feat[ir]: add sanity check

vyper/codegen/ir_node.py

@@ -139,6 +139,9 @@ class IRnode:
     func_ir: Any
     common_ir: Any
 
+    _id: int
+    _next_id: int = -1
+
     def __init__(
         self,
         value: Union[str, int],
@@ -153,6 +156,7 @@ def __init__(
         encoding: Encoding = Encoding.VYPER,
         is_self_call: bool = False,
         passthrough_metadata: dict[str, Any] = None,
+        _id: int = None,
     ):
         if args is None:
             args = []
@@ -175,6 +179,12 @@ def __init__(
         self.func_ir = None
         self.common_ir = None
 
+        self._id = _id  # type: ignore[assignment]
+        self.ensure_id()
+
+        # helpful for debugging:
+        # self._origin = traceback.extract_stack()
+
         assert self.value is not None, "None is not allowed as IRnode value"
 
         # Determine this node's valency (1 if it pushes a value on the stack,
@@ -358,6 +368,15 @@ def __deepcopy__(self, memo):
         ret.args = [copy.deepcopy(arg) for arg in ret.args]
         return ret
 
+    @classmethod
+    def generate_id(cls):
+        cls._next_id += 1
+        return cls._next_id
+
+    def ensure_id(self):
+        if self._id is None:
+            self._id = self.generate_id()
+
     # TODO would be nice to rename to `gas_estimate` or `gas_bound`
     @property
     def gas(self):
@@ -489,6 +508,8 @@ def repr_value(self):
             return hex(self.value)
         if not isinstance(self.value, str):
             return str(self.value)
+        # useful for debugging
+        # return f"{self._id}:{self.value}"
         return self.value
 
     @staticmethod
@@ -559,11 +580,13 @@ def from_list(
         is_self_call: bool = False,
         passthrough_metadata: dict[str, Any] = None,
         encoding: Encoding = Encoding.VYPER,
+        _id=None,
     ) -> "IRnode":
         if isinstance(typ, str):  # pragma: nocover
             raise CompilerPanic(f"Expected type, not string: {typ}")
 
         if isinstance(obj, IRnode):
+            assert _id is None
             # note: this modify-and-returnclause is a little weird since
             # the input gets modified. CC 20191121.
             if typ is not None:
@@ -592,6 +615,7 @@ def from_list(
                 error_msg=error_msg,
                 is_self_call=is_self_call,
                 passthrough_metadata=passthrough_metadata,
+                _id=_id,
             )
         else:
             return cls(
@@ -607,4 +631,5 @@ def from_list(
                 error_msg=error_msg,
                 is_self_call=is_self_call,
                 passthrough_metadata=passthrough_metadata,
+                _id=_id,
             )

vyper/ir/compile_ir.py

@@ -211,11 +211,24 @@
     return apply_line_no_wrapper
 
 
+def check_duplicated_nodes(ir_node, seen=None):
+    seen = seen or set()
+
+    if ir_node.is_complex_ir and ir_node._id in seen:
+        raise CompilerPanic(f"bad code {ir_node}", ir_node.ast_source)
+    seen.add(ir_node._id)
+
+    for arg in ir_node.args:
+        check_duplicated_nodes(arg, seen)
+
+
 @apply_line_numbers
 def compile_to_assembly(code, optimize=OptimizationLevel.GAS):
     global _revert_label
     _revert_label = mksymbol("revert")
 
+    check_duplicated_nodes(code)
+
     # don't overwrite ir since the original might need to be output, e.g. `-f ir,asm`
     code = copy.deepcopy(code)
     _rewrite_return_sequences(code)

vyper/ir/optimizer.py

@@ -442,6 +442,7 @@ def _optimize(node: IRnode, parent: Optional[IRnode]) -> Tuple[bool, IRnode]:
     add_gas_estimate = node.add_gas_estimate
     is_self_call = node.is_self_call
     passthrough_metadata = node.passthrough_metadata
+    _id = node._id
 
     changed = False
 
@@ -466,6 +467,7 @@ def finalize(val, args):
             add_gas_estimate=add_gas_estimate,
             is_self_call=is_self_call,
             passthrough_metadata=passthrough_metadata,
+            _id=_id,
         )
 
         if should_check_symbols: