code_source rename to code_hash

specs/tables.md

@@ -94,7 +94,7 @@ Type sizes:
 |          |             | *CallContext state*        |          |                | *CallContextFieldTag* (rw) |             |            |            |                 |
 | $counter | $isWrite    | CallContext                | $callID  |                | IsRoot                     |             | $value     | 0          | 0               |
 | $counter | $isWrite    | CallContext                | $callID  |                | IsCreate                   |             | $value     | 0          | 0               |
-| $counter | $isWrite    | CallContext                | $callID  |                | CodeSource                 |             | $value     | 0          | 0               |
+| $counter | $isWrite    | CallContext                | $callID  |                | CodeHash                   |             | $value     | 0          | 0               |
 | $counter | $isWrite    | CallContext                | $callID  |                | ProgramCounter             |             | $value     | 0          | 0               |
 | $counter | $isWrite    | CallContext                | $callID  |                | StackPointer               |             | $value     | 0          | 0               |
 | $counter | $isWrite    | CallContext                | $callID  |                | GasLeft                    |             | $value     | 0          | 0               |

src/zkevm_specs/evm/execution/begin_tx.py

@@ -58,7 +58,10 @@ def begin_tx(instruction: Instruction):
 
     if tx_is_create == 1:
         # TODO: Verify created address
-        # TODO: Decide what code_source should be (tx_id or hash of creation code)
+        # code_hash represents the contract creation code
+        # 1. In the case of root call, code_hash is the hash of the tx calldata.
+        # 2. In the case of internal call, code_hash is the hash of the chunk of
+        #    caller memory corresponding to the contract init code.
         raise NotImplementedError
     elif tx_callee_address in list(PrecompiledAddress):
         # TODO: Handle precompile
@@ -99,7 +102,7 @@ def begin_tx(instruction: Instruction):
                 (CallContextFieldTag.LastCalleeReturnDataLength, FQ(0)),
                 (CallContextFieldTag.IsRoot, FQ(True)),
                 (CallContextFieldTag.IsCreate, FQ(False)),
-                (CallContextFieldTag.CodeSource, code_hash),
+                (CallContextFieldTag.CodeHash, code_hash),
             ]:
                 instruction.constrain_equal(
                     instruction.call_context_lookup(tag, call_id=call_id), value
@@ -110,7 +113,7 @@ def begin_tx(instruction: Instruction):
                 call_id=Transition.to(call_id),
                 is_root=Transition.to(True),
                 is_create=Transition.to(False),
-                code_source=Transition.to(code_hash),
+                code_hash=Transition.to(code_hash),
                 gas_left=Transition.to(gas_left),
                 reversible_write_counter=Transition.to(2),
                 log_id=Transition.to(0),

src/zkevm_specs/evm/execution/call.py

@@ -142,7 +142,7 @@ def call(instruction: Instruction):
             call_id=Transition.same(),
             is_root=Transition.same(),
             is_create=Transition.same(),
-            code_source=Transition.same(),
+            code_hash=Transition.same(),
         )
     else:
         # Save caller's call state
@@ -181,7 +181,7 @@ def call(instruction: Instruction):
             (CallContextFieldTag.LastCalleeReturnDataLength, FQ(0)),
             (CallContextFieldTag.IsRoot, FQ(False)),
             (CallContextFieldTag.IsCreate, FQ(False)),
-            (CallContextFieldTag.CodeSource, callee_code_hash.expr()),
+            (CallContextFieldTag.CodeHash, callee_code_hash.expr()),
         ]:
             instruction.constrain_equal(
                 instruction.call_context_lookup(field_tag, call_id=callee_call_id),
@@ -196,7 +196,7 @@ def call(instruction: Instruction):
             call_id=Transition.to(callee_call_id),
             is_root=Transition.to(False),
             is_create=Transition.to(False),
-            code_source=Transition.to(callee_code_hash),
+            code_hash=Transition.to(callee_code_hash),
             gas_left=Transition.to(callee_gas_left),
             reversible_write_counter=Transition.to(2),
             log_id=Transition.same(),

src/zkevm_specs/evm/execution/codecopy.py

@@ -17,7 +17,7 @@ def codecopy(instruction: Instruction):
     memory_offset, size = instruction.memory_offset_and_length(memory_offset_word, size_word)
     code_offset = instruction.rlc_to_fq(code_offset_word, N_BYTES_MEMORY_ADDRESS)
 
-    code_size = instruction.bytecode_length(instruction.curr.code_source)
+    code_size = instruction.bytecode_length(instruction.curr.code_hash)
 
     next_memory_size, memory_expansion_gas_cost = instruction.memory_expansion_dynamic_length(
         memory_offset, size
@@ -35,7 +35,7 @@ def codecopy(instruction: Instruction):
         instruction.constrain_equal(next_aux.dst_addr, memory_offset)
         instruction.constrain_equal(next_aux.src_addr_end, code_size)
         instruction.constrain_equal(next_aux.bytes_left, size)
-        instruction.constrain_equal(next_aux.code_source, instruction.curr.code_source)
+        instruction.constrain_equal(next_aux.code_hash, instruction.curr.code_hash)
 
     instruction.step_state_transition_in_same_context(
         opcode,

src/zkevm_specs/evm/execution/copy_code_to_memory.py

@@ -20,7 +20,7 @@ def copy_code_to_memory(instruction: Instruction):
     for idx in range(MAX_N_BYTES_COPY_CODE_TO_MEMORY):
         if buffer_reader.read_flag(idx) == 1:
             byte = instruction.bytecode_lookup(
-                aux.code_source,
+                aux.code_hash,
                 aux.src_addr + idx,
             )
             buffer_reader.constrain_byte(idx, byte)
@@ -47,14 +47,14 @@ def copy_code_to_memory(instruction: Instruction):
         instruction.constrain_equal(next_aux.dst_addr, aux.dst_addr + copied_bytes)
         instruction.constrain_equal(next_aux.bytes_left + copied_bytes, aux.bytes_left)
         instruction.constrain_equal(next_aux.src_addr_end, aux.src_addr_end)
-        instruction.constrain_equal(next_aux.code_source, aux.code_source)
+        instruction.constrain_equal(next_aux.code_hash, aux.code_hash)
 
     instruction.constrain_step_state_transition(
         rw_counter=Transition.delta(instruction.rw_counter_offset),
         call_id=Transition.same(),
         is_root=Transition.same(),
         is_create=Transition.same(),
-        code_source=Transition.same(),
+        code_hash=Transition.same(),
         program_counter=Transition.same(),
         stack_pointer=Transition.same(),
         memory_size=Transition.same(),

src/zkevm_specs/evm/execution/memory_copy.py

@@ -48,7 +48,7 @@ def copy_to_memory(instruction: Instruction):
         call_id=Transition.same(),
         is_root=Transition.same(),
         is_create=Transition.same(),
-        code_source=Transition.same(),
+        code_hash=Transition.same(),
         program_counter=Transition.same(),
         stack_pointer=Transition.same(),
         memory_size=Transition.same(),

src/zkevm_specs/evm/instruction.py

@@ -164,7 +164,7 @@ def constrain_step_state_transition(self, **kwargs: Transition):
                 "call_id",
                 "is_root",
                 "is_create",
-                "code_source",
+                "code_hash",
                 "program_counter",
                 "stack_pointer",
                 "gas_left",
@@ -209,7 +209,7 @@ def step_state_transition_to_new_context(
         call_id: Transition,
         is_root: Transition,
         is_create: Transition,
-        code_source: Transition,
+        code_hash: Transition,
         gas_left: Transition,
         reversible_write_counter: Transition,
         log_id: Transition,
@@ -219,7 +219,7 @@ def step_state_transition_to_new_context(
             call_id=call_id,
             is_root=is_root,
             is_create=is_create,
-            code_source=code_source,
+            code_hash=code_hash,
             gas_left=gas_left,
             reversible_write_counter=reversible_write_counter,
             log_id=log_id,
@@ -257,7 +257,7 @@ def step_state_transition_in_same_context(
             call_id=Transition.same(),
             is_root=Transition.same(),
             is_create=Transition.same(),
-            code_source=Transition.same(),
+            code_hash=Transition.same(),
         )
 
     def sum(self, values: Sequence[IntOrFQ]) -> FQ:
@@ -509,7 +509,7 @@ def opcode_lookup_at(self, index: FQ, is_code: bool) -> FQ:
                 "The opcode source when is_root and is_create (root creation call) is not determined yet"
             )
         else:
-            return self.bytecode_lookup(self.curr.code_source, index, FQ(is_code)).expr()
+            return self.bytecode_lookup(self.curr.code_hash, index, FQ(is_code)).expr()
 
     def rw_lookup(
         self,

src/zkevm_specs/evm/step.py

@@ -9,7 +9,7 @@ class StepState:
     Step state EVM circuit tracks step by step and used to ensure the
     execution trace is verified continuously and chronologically.
     It includes fields that are used from beginning to end like is_root,
-    is_create and code_source.
+    is_create and code_hash.
     It also includes call's mutable states which change almost every step like
     program_counter and stack_pointer.
     """
@@ -18,18 +18,15 @@ class StepState:
     rw_counter: FQ
     call_id: FQ
 
-    # The following 3 fields decide the opcode source. There are 2 possible
-    # cases:
-    # 1. Root creation call (is_root and is_create)
-    #   It was planned to set the code_source to tx_id, then lookup tx_table's
-    #   CallData field directly, but is still yet to be determined.
-    #   See the issue https://github.com/appliedzkp/zkevm-specs/issues/73 for
-    #   further discussion.
-    # 2. Deployed contract interaction or internal creation call
-    #   We set code_source to bytecode_hash and lookup bytecode_table.
     is_root: bool
     is_create: bool
-    code_source: RLC
+
+    # code_hash represents the bytecode hash of the bytecode. This is straightforward
+    # for a contract call that does not create a contract. For a creation call,
+    # we already populate the bytecode table with the contract creation code either
+    # from the tx calldata (in the case of a root call) or from the caller's memory
+    # (in the case of an internal call).
+    code_hash: RLC
 
     # The following fields change almost every step.
     program_counter: FQ
@@ -55,7 +52,7 @@ def __init__(
         call_id: int = 0,
         is_root: bool = False,
         is_create: bool = False,
-        code_source: RLC = RLC(0),
+        code_hash: RLC = RLC(0),
         program_counter: int = 0,
         stack_pointer: int = 1024,
         gas_left: int = 0,
@@ -69,7 +66,7 @@ def __init__(
         self.call_id = FQ(call_id)
         self.is_root = is_root
         self.is_create = is_create
-        self.code_source = code_source
+        self.code_hash = code_hash
         self.program_counter = FQ(program_counter)
         self.stack_pointer = FQ(stack_pointer)
         self.gas_left = FQ(gas_left)
@@ -131,18 +128,18 @@ class CopyCodeToMemoryAuxData:
     dst_addr: FQ
     bytes_left: FQ
     src_addr_end: FQ
-    code_source: RLC
+    code_hash: RLC
 
     def __init__(
         self,
         src_addr: int,
         dst_addr: int,
         bytes_left: int,
         src_addr_end: int,
-        code_source: RLC,
+        code_hash: RLC,
     ):
         self.src_addr = FQ(src_addr)
         self.dst_addr = FQ(dst_addr)
         self.bytes_left = FQ(bytes_left)
         self.src_addr_end = FQ(src_addr_end)
-        self.code_source = code_source
+        self.code_hash = code_hash

src/zkevm_specs/evm/table.py

@@ -223,7 +223,7 @@ class CallContextFieldTag(IntEnum):
     IsStatic = auto()  # to know if state modification is within static call or not
     IsRoot = auto()
     IsCreate = auto()
-    CodeSource = auto()
+    CodeHash = auto()
 
     # The following are read-only data inside a call like previous section for
     # opcode RETURNDATASIZE and RETURNDATACOPY, except they will be updated when

tests/evm/test_add_sub.py

@@ -59,7 +59,7 @@ def test_add_sub(opcode: Opcode, a: int, b: int):
                 call_id=1,
                 is_root=True,
                 is_create=False,
-                code_source=bytecode_hash,
+                code_hash=bytecode_hash,
                 program_counter=66,
                 stack_pointer=1022,
                 gas_left=3,
@@ -70,7 +70,7 @@ def test_add_sub(opcode: Opcode, a: int, b: int):
                 call_id=1,
                 is_root=True,
                 is_create=False,
-                code_source=bytecode_hash,
+                code_hash=bytecode_hash,
                 program_counter=67,
                 stack_pointer=1023,
                 gas_left=0,

tests/evm/test_begin_tx.py

@@ -131,7 +131,7 @@ def test_begin_tx(tx: Transaction, callee: Account, is_success: bool):
         .call_context_read(1, CallContextFieldTag.LastCalleeReturnDataLength, 0) \
         .call_context_read(1, CallContextFieldTag.IsRoot, True) \
         .call_context_read(1, CallContextFieldTag.IsCreate, False) \
-        .call_context_read(1, CallContextFieldTag.CodeSource, bytecode_hash)
+        .call_context_read(1, CallContextFieldTag.CodeHash, bytecode_hash)
     # fmt: on
 
     tables = Tables(
@@ -157,7 +157,7 @@ def test_begin_tx(tx: Transaction, callee: Account, is_success: bool):
                 call_id=1,
                 is_root=True,
                 is_create=False,
-                code_source=bytecode_hash,
+                code_hash=bytecode_hash,
                 program_counter=0,
                 stack_pointer=1024,
                 gas_left=0,

tests/evm/test_block_ctx.py

@@ -139,7 +139,7 @@ def verify_block_ctx(
                 call_id=1,
                 is_root=True,
                 is_create=False,
-                code_source=bytecode_hash,
+                code_hash=bytecode_hash,
                 program_counter=0,
                 stack_pointer=1024,
                 gas_left=2,
@@ -150,7 +150,7 @@ def verify_block_ctx(
                 call_id=1,
                 is_root=True,
                 is_create=False,
-                code_source=bytecode_hash,
+                code_hash=bytecode_hash,
                 program_counter=1,
                 stack_pointer=1023,
                 gas_left=0,

tests/evm/test_call.py

@@ -238,7 +238,7 @@ def test_call(
         .call_context_read(24, CallContextFieldTag.LastCalleeReturnDataLength, 0) \
         .call_context_read(24, CallContextFieldTag.IsRoot, False) \
         .call_context_read(24, CallContextFieldTag.IsCreate, False) \
-        .call_context_read(24, CallContextFieldTag.CodeSource, callee_bytecode_hash)
+        .call_context_read(24, CallContextFieldTag.CodeHash, callee_bytecode_hash)
     # fmt: on
 
     tables = Tables(
@@ -263,7 +263,7 @@ def test_call(
                 call_id=1,
                 is_root=True,
                 is_create=False,
-                code_source=caller_bytecode_hash,
+                code_hash=caller_bytecode_hash,
                 program_counter=231,
                 stack_pointer=1017,
                 gas_left=caller_ctx.gas_left,
@@ -277,7 +277,7 @@ def test_call(
                     call_id=1,
                     is_root=True,
                     is_create=False,
-                    code_source=caller_bytecode_hash,
+                    code_hash=caller_bytecode_hash,
                     program_counter=232,
                     stack_pointer=1023,
                     gas_left=expected.caller_gas_left,
@@ -293,7 +293,7 @@ def test_call(
                     call_id=24,
                     is_root=False,
                     is_create=False,
-                    code_source=callee_bytecode_hash,
+                    code_hash=callee_bytecode_hash,
                     program_counter=0,
                     stack_pointer=1024,
                     gas_left=expected.callee_gas_left,