add vyper signature to bytecode (#2860)

add a vyper signature to vyper bytecode. this makes it easier to
identify on block explorers. the format is the CBOR encoding of
`{"vyper": (major,minor,patch)}`. for instance, a vyper 0.3.4
signature will be: `b"\xa1\x65vyper\x83\x00\x03\x04"`

vyper/compiler/output.py

@@ -184,7 +184,9 @@ def _build_asm(asm_list):
 
 
 def build_source_map_output(compiler_data: CompilerData) -> OrderedDict:
-    _, line_number_map = compile_ir.assembly_to_evm(compiler_data.assembly_runtime)
+    _, line_number_map = compile_ir.assembly_to_evm(
+        compiler_data.assembly_runtime, insert_vyper_signature=True
+    )
     # Sort line_number_map
     out = OrderedDict()
     for k in sorted(line_number_map.keys()):

vyper/compiler/phases.py

@@ -164,13 +164,13 @@ def assembly_runtime(self) -> list:
     @property
     def bytecode(self) -> bytes:
         if not hasattr(self, "_bytecode"):
-            self._bytecode = generate_bytecode(self.assembly)
+            self._bytecode = generate_bytecode(self.assembly, is_runtime=False)
         return self._bytecode
 
     @property
     def bytecode_runtime(self) -> bytes:
         if not hasattr(self, "_bytecode_runtime"):
-            self._bytecode_runtime = generate_bytecode(self.assembly_runtime)
+            self._bytecode_runtime = generate_bytecode(self.assembly_runtime, is_runtime=True)
         return self._bytecode_runtime
 
 
@@ -323,7 +323,7 @@ def _find_nested_opcode(assembly, key):
         return any(_find_nested_opcode(x, key) for x in sublists)
 
 
-def generate_bytecode(assembly: list) -> bytes:
+def generate_bytecode(assembly: list, is_runtime: bool = False) -> bytes:
     """
     Generate bytecode from assembly instructions.
 
@@ -337,4 +337,4 @@ def generate_bytecode(assembly: list) -> bytes:
     bytes
         Final compiled bytecode.
     """
-    return compile_ir.assembly_to_evm(assembly)[0]
+    return compile_ir.assembly_to_evm(assembly, insert_vyper_signature=is_runtime)[0]

vyper/ir/compile_ir.py

@@ -5,6 +5,7 @@
 from vyper.evm.opcodes import get_opcodes
 from vyper.exceptions import CodegenPanic, CompilerPanic
 from vyper.utils import MemoryPositions
+from vyper.version import version_tuple
 
 PUSH_OFFSET = 0x5F
 DUP_OFFSET = 0x7F
@@ -507,7 +508,7 @@ def _height_of(witharg):
         o.extend(["_sym_subcode_size", begincode, "_mem_deploy_start", "CODECOPY"])
 
         # calculate the len of runtime code
-        o.extend(["_sym_subcode_size"] + PUSH(padding) + ["ADD"])  # stack: len
+        o.extend(["_OFST", "_sym_subcode_size", padding])  # stack: len
         o.extend(["_mem_deploy_start"])  # stack: len mem_ofst
         o.extend(["RETURN"])
 
@@ -918,7 +919,7 @@ def _optimize_assembly(assembly):
 
 
 # Assembles assembly into EVM
-def assembly_to_evm(assembly, start_pos=0):
+def assembly_to_evm(assembly, start_pos=0, insert_vyper_signature=False):
     line_number_map = {
         "breakpoints": set(),
         "pc_breakpoints": set(),
@@ -930,6 +931,11 @@ def assembly_to_evm(assembly, start_pos=0):
     runtime_code, runtime_code_start, runtime_code_end = None, None, None
     pos = start_pos
 
+    bytecode_suffix = b""
+    if insert_vyper_signature:
+        # CBOR encoded: {"vyper": [major,minor,patch]}
+        bytecode_suffix += b"\xa1\x65vyper\x83" + bytes(list(version_tuple))
+
     # go through the code, resolving symbolic locations
     # (i.e. JUMPDEST locations) to actual code locations
     for i, item in enumerate(assembly):
@@ -977,7 +983,10 @@ def assembly_to_evm(assembly, start_pos=0):
             pos += 0
         elif isinstance(item, list):
             assert runtime_code is None, "Multiple subcodes"
-            runtime_code, sub_map = assembly_to_evm(item, start_pos=pos)
+            runtime_code, sub_map = assembly_to_evm(
+                item, start_pos=pos, insert_vyper_signature=True
+            )
+
             assert item[0].startswith("_DEPLOY_MEM_OFST_")
             ctor_mem_size = int(item[0][len("_DEPLOY_MEM_OFST_") :])
 
@@ -991,6 +1000,8 @@ def assembly_to_evm(assembly, start_pos=0):
         else:
             pos += 1
 
+    pos += len(bytecode_suffix)
+
     code_end = pos - start_pos
     posmap["_sym_code_end"] = code_end
     posmap["_mem_deploy_start"] = runtime_code_start
@@ -1045,6 +1056,8 @@ def assembly_to_evm(assembly, start_pos=0):
             # Should never reach because, assembly is create in _compile_to_assembly.
             raise Exception("Weird symbol in assembly: " + str(item))  # pragma: no cover
 
+    o += bytecode_suffix
+
     assert len(o) == pos - start_pos, (len(o), pos, start_pos)
     line_number_map["breakpoints"] = list(line_number_map["breakpoints"])
     line_number_map["pc_breakpoints"] = list(line_number_map["pc_breakpoints"])