), Error> {
+    pub fn push_op_reversible<T: Op>(
+        &mut self,
+        step: &mut ExecStep,
+        rw: RW,
+        op: T,
+    ) -> Result<(), Error> {
         let op_ref = self.block.container.insert(Operation::new_reversible(
             self.block_ctx.rwc.inc_pre(),
             rw,
             op,
         ));
-        self.step.bus_mapping_instance.push(op_ref);
+        step.bus_mapping_instance.push(op_ref);
 
         // Increase state_write_counter
         self.call_ctx_mut()?.swc += 1;
@@ -710,12 +826,13 @@ impl<'a> CircuitInputStateRef<'a> {
     /// [`RWCounter`] by one.
     pub fn push_memory_op(
         &mut self,
+        step: &mut ExecStep,
         rw: RW,
         address: MemoryAddress,
         value: u8,
     ) -> Result<(), Error> {
         let call_id = self.call()?.call_id;
-        self.push_op(rw, MemoryOp::new(call_id, address, value));
+        self.push_op(step, rw, MemoryOp::new(call_id, address, value));
         Ok(())
     }
 
@@ -726,12 +843,13 @@ impl<'a> CircuitInputStateRef<'a> {
     /// [`RWCounter`] by one.
     pub fn push_stack_op(
         &mut self,
+        step: &mut ExecStep,
         rw: RW,
         address: StackAddress,
         value: Word,
     ) -> Result<(), Error> {
         let call_id = self.call()?.call_id;
-        self.push_op(rw, StackOp::new(call_id, address, value));
+        self.push_op(step, rw, StackOp::new(call_id, address, value));
         Ok(())
     }
 
@@ -1254,7 +1372,6 @@ impl<'a> CircuitInputBuilder {
         &'a mut self,
         tx: &'a mut Transaction,
         tx_ctx: &'a mut TransactionContext,
-        step: &'a mut ExecStep,
     ) -> CircuitInputStateRef {
         CircuitInputStateRef {
             sdb: &mut self.sdb,
@@ -1263,7 +1380,6 @@ impl<'a> CircuitInputBuilder {
             block_ctx: &mut self.block_ctx,
             tx,
             tx_ctx,
-            step,
         }
     }
 
@@ -1342,50 +1458,25 @@ impl<'a> CircuitInputBuilder {
         // - execution_state: BeginTx
         // - op: None
         // Generate BeginTx step
-        let mut step = ExecStep {
-            gas_left: Gas(tx.gas),
-            rwc: self.block_ctx.rwc,
-            ..Default::default()
-        };
-        gen_begin_tx_ops(&mut self.state_ref(&mut tx, &mut tx_ctx, &mut step))?;
-        tx.steps.push(step);
+        let begin_tx_step = gen_begin_tx_ops(&mut self.state_ref(&mut tx, &mut tx_ctx))?;
+        tx.steps.push(begin_tx_step);
 
         for (index, geth_step) in geth_trace.struct_logs.iter().enumerate() {
-            let call_ctx = tx_ctx.call_ctx()?;
-            let mut step =
-                ExecStep::new(geth_step, call_ctx.index, self.block_ctx.rwc, call_ctx.swc);
-            let mut state_ref = self.state_ref(&mut tx, &mut tx_ctx, &mut step);
-
-            gen_associated_ops(
+            let mut state_ref = self.state_ref(&mut tx, &mut tx_ctx);
+            let exec_steps = gen_associated_ops(
                 &geth_step.op,
                 &mut state_ref,
                 &geth_trace.struct_logs[index..],
             )?;
-
-            tx.steps.push(step);
+            tx.steps.extend(exec_steps);
         }
 
         // TODO: Move into gen_associated_steps with
         // - execution_state: EndTx
         // - op: None
         // Generate EndTx step
-        let step_prev = tx
-            .steps
-            .last()
-            .expect("steps should have at least one BeginTx step");
-        let mut step = ExecStep {
-            gas_left: Gas(step_prev.gas_left.0 - step_prev.gas_cost.0),
-            rwc: self.block_ctx.rwc,
-            // For tx without code execution
-            swc: if let Some(call_ctx) = tx_ctx.calls.last() {
-                call_ctx.swc
-            } else {
-                0
-            },
-            ..Default::default()
-        };
-        gen_end_tx_ops(&mut self.state_ref(&mut tx, &mut tx_ctx, &mut step))?;
-        tx.steps.push(step);
+        let end_tx_step = gen_end_tx_ops(&mut self.state_ref(&mut tx, &mut tx_ctx))?;
+        tx.steps.push(end_tx_step);
 
         self.block.txs.push(tx);
         self.sdb.clear_access_list_and_refund();
@@ -1898,8 +1989,7 @@ mod tracer_tests {
         }
 
         fn state_ref(&mut self) -> CircuitInputStateRef {
-            self.builder
-                .state_ref(&mut self.tx, &mut self.tx_ctx, &mut self.step)
+            self.builder.state_ref(&mut self.tx, &mut self.tx_ctx)
         }
     }
 
@@ -1970,7 +2060,7 @@ mod tracer_tests {
                  STOP
         };
         let block =
-            mock::new_single_tx_trace_code_gas(&code, Gas(1_000_000_000_000_000u64)).unwrap();
+            mock::new_single_tx_trace_code_gas(&code, Gas(1_000_000_000_000_000u64), None).unwrap();
         let struct_logs = &block.geth_traces[0].struct_logs;
 
         // get last CALL
@@ -2830,7 +2920,7 @@ mod tracer_tests {
             PUSH1(0x1)
             PUSH1(0x2)
         };
-        let block = mock::new_single_tx_trace_code_gas(&code, Gas(21004)).unwrap();
+        let block = mock::new_single_tx_trace_code_gas(&code, Gas(21004), None).unwrap();
         let struct_logs = &block.geth_traces[0].struct_logs;
 
         assert_eq!(struct_logs[1].error, Some(GETH_ERR_OUT_OF_GAS.to_string()));