feat: error map (#2939)

add an error map to source_map output. this makes it easier for
downstream tooling to detect why a revert was issued (without having to
rely on heuristics or explicit returndata). this also paves the way for
brownie-style dev revert strings to be supported by at the language
level.

Author: charles-cooper

tests/compiler/test_source_map.py

@@ -68,6 +68,19 @@ def test_pos_map_offsets():
             )
 
 
+def test_error_map():
+    code = """
+foo: uint256
+
+@external
+def update_foo():
+    self.foo += 1
+    """
+    error_map = compile_code(code, ["source_map"])["source_map"]["error_map"]
+    assert "safeadd" in list(error_map.values())
+    assert "fallback function" in list(error_map.values())
+
+
 def test_compress_source_map():
     code = """
 @external

vyper/codegen/arithmetic.py

@@ -155,7 +155,8 @@ def safe_add(x, y):
             # TODO push down into optimizer rules.
             ok = ["ge", res, x]
 
-        ret = IRnode.from_list(["seq", ["assert", ok], res])
+        check = IRnode.from_list(["assert", ok], error_msg="safeadd")
+        ret = IRnode.from_list(["seq", check, res])
         return b1.resolve(ret)
 
 
@@ -184,7 +185,8 @@ def safe_sub(x, y):
             # TODO push down into optimizer rules.
             ok = ["le", res, x]
 
-        ret = IRnode.from_list(["seq", ["assert", ok], res])
+        check = IRnode.from_list(["assert", ok], error_msg="safesub")
+        ret = IRnode.from_list(["seq", check, res])
         return b1.resolve(ret)
 
 
@@ -250,7 +252,8 @@ def safe_mul(x, y):
         # (if bits == 256, clamp_basetype is a no-op)
         res = clamp_basetype(res)
 
-        res = IRnode.from_list(["seq", ["assert", ok], res], typ=res.typ)
+        check = IRnode.from_list(["assert", ok], error_msg="safediv")
+        res = IRnode.from_list(["seq", check, res], typ=res.typ)
 
         return b1.resolve(res)
 
@@ -308,7 +311,7 @@ def safe_div(x, y):
             # TODO maybe use safe_mul
             res = clamp_basetype(res)
 
-        check = ["assert", ok]
+        check = IRnode.from_list(["assert", ok], error_msg="safemul")
         return IRnode.from_list(b1.resolve(["seq", check, res]))
 
 

vyper/codegen/core.py

@@ -976,7 +976,7 @@ def clamp_basetype(ir_node):
     else:  # pragma: nocover
         raise CompilerPanic(f"{t} passed to clamp_basetype")
 
-    return IRnode.from_list(ret, typ=ir_node.typ)
+    return IRnode.from_list(ret, typ=ir_node.typ, error_msg=f"validate {t}")
 
 
 def int_clamp(ir_node, bits, signed=False):
@@ -1024,15 +1024,16 @@ def promote_signed_int(x, bits):
 # general clamp function for all ops and numbers
 def clamp(op, arg, bound):
     with IRnode.from_list(arg).cache_when_complex("clamp_arg") as (b1, arg):
-        assertion = ["assert", [op, arg, bound]]
-        ret = ["seq", assertion, arg]
+        check = IRnode.from_list(["assert", [op, arg, bound]], error_msg=f"clamp {op} {bound}")
+        ret = ["seq", check, arg]
         return IRnode.from_list(b1.resolve(ret), typ=arg.typ)
 
 
 def clamp_nonzero(arg):
     # TODO: use clamp("ne", arg, 0) once optimizer rules can handle it
     with IRnode.from_list(arg).cache_when_complex("should_nonzero") as (b1, arg):
-        ret = ["seq", ["assert", arg], arg]
+        check = IRnode.from_list(["assert", arg], error_msg="clamp_nonzero")
+        ret = ["seq", check, arg]
         return IRnode.from_list(b1.resolve(ret), typ=arg.typ)
 
 

vyper/codegen/ir_node.py

@@ -115,6 +115,7 @@ def __init__(
         location: Optional[AddrSpace] = None,
         source_pos: Optional[Tuple[int, int]] = None,
         annotation: Optional[str] = None,
+        error_msg: Optional[str] = None,
         mutable: bool = True,
         add_gas_estimate: int = 0,
         encoding: Encoding = Encoding.VYPER,
@@ -129,6 +130,7 @@ def __init__(
         self.typ = typ
         self.location = location
         self.source_pos = source_pos
+        self.error_msg = error_msg
         self.annotation = annotation
         self.mutable = mutable
         self.add_gas_estimate = add_gas_estimate
@@ -494,6 +496,7 @@ def from_list(
         location: Optional[AddrSpace] = None,
         source_pos: Optional[Tuple[int, int]] = None,
         annotation: Optional[str] = None,
+        error_msg: Optional[str] = None,
         mutable: bool = True,
         add_gas_estimate: int = 0,
         encoding: Encoding = Encoding.VYPER,
@@ -512,6 +515,8 @@ def from_list(
                 obj.location = location
             if obj.encoding is None:
                 obj.encoding = encoding
+            if obj.error_msg is None:
+                obj.error_msg = error_msg
 
             return obj
         elif not isinstance(obj, list):
@@ -523,7 +528,9 @@ def from_list(
                 annotation=annotation,
                 mutable=mutable,
                 add_gas_estimate=add_gas_estimate,
+                source_pos=source_pos,
                 encoding=encoding,
+                error_msg=error_msg,
             )
         else:
             return cls(
@@ -536,4 +543,5 @@ def from_list(
                 source_pos=source_pos,
                 add_gas_estimate=add_gas_estimate,
                 encoding=encoding,
+                error_msg=error_msg,
             )

vyper/codegen/module.py

@@ -150,7 +150,9 @@ def _runtime_ir(runtime_functions, all_sigs, global_ctx):
             default_function, all_sigs, global_ctx, skip_nonpayable_check
         )
     else:
-        fallback_ir = IRnode.from_list(["revert", 0, 0], annotation="Default function")
+        fallback_ir = IRnode.from_list(
+            ["revert", 0, 0], annotation="Default function", error_msg="fallback function"
+        )
 
     # ensure the external jumptable section gets closed out
     # (for basic block hygiene and also for zksync interpreter)

vyper/codegen/stmt.py

@@ -160,7 +160,9 @@ def parse_Call(self):
 
     def _assert_reason(self, test_expr, msg):
         if isinstance(msg, vy_ast.Name) and msg.id == "UNREACHABLE":
-            return IRnode.from_list(["assert_unreachable", test_expr])
+            return IRnode.from_list(
+                ["assert_unreachable", test_expr], error_msg="assert unreachable"
+            )
 
         # set constant so that revert reason str is well behaved
         try:
@@ -209,21 +211,21 @@ def _get_last(ir):
         else:
             ir_node = revert_seq
 
-        return IRnode.from_list(ir_node)
+        return IRnode.from_list(ir_node, error_msg="user revert with reason")
 
     def parse_Assert(self):
         test_expr = Expr.parse_value_expr(self.stmt.test, self.context)
 
         if self.stmt.msg:
             return self._assert_reason(test_expr, self.stmt.msg)
         else:
-            return IRnode.from_list(["assert", test_expr])
+            return IRnode.from_list(["assert", test_expr], error_msg="user assert")
 
     def parse_Raise(self):
         if self.stmt.exc:
             return self._assert_reason(None, self.stmt.exc)
         else:
-            return IRnode.from_list(["revert", 0, 0])
+            return IRnode.from_list(["revert", 0, 0], error_msg="user raise")
 
     def _check_valid_range_constant(self, arg_ast_node):
         with self.context.range_scope():

vyper/ir/compile_ir.py

@@ -164,8 +164,10 @@ class Instruction(str):
     def __new__(cls, sstr, *args, **kwargs):
         return super().__new__(cls, sstr)
 
-    def __init__(self, sstr, source_pos=None):
+    def __init__(self, sstr, source_pos=None, error_msg=None):
+        self.error_msg = error_msg
         self.pc_debugger = False
+
         if source_pos is not None:
             self.lineno, self.col_offset, self.end_lineno, self.end_col_offset = source_pos
         else:
@@ -177,8 +179,9 @@ def apply_line_numbers(func):
     def apply_line_no_wrapper(*args, **kwargs):
         code = args[0]
         ret = func(*args, **kwargs)
+
         new_ret = [
-            Instruction(i, code.source_pos)
+            Instruction(i, code.source_pos, code.error_msg)
             if isinstance(i, str) and not isinstance(i, Instruction)
             else i
             for i in ret
@@ -732,7 +735,12 @@ def note_line_num(line_number_map, item, pos):
             offsets = (item.lineno, item.col_offset, item.end_lineno, item.end_col_offset)
         else:
             offsets = None
+
         line_number_map["pc_pos_map"][pos] = offsets
+
+        if item.error_msg is not None:
+            line_number_map["error_map"][pos] = item.error_msg
+
     added_line_breakpoint = note_breakpoint(line_number_map, item, pos)
     return added_line_breakpoint
 
@@ -944,6 +952,7 @@ def assembly_to_evm(assembly, start_pos=0, insert_vyper_signature=False):
         "pc_breakpoints": set(),
         "pc_jump_map": {0: "-"},
         "pc_pos_map": {},
+        "error_map": {},
     }
 
     posmap = {}

vyper/ir/optimizer.py

@@ -424,6 +424,7 @@ def _optimize(node: IRnode, parent: Optional[IRnode]) -> Tuple[bool, IRnode]:
     typ = node.typ
     location = node.location
     source_pos = node.source_pos
+    error_msg = node.error_msg
     annotation = node.annotation
     add_gas_estimate = node.add_gas_estimate
 
@@ -445,6 +446,7 @@ def finalize(val, args):
             typ=typ,
             location=location,
             source_pos=source_pos,
+            error_msg=error_msg,
             annotation=annotation,
             add_gas_estimate=add_gas_estimate,
         )