extract_metadata: Improve detection of mainReadsParams (#17702)

In non-optimizing modes llvm can generate extra local.set/local.get
instructions which was throwing off this function.

Author: sbc100

test/other/metadce/test_metadce_hello_O0.jssize

@@ -1 +1 @@
-27498
+25683

test/other/metadce/test_metadce_mem_O3.exports

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

test/other/metadce/test_metadce_mem_O3.funcs

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

test/other/metadce/test_metadce_mem_O3.jssize

@@ -1 +1 @@
-6153
+5284

test/other/metadce/test_metadce_mem_O3.size

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

test/other/metadce/test_metadce_mem_O3_grow.exports

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

test/other/metadce/test_metadce_mem_O3_grow.funcs

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

test/other/metadce/test_metadce_mem_O3_grow.jssize

@@ -1 +1 @@
-6484
+5606

test/other/metadce/test_metadce_mem_O3_grow.size

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

test/other/test_unoptimized_code_size.js.size

@@ -1 +1 @@
-83875
+83653

test/other/test_unoptimized_code_size_no_asserts.js.size

@@ -1 +1 @@
-56428
+56206

test/other/test_unoptimized_code_size_strict.js.size

@@ -1 +1 @@
-65940
+65556

test/test_other.py

@@ -10055,7 +10055,7 @@ def test_main_reads_params(self):
     no = os.path.getsize('no.js')
     create_file('yes.c', '''
       int main(int argc, char **argv) {
-        return argc;
+        return (long)argv[argc-1];
       }
     ''')
     self.run_process([EMCC, 'yes.c', '-O3', '-o', 'yes.js'])

tools/extract_metadata.py

@@ -17,7 +17,7 @@ def skip_function_header(module):
     num_local_decls -= 1
 
 
-def is_wrapper_function(module, function):
+def is_orig_main_wrapper(module, function):
   module.seek(function.offset)
   skip_function_header(module)
   end = function.offset + function.size
@@ -29,11 +29,15 @@ def is_wrapper_function(module, function):
       print(e)
       return False
     if opcode == OpCode.CALL:
-      callee = module.read_uleb()  # noqa
-    elif opcode == OpCode.END:
-      break
+      module.read_uleb()  # callee
+    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
       return False
+
   assert opcode == OpCode.END
   return True
 
@@ -199,17 +203,20 @@ def get_asm_strings(module, export_map):
 
 def get_main_reads_params(module, export_map):
   if settings.STANDALONE_WASM:
-    return 1
+    return True
 
   main = export_map.get('main') or export_map.get('__main_argc_argv')
   if not main or main.kind != webassembly.ExternType.FUNC:
-    return 0
+    return False
 
   main_func = module.get_function(main.index)
-  if is_wrapper_function(module, main_func):
-    return 0
-  else:
-    return 1
+  if is_orig_main_wrapper(module, main_func):
+    # If main is simple wrapper function then we know that __orginial_main
+    # doesn't read arguments.
+    return False
+
+  # By default assume params are read
+  return True
 
 
 def get_named_globals(module, exports):