Fix detection of main wrapper function

Note that the expectations for test_metadce_mem grew because we were
previously mis-detecting mem.c as not read params, when in fact it does.

Fixes: #17720

Author: sbc100

test/core/test_main_reads_args.c

@@ -0,0 +1,5 @@
+extern int real_main(int, char**);
+
+int main(int argc, char** argv) {
+  return real_main(argc, argv);
+}

test/core/test_main_reads_args.out

@@ -0,0 +1,2 @@
+argc: 1
+argv[0]: test_main_reads_args.js

test/core/test_main_reads_args_real.c

@@ -0,0 +1,11 @@
+#include <stdio.h>
+#include <string.h>
+
+int real_main(int argc, char** argv) {
+  printf("argc: %d\n", argc);
+  const char* arg0 = strrchr(argv[0], '/');
+  if (!arg0)
+    arg0 = strrchr(argv[0], '\\');
+  printf("argv[0]: %s\n", arg0 + 1);
+  return 0;
+}

test/other/metadce/test_metadce_mem_O3.exports

@@ -2,3 +2,4 @@ b
 c
 d
 e
+f

test/other/metadce/test_metadce_mem_O3.funcs

@@ -1,3 +1,4 @@
 $__wasm_call_ctors
 $main
 $sbrk
+$stackAlloc

test/other/metadce/test_metadce_mem_O3.size

@@ -1 +1 @@
-5935
+5957

test/other/metadce/test_metadce_mem_O3_grow.exports

@@ -2,3 +2,4 @@ b
 c
 d
 e
+f

test/other/metadce/test_metadce_mem_O3_grow.funcs

@@ -1,3 +1,4 @@
 $__wasm_call_ctors
 $main
 $sbrk
+$stackAlloc

test/other/metadce/test_metadce_mem_O3_grow.size

@@ -1 +1 @@
-5936
+5958

test/test_core.py

@@ -9489,6 +9489,10 @@ def test_js_library_i64_params(self):
     self.emcc_args += ['--js-library=' + test_file('core/js_library_i64_params.js')]
     self.do_core_test('js_library_i64_params.c')
 
+  def test_main_reads_args(self):
+    self.run_process([EMCC, '-c', test_file('core/test_main_reads_args_real.c'), '-o', 'real.o'] + self.get_emcc_args(ldflags=False))
+    self.do_core_test('test_main_reads_args.c', emcc_args=['real.o', '-sEXIT_RUNTIME'])
+
 
 # Generate tests for everything
 def make_run(name, emcc_args, settings=None, env=None, node_args=None, require_v8=False, v8_args=None):

tools/extract_metadata.py

@@ -18,24 +18,29 @@ def skip_function_header(module):
 
 
 def is_orig_main_wrapper(module, function):
+  module.get_types()
+  module.get_function_types()
   module.seek(function.offset)
   skip_function_header(module)
   end = function.offset + function.size
   while module.tell() != end:
     opcode = module.read_byte()
     try:
       opcode = OpCode(opcode)
-    except ValueError as e:
-      print(e)
+    except ValueError:
       return False
     if opcode == OpCode.CALL:
-      module.read_uleb()  # callee
+      callee = module.read_uleb()
+      callee_type = module.get_function_type(callee)
+      if len(callee_type.params) != 0:
+        return False
     elif opcode in (OpCode.LOCAL_GET, OpCode.LOCAL_SET):
       module.read_uleb()  # local index
     elif opcode in (OpCode.END, OpCode.RETURN):
       pass
     else:
       # Any other opcodes and we assume this not a simple wrapper
+      print("Uknown opcode: %s" % opcode)
       return False
 
   assert opcode == OpCode.END

tools/webassembly.py

@@ -272,6 +272,7 @@ def sections(self):
       yield Section(section_type, section_size, section_offset, name)
       offset = section_offset + section_size
 
+  @memoize
   def get_types(self):
     type_section = self.get_section(SecType.TYPE)
     if not type_section:
@@ -280,17 +281,21 @@ def get_types(self):
     num_types = self.read_uleb()
     types = []
     for _ in range(num_types):
-      params = []
-      returns = []
       type_form = self.read_byte()
       assert type_form == 0x60
+
+      params = []
       num_params = self.read_uleb()
       for _ in range(num_params):
         params.append(self.read_type())
+
+      returns = []
       num_returns = self.read_uleb()
       for _ in range(num_returns):
         returns.append(self.read_type())
+
       types.append(FuncType(params, returns))
+
     return types
 
   def parse_features_section(self):
@@ -503,39 +508,55 @@ def get_tables(self):
 
     return tables
 
+  @memoize
+  def get_function_types(self):
+    function_section = self.get_section(SecType.FUNCTION)
+    if not function_section:
+      return []
+
+    self.seek(function_section.offset)
+    num_types = self.read_uleb()
+    func_types = []
+    for _ in range(num_types):
+      func_types.append(self.read_uleb())
+    return func_types
+
   def has_name_section(self):
     return self.get_custom_section('name') is not None
 
   @once
   def _calc_indexes(self):
-    self.num_imported_funcs = 0
-    self.num_imported_globals = 0
-    self.num_imported_memories = 0
-    self.num_imported_tables = 0
-    self.num_imported_tags = 0
+    self.imports_by_kind = {}
     for i in self.get_imports():
-      if i.kind == ExternType.FUNC:
-        self.num_imported_funcs += 1
-      elif i.kind == ExternType.GLOBAL:
-        self.num_imported_globals += 1
-      elif i.kind == ExternType.MEMORY:
-        self.num_imported_memories += 1
-      elif i.kind == ExternType.TABLE:
-        self.num_imported_tables += 1
-      elif i.kind == ExternType.TAG:
-        self.num_imported_tags += 1
-      else:
-        assert False, 'unhandled export type: %s' % i.kind
+      self.imports_by_kind.setdefault(i.kind, [])
+      self.imports_by_kind[i.kind].append(i)
+
+  def num_imported_funcs(self):
+    self._calc_indexes()
+    return len(self.imports_by_kind.get(ExternType.FUNC, []))
+
+  def num_imported_globals(self):
+    self._calc_indexes()
+    return len(self.imports_by_kind.get(ExternType.GLOBAL, []))
 
   def get_function(self, idx):
     self._calc_indexes()
-    assert idx >= self.num_imported_funcs
-    return self.get_functions()[idx - self.num_imported_funcs]
+    assert idx >= self.num_imported_funcs()
+    return self.get_functions()[idx - self.num_imported_funcs()]
 
   def get_global(self, idx):
     self._calc_indexes()
-    assert idx >= self.num_imported_globals
-    return self.get_globals()[idx - self.num_imported_globals]
+    assert idx >= self.num_imported_globals()
+    return self.get_globals()[idx - self.num_imported_globals()]
+
+  def get_function_type(self, idx):
+    self._calc_indexes()
+    if idx < self.num_imported_funcs():
+      imp = self.imports_by_kind[ExternType.FUNC][idx]
+      func_type = imp.type
+    else:
+      func_type = self.get_function_types()[idx - self.num_imported_funcs()]
+    return self.get_types()[func_type]
 
 
 def parse_dylink_section(wasm_file):