Enable constructor/destructor on WASM

Output `WASM_INIT_FUNCS` in linking custom section: https://github.com/WebAssembly/tool-conventions/blob/main/Linking.md#start-section On WASM, no `.init_array` nor `.fini_array`, call `__wasm_call_ctors` function on `_start` to call constructors. Generate a function to call destructors and register the function using `__cxa_atexit`.
tyfkda · Sep 14, 2024 · 670952b · 670952b
1 parent 1b73957
commit 670952b
Show file tree

Hide file tree

Showing 14 changed files with 393 additions and 107 deletions.
diff --git a/Makefile b/Makefile
@@ -236,7 +236,7 @@ wcc-libs:
 	$(MAKE) CC=../wcc AR=llvm-ar -C libsrc wcc-libs
 
 WCCLD_SRCS:=$(DEBUG_DIR)/wcc-ld.c $(WCC_DIR)/wasm_linker.c \
-	$(WCC_DIR)/wcc_util.c $(WCC_DIR)/emit_wasm.c $(WCC_DIR)/traverse.c $(WCC_DIR)/traverse_setjmp.c \
+	$(WCC_DIR)/wcc_util.c $(WCC_DIR)/emit_wasm.c $(WCC_DIR)/gen_wasm.c $(WCC_DIR)/traverse.c $(WCC_DIR)/traverse_setjmp.c \
 	$(wildcard $(CC1_FE_DIR)/*.c) \
 	$(UTIL_DIR)/util.c $(UTIL_DIR)/table.c $(UTIL_DIR)/archive.c
 WCCLD_OBJS:=$(addprefix $(WCC_OBJ_DIR)/,$(notdir $(WCCLD_SRCS:.c=.o)))

diff --git a/libsrc/_wasm/crt0/_start.c b/libsrc/_wasm/crt0/_start.c
@@ -1,24 +1,9 @@
 #include "alloca.h"  // alloca
-#include "stdio.h"   // fflush
-#include "stdlib.h"  // atexit, exit
+#include "stdlib.h"  // exit
 
-#include "../../stdio/_fileman.h"
 #include "../wasi.h"
 
-extern FILEMAN __fileman;
-
-inline void __flush_all_files(void) {
-  fflush(stdout);
-  fflush(stderr);
-
-  struct FILE **files = __fileman.opened;
-  for (int i = 0, length = __fileman.length; i < length; ++i)
-    fflush(files[i]);
-}
-
-static void _atexit_proc(void) {
-  __flush_all_files();
-}
+extern void __wasm_call_ctors(void);
 
 int __max_preopen_fd = 3;
 
@@ -55,7 +40,8 @@ void _start(void) {
 
   __max_preopen_fd = find_preopens();
 
-  atexit(_atexit_proc);
+  __wasm_call_ctors();
+
   int ec = main(argc, argv);
   exit(ec);
 #undef main

diff --git a/libsrc/stdlib/__cxa_atexit.c b/libsrc/stdlib/__cxa_atexit.c
@@ -0,0 +1,39 @@
+#include "stdlib.h"
+#include "_exit.h"
+
+#define MAX  (8)
+
+typedef struct {
+  void (*dtor)(void*);
+  void *arg;
+  void *dso_handle;
+} CxaAtexitBuf;
+
+static CxaAtexitBuf buf[MAX];
+static int count;
+
+void *__dso_handle;
+
+int __cxa_atexit(void (*func)(void*), void *arg, void *d) {
+  if (count >= MAX)
+    return 1;
+  CxaAtexitBuf *p = &buf[count++];
+  p->dtor = func;
+  p->arg = arg;
+  p->dso_handle = d;
+  return 0;
+}
+
+static void call_cxa_atexit_funcs(void) {
+  for (CxaAtexitBuf *p = &buf[count]; p > buf; ) {
+    --p;
+    (*p->dtor)(p->arg);
+  }
+}
+
+__attribute__((constructor))
+static void register_atexit(void) {
+  static OnExitChain chain = {NULL, call_cxa_atexit_funcs};
+  chain.next = __on_exit_chain;
+  __on_exit_chain = &chain;
+}
diff --git a/libsrc/stdlib/exit.c b/libsrc/stdlib/exit.c
@@ -28,6 +28,12 @@ OnExitChain *__on_exit_chain;
 void exit(int code) {
   // TODO: Guard multiple calls
 
+#if defined(__WASM)
+  // On wcc, if there is a indirect function call but no actual function reference exists,
+  // table/elem section are not emitted and cause a load error.
+  // To avoid this, make sure a function reference exists and table/elem section are emitted.
+  (void)exit;
+#endif
   OnExitChain *chain = __on_exit_chain;
   __on_exit_chain = NULL;
   for (; chain != NULL; chain = chain->next) {

diff --git a/src/cc/frontend/parser.c b/src/cc/frontend/parser.c
@@ -948,79 +948,148 @@ static Declaration *parse_declaration(void) {
   return NULL;
 }
 
-#if XCC_TARGET_PLATFORM == XCC_PLATFORM_APPLE
-static void generate_dtor_func(Vector *decls) {
-  const Name *destructor_name = alloc_name("destructor", NULL, false);
-  Vector *dtors = NULL;
-  for (int i = 0, len = decls->len; i < len; ++i) {
-    Declaration *decl = decls->data[i];
-    if (decl == NULL || decl->kind != DCL_DEFUN)
-      continue;
-    Function *func = decl->defun.func;
-    if (func->attributes != NULL) {
-      if (table_try_get(func->attributes, destructor_name, NULL)) {
-        if (dtors == NULL)
-          dtors = new_vector();
-        vec_push(dtors, func);
-      }
-    }
+#if XCC_TARGET_PLATFORM == XCC_PLATFORM_APPLE || XCC_TARGET_PLATFORM == XCC_PLATFORM_WASI
+static Function *generate_dtor_caller_func(Vector *dtors) {
+  // Generate function:
+  //  void dtor_caller(void*) {
+  //    dtor1();
+  //    ...
+  //  }
+
+  Vector *param_types = new_vector();
+  vec_push(param_types, &tyVoidPtr);
+  Type *functype = new_func_type(&tyVoid, param_types, false);
+
+  Vector *top_vars = new_vector();
+  var_add(top_vars, alloc_name("p", NULL, false), &tyVoidPtr, VS_PARAM);
+
+  const Token *functok = alloc_dummy_ident();
+  Table *attributes = NULL;
+  Function *func = define_func(functype, functok, top_vars, VS_STATIC, attributes);
+
+  assert(curfunc == NULL);
+  assert(is_global_scope(curscope));
+  curfunc = func;
+
+  func->scopes = new_vector();
+  Scope *scope = enter_scope(func);
+  scope->vars = top_vars;
+
+  // Construct function body: call destructors.
+  Vector *stmts = new_vector();
+  for (int i = 0; i < dtors->len; ++i) {
+    Function *dtor = dtors->data[i];
+    const Token *token = NULL;
+    Vector *args = new_vector();
+    Expr *func = new_expr_variable(dtor->name, dtor->type, token, global_scope);
+    Expr *call = new_expr_funcall(token, func, args);
+    vec_push(stmts, new_stmt_expr(call));
   }
-  if (dtors == NULL)
-    return;
 
-  // Declare: extern void __dso_handle();
+  func->body_block = new_stmt_block(NULL, stmts, scope, NULL);
+
+  exit_scope();
+  curfunc = NULL;
+
+  return func;
+}
+
+static Function *generate_dtor_register_func(Function *dtor_caller_func) {
+  // Generate function:
+  //  __attribute__((constructor))
+  //  void dtor_register(void) {
+  //    __cxa_atexit(dtor_caller, NULL, &__dso_handle);
+  //  }
+
+  // Declare: extern void *__dso_handle;
   const Name *dso_handle_name = alloc_name("__dso_handle", NULL, false);
   scope_add(global_scope, dso_handle_name, &tyVoidPtr, VS_EXTERN);
 
-  // Declare: extern int __cxa_atexit(void*, void*, void*);
+  // Declare: extern int __cxa_atexit(void (*)(void*), void*, void*);
   const Name *cxa_atexit_name = alloc_name("__cxa_atexit", NULL, false);
-  Type *cxa_atexit_functype = new_func_type(&tyVoid, NULL, false);
-  define_func(cxa_atexit_functype, alloc_ident(cxa_atexit_name, NULL, cxa_atexit_name->chars, NULL), new_vector(), VS_EXTERN, NULL);
+  Type *cxa_atexit_functype;
+  {
+    Vector *cxa_atexit_param_types = new_vector();
+    vec_push(cxa_atexit_param_types, &tyVoidPtr);  // void (*func)(void*)
+    vec_push(cxa_atexit_param_types, &tyVoidPtr);
+    vec_push(cxa_atexit_param_types, &tyVoidPtr);
+
+    Vector *param_vars = new_vector();
+    var_add(param_vars, alloc_name("x", NULL, false), &tyVoidPtr, VS_PARAM);
+    var_add(param_vars, alloc_name("y", NULL, false), &tyVoidPtr, VS_PARAM);
+    var_add(param_vars, alloc_name("z", NULL, false), &tyVoidPtr, VS_PARAM);
+
+    cxa_atexit_functype = new_func_type(&tyInt, cxa_atexit_param_types, false);
+    define_func(cxa_atexit_functype, alloc_ident(cxa_atexit_name, NULL, cxa_atexit_name->chars, NULL), param_vars, VS_EXTERN, NULL);
+  }
 
-  // Generate function that calls destructors. It is marked as `constructor` to register thenm.
+  Vector *param_types = new_vector();
+  Type *functype = new_func_type(&tyVoid, param_types, false);
+  Vector *top_vars = new_vector();
 
-  Type *functype = new_func_type(&tyVoid, new_vector(), false);
   const Token *functok = alloc_dummy_ident();
   Table *attributes = alloc_table();
   assert(attributes != NULL);
   table_put(attributes, alloc_name("constructor", NULL, false), NULL);
-  Function *func = define_func(functype, functok, new_vector(), VS_STATIC, attributes);
-
+  Function *func = define_func(functype, functok, top_vars, VS_STATIC, attributes);
   func->flag |= FUNCF_HAS_FUNCALL;  // Make frame pointer saved on prologue.
 
   assert(curfunc == NULL);
   assert(is_global_scope(curscope));
   curfunc = func;
 
-  Vector *top_vars = new_vector();
   func->scopes = new_vector();
   Scope *scope = enter_scope(func);
   scope->vars = top_vars;
 
-  // Construct function body: call destructors.
-  Vector *cxa_atexit_param_types = new_vector();
-  vec_push(cxa_atexit_param_types, &tyVoidPtr);  // void (*func)(void*)
-  vec_push(cxa_atexit_param_types, &tyVoidPtr);
-  vec_push(cxa_atexit_param_types, &tyVoidPtr);
-  Type *cxa_atexit_type = new_func_type(&tyInt, cxa_atexit_param_types, false);
   Vector *stmts = new_vector();
-  for (int i = 0; i < dtors->len; ++i) {
-    Function *dtor = dtors->data[i];
-    // __cxa_atexit(func, NULL, &__dso_handle);
-    const Token *token = NULL;
-    Vector *args = new_vector();
-    vec_push(args, new_expr_variable(dtor->name, dtor->type, token, global_scope));
-    vec_push(args, new_expr_fixlit(&tyVoidPtr, token, 0));
-    vec_push(args, new_expr_unary(EX_REF, &tyVoidPtr, NULL, new_expr_variable(dso_handle_name, &tyVoidPtr, token, global_scope)));
-    Expr *func = new_expr_variable(cxa_atexit_name, cxa_atexit_type, token, global_scope);
-    Expr *call = new_expr_funcall(token, func, args);
-    vec_push(stmts, new_stmt_expr(call));
-  }
+  const Token *token = NULL;
+  Vector *args = new_vector();
+  vec_push(args, make_refer(token, new_expr_variable(dtor_caller_func->name, dtor_caller_func->type, token, global_scope)));
+  vec_push(args, new_expr_fixlit(&tyVoidPtr, token, 0));
+  vec_push(args, new_expr_unary(EX_REF, &tyVoidPtr, NULL, new_expr_variable(dso_handle_name, &tyVoidPtr, token, global_scope)));
+  // __cxa_atexit(dtor_caller, NULL, &__dso_handle);
+  vec_push(stmts, new_stmt_expr(
+      new_expr_funcall(token, new_expr_variable(cxa_atexit_name, cxa_atexit_functype, token, global_scope), args)));
+
   func->body_block = new_stmt_block(NULL, stmts, scope, NULL);
 
+  exit_scope();
   curfunc = NULL;
 
-  vec_push(decls, new_decl_defun(func));
+  return func;
+}
+
+static void modify_dtor_func(Vector *decls) {
+  const Name *destructor_name = alloc_name("destructor", NULL, false);
+  Vector *dtors = NULL;
+  for (int i = 0, len = decls->len; i < len; ++i) {
+    Declaration *decl = decls->data[i];
+    if (decl == NULL || decl->kind != DCL_DEFUN)
+      continue;
+    Function *func = decl->defun.func;
+    if (func->attributes != NULL) {
+      if (table_try_get(func->attributes, destructor_name, NULL)) {
+        const Type *type = func->type;
+        if (type->func.params == NULL || type->func.params->len > 0 || type->func.ret->kind != TY_VOID) {
+          const Token *token = func->body_block != NULL ? func->body_block->token : NULL;
+          parse_error(PE_NOFATAL, token, "destructor must have no parameters and return void");
+        } else {
+          if (dtors == NULL)
+            dtors = new_vector();
+          vec_push(dtors, func);
+        }
+      }
+    }
+  }
+  if (dtors == NULL)
+    return;
+
+  Function *caller_func = generate_dtor_caller_func(dtors);
+  vec_push(decls, new_decl_defun(caller_func));
+
+  Function *register_func = generate_dtor_register_func(caller_func);
+  vec_push(decls, new_decl_defun(register_func));
 }
 #endif
 
@@ -1033,7 +1102,7 @@ void parse(Vector *decls) {
       vec_push(decls, decl);
   }
 
-#if XCC_TARGET_PLATFORM == XCC_PLATFORM_APPLE
-  generate_dtor_func(decls);
+#if XCC_TARGET_PLATFORM == XCC_PLATFORM_APPLE || XCC_TARGET_PLATFORM == XCC_PLATFORM_WASI
+  modify_dtor_func(decls);
 #endif
 }
diff --git a/src/wcc/emit_wasm.c b/src/wcc/emit_wasm.c
@@ -660,6 +660,21 @@ static void emit_linking_section(EmitWasm *ew) {
     data_close_chunk(&linking_section, -1);
   }
 
+  if (init_funcs != NULL) {
+    data_push(&linking_section, LT_WASM_INIT_FUNCS);  // subsec type
+    data_open_chunk(&linking_section);  // Payload start.
+    data_uleb128(&linking_section, -1, init_funcs->len);  // Count
+    for (int i = 0; i < init_funcs->len; ++i) {
+      Function *func = init_funcs->data[i];
+      FuncInfo *info;
+      info = table_get(&func_info_table, func->name);
+      assert(info != NULL);
+      data_uleb128(&linking_section, -1, 65535);  // Priority
+      data_uleb128(&linking_section, -1, info->index);  // Symbol index
+    }
+    data_close_chunk(&linking_section, -1);
+  }
+
   if (linking_section.len > 0) {
     data_close_chunk(&linking_section, -1);
 

diff --git a/src/wcc/gen_wasm.c b/src/wcc/gen_wasm.c
@@ -22,7 +22,7 @@
 #include "wasm.h"
 #include "wasm_obj.h"
 
-#define CODE  (((FuncExtra*)curfunc->extra)->code)
+#define CODE  (curcodeds)
 
 #define ADD_LEB128(x)  data_leb128(CODE, -1, x)
 #define ADD_ULEB128(x) data_uleb128(CODE, -1, x)
@@ -34,6 +34,8 @@
 
 static void gen_lval(Expr *expr);
 
+DataStorage *curcodeds;
+
 void add_code(const unsigned char *buf, size_t size) {
   data_append(CODE, buf, size);
 }
@@ -1611,16 +1613,19 @@ static void gen_defun(Function *func) {
   if (func->scopes == NULL)  // Prototype definition
     return;
 
-  curfunc = func;
-
   DataStorage *code = malloc_or_die(sizeof(*code));
   data_init(code);
+  data_open_chunk(code);
+
   FuncExtra *extra = func->extra;
   assert(extra != NULL);
   extra->code = code;
   func->extra = extra;
   uint32_t frame_size = allocate_local_variables(func, code);
 
+  curfunc = func;
+  curcodeds = code;
+
   // Prologue
 
   const Type *functype = func->type;
@@ -1701,10 +1706,11 @@ static void gen_defun(Function *func) {
   ADD_CODE(OP_END);
 
   size_t before = code->len;
-  data_uleb128(code, 0, code->len);  // Insert code size at the top.
+  data_close_chunk(code, -1);
   extra->offset = code->len - before;
 
   curfunc = NULL;
+  curcodeds = NULL;
   assert(cur_depth == 0);
 }