[add] union

README.md

@@ -8,7 +8,7 @@ Rui Ueyama さんの「低レイヤを知りたい人のための C コンパイ
 - if, for, 配列, ポインター
 - グローバル変数の定義
 - 関数の定義、呼び出し
-- 構造体の基礎
+- struct, union
 - ブロックスコープ
 - 配列の初期化式
 - typedef, enum
@@ -18,7 +18,6 @@ Rui Ueyama さんの「低レイヤを知りたい人のための C コンパイ
 ## TODO
 
 - switch
-- union
 - 構造体の初期化式
 - \_Bool
 
@@ -52,8 +51,8 @@ $ make diff
                  | <storage_class>? "void"
                  | <storage_class>? "short"
                  | <storage_class>? "long" "long"? "int"?
-                 | <storage_class>? "struct" <ident>
-                 | <storage_class>? "struct" <ident> "{" <struct_declaration>* "}"
+                 | <storage_class>? ("struct" | "union") <ident>
+                 | <storage_class>? ("struct" | "union") <ident> "{" <struct_declaration>* "}"
                  | <storage_class>? "enum" <ident>
                  | <storage_class>? "enum" <ident>? "{" <enumerator_list> "}"
 <type_name> = <type_specifier> <pointer> <type_suffix>

src/codegen.c

@@ -167,7 +167,7 @@ static void gen_addr(Node *node) {
 }
 
 static void load(Type *ty) {
-    if (ty->kind == TYPE_ARRAY || ty->kind == TYPE_STRUCT) {
+    if (ty->kind == TYPE_ARRAY || ty->kind == TYPE_STRUCT || ty->kind == TYPE_UNION) {
         // アドレスのまま読みこむようにする
         return;
     }
@@ -232,7 +232,7 @@ static void gen(Node *node) {
         pop_rdi();
         pop();
         add_type(node->lhs);
-        if (node->type->kind == TYPE_STRUCT) {
+        if (node->type->kind == TYPE_STRUCT || node->type->kind == TYPE_UNION) {
             // メモリコピー
             for (int i = 0; i < node->type->size; i++) {
                 printf("  mov r8, [rdi+%d]\n", i);

src/debug.c

@@ -196,6 +196,8 @@ void print_type_kind(TypeKind kind) {
         fprintf(stderr, "TYPE_VOID");
     else if (kind == TYPE_STRUCT)
         fprintf(stderr, "TYPE_STRUCT");
+    else if (kind == TYPE_UNION)
+        fprintf(stderr, "TYPE_UNION");
     else if (kind == TYPE_ENUM)
         fprintf(stderr, "TYPE_ENUM");
     else
@@ -274,7 +276,7 @@ void debug_type(Type *ty, int depth) {
     print_type_kind(ty->kind);
     puts("");
 
-    if (ty->kind == TYPE_STRUCT) {
+    if (ty->kind == TYPE_STRUCT || ty->kind == TYPE_UNION) {
         recursion_line_printf(depth, "name -> %s\n", ty->name);
         recursion_line_printf(depth, "size -> %d\n", ty->size);
         for (Var *member = ty->member; member; member = member->next) {

src/kcc.h

@@ -77,6 +77,7 @@ enum TypeKind {
     TYPE_ARRAY,
     TYPE_VOID,
     TYPE_STRUCT,
+    TYPE_UNION,
     TYPE_ENUM,
 };
 
@@ -306,6 +307,8 @@ Vector *string_literal;
 Vector *funcs;  // Function型のVector
 Vector *struct_global_lists;
 Vector *struct_local_lists;
+Vector *union_global_lists;
+Vector *union_local_lists;
 Vector *enum_global_lists;
 Vector *enum_local_lists;  // 既出の列挙型
 Vector *typedef_alias;

src/main.c

@@ -6,6 +6,8 @@ static void init() {
     globals = NULL;        // グローバル変数の初期化
     struct_global_lists = new_vec();
     struct_local_lists = new_vec();
+    union_global_lists = new_vec();
+    union_local_lists = new_vec();
     enum_global_lists = new_vec();
     enum_local_lists = new_vec();
     funcs = new_vec();

src/parse.c

@@ -199,7 +199,8 @@ static void new_struct_member(Token *tok, Type *member_type, Type *struct_type)
     member->name = my_strndup(tok->str, tok->len);
     member->len = tok->len;
     member->type = member_type;
-    // offsetはalignmentを考慮するので後で決める
+    // offsetはalignmentを考慮するので後で決める。unionは0。
+    member->offset = 0;
     struct_type->member = member;
     struct_type->size += member_type->size;
 }
@@ -287,6 +288,14 @@ static Type *find_gstruct_type(char *name) {
     return find_aggregate_type(name, struct_global_lists);
 }
 
+static Type *find_lunion_type(char *name) {
+    return find_aggregate_type(name, union_local_lists);
+}
+
+static Type *find_gunion_type(char *name) {
+    return find_aggregate_type(name, union_global_lists);
+}
+
 static Type *find_lenum_type(char *name) {
     return find_aggregate_type(name, enum_local_lists);
 }
@@ -567,12 +576,12 @@ static void create_lvar_from_params(Var *params) {
     create_lvar_from_params(params->next);
 }
 
-bool is_already_defined_global_obj(Token *tok) {
+static bool is_already_defined_global_obj(Token *tok) {
     char *name = my_strndup(tok->str, tok->len);
     return find_gvar(tok) || find_func(name);
 }
 
-bool is_same_params(Var *params1, Var *params2) {
+static bool is_same_params(Var *params1, Var *params2) {
     for (Var *v1 = params1, *v2 = params2;; v1 = v1->next, v2 = v2->next) {
         if (v1 == NULL || v2 == NULL) {
             // NULL == NULL -> params1とparams2は等しい
@@ -585,13 +594,56 @@ bool is_same_params(Var *params1, Var *params2) {
     }
 }
 
-bool has_lvar_in_all_params(Var *params) {
+static bool has_lvar_in_all_params(Var *params) {
     for (Var *v = params; v; v = v->next) {
         if (v->is_only_type) return false;
     }
     return true;
 }
 
+static Var *reverse_linked_list_var(Var *cur) {
+    Var *reversed = NULL;
+    while (cur) {
+        Var *tmp = cur->next;
+        cur->next = reversed;
+        reversed = cur;
+        cur = tmp;
+    }
+    return reversed->next;
+}
+
+static Type *struct_defined_or_forward(Type *type) {
+    if (is_global) {
+        Type *t = find_gstruct_type(type->name);
+        if (t != NULL && !t->is_forward) {
+            error("find_gstruct_type() failure: %s構造体は既に宣言済みです。", type->name);
+        }
+        return t;
+    } else {
+        Type *t = find_lstruct_type(type->name);
+        if (t != NULL && !t->is_forward) {
+            error("find_lstruct_type() failure: %s構造体は既に宣言済みです。", type->name);
+        }
+        return t;
+    }
+}
+
+static Type *union_defined_or_forward(Type *type) {
+    if (is_global) {
+        Type *t = find_gunion_type(type->name);
+        if (t != NULL && !t->is_forward) {
+            error("find_gunion_type() failure: %s構造体は既に宣言済みです。", type->name);
+        }
+        return t;
+    } else {
+        Type *t = find_lunion_type(type->name);
+        if (t != NULL && !t->is_forward) {
+            error("find_lunion_type() failure: %s構造体は既に宣言済みです。", type->name);
+        }
+        return t;
+    }
+}
+
 /*************************************/
 /******                         ******/
 /******        NEW_NODE         ******/
@@ -840,7 +892,7 @@ void program() {
             if (fn != NULL) vec_push(funcs, fn);
             is_global = true;
         } else {
-            if ((type->kind == TYPE_STRUCT || type->kind == TYPE_ENUM) &&
+            if ((type->kind == TYPE_STRUCT || type->kind == TYPE_UNION || type->kind == TYPE_ENUM) &&
                 consume(';')) {
                 // 構造体・列強型の作成 or 宣言
                 continue;
@@ -1046,6 +1098,7 @@ static Type *pointer(Type *type) {
  *                  | <storage_class>? "enum" <ident>? "{" <enumerator_list> "}"
  */
 static Type *type_specifier() {
+    // storage class
     if (consume(TK_TYPEDEF))
         current_storage = STORAGE_TYPEDEF;
     else if (consume(TK_EXTERN)) {
@@ -1070,19 +1123,8 @@ static Type *type_specifier() {
     }
 
     if (type->kind == TYPE_STRUCT && consume('{')) {
-        if (is_global) {
-            Type *t = find_gstruct_type(type->name);
-            if (t != NULL && !t->is_forward) {
-                error("find_gstruct_type() failure: %s構造体は既に宣言済みです。", type->name);
-            }
-            if (t) type = t;
-        } else {
-            Type *t = find_lstruct_type(type->name);
-            if (t != NULL && !t->is_forward) {
-                error("find_lstruct_type() failure: %s構造体は既に宣言済みです。", type->name);
-            }
-            if (t) type = t;
-        }
+        Type *t = struct_defined_or_forward(type);
+        if (t != NULL) type = t;
 
         if (!type->is_forward) {
             vec_push(is_global ? struct_global_lists : struct_local_lists, type);
@@ -1096,16 +1138,31 @@ static Type *type_specifier() {
             type = struct_declaration(type);
         }
         // 定義した順に並べ直す
-        Var *reverse_member = NULL;
-        while (type->member) {
-            Var *tmp = type->member->next;
-            type->member->next = reverse_member;
-            reverse_member = type->member;
-            type->member = tmp;
+        type->member = reverse_linked_list_var(type->member);
+        // memberのアライメントを考慮したoffsetを決定する
+        apply_align_struct(type);
+
+        return type;
+    }
+
+    if (type->kind == TYPE_UNION && consume('{')) {
+        Type *t = union_defined_or_forward(type);
+        if (t != NULL) type = t;
+
+        if (!type->is_forward) {
+            vec_push(is_global ? union_global_lists : union_local_lists, type);
+        } else {
+            type->is_forward = false;
         }
-        type->member = reverse_member->next;
 
-        // memberのアライメントを考慮したoffsetを決定する
+        // 構造体のメンバーの宣言
+        type->member = memory_alloc(sizeof(Var));
+        while (!consume('}')) {
+            type = struct_declaration(type);
+        }
+        // 定義した順に並べ直す
+        type->member = reverse_linked_list_var(type->member);
+
         apply_align_struct(type);
 
         return type;
@@ -1142,6 +1199,19 @@ static Type *type_specifier() {
         type = stype;
     }
 
+    if (type->kind == TYPE_UNION) {
+        Type *stype = find_lunion_type(type->name);
+        if (stype == NULL) {
+            stype = find_gunion_type(type->name);
+            if (stype == NULL) {
+                type->is_forward = true;
+                stype = type;
+                vec_push(is_global ? union_global_lists : union_local_lists, stype);
+            }
+        }
+        type = stype;
+    }
+
     if (type->kind == TYPE_ENUM) {
         Type *etype = find_enum_type(type->name);
         if (etype == NULL) {
@@ -1364,6 +1434,7 @@ static Function *func_define(Type *type) {
 
     locals = memory_alloc(sizeof(Var));
     struct_local_lists = new_vec();  // 関数毎に構造体を初期化
+    union_local_lists = new_vec();
     enum_local_lists = new_vec();
     locals->offset = 0;
     start_local_scope();
@@ -1396,7 +1467,7 @@ static Node *compound_stmt() {
         Node *n;
         if (consume_is_type_nostep(token)) {
             Type *type = type_specifier();
-            if ((type->kind == TYPE_STRUCT || type->kind == TYPE_ENUM) &&
+            if ((type->kind == TYPE_STRUCT || type->kind == TYPE_UNION || type->kind == TYPE_ENUM) &&
                 consume_nostep(';')) {
                 // 構造体か列挙型の作成 or 宣言
                 continue;
@@ -1471,7 +1542,7 @@ static Node *stmt() {
         if (consume_is_type_nostep(token)) {
             // <declaration>
             Type *type = type_specifier();
-            if ((type->kind == TYPE_STRUCT || type->kind == TYPE_ENUM) &&
+            if ((type->kind == TYPE_STRUCT || type->kind == TYPE_UNION || type->kind == TYPE_ENUM) &&
                 consume_nostep('{')) {
                 // 構造体か列挙型の作成
                 error("stmt() failure: failure");
@@ -1858,8 +1929,8 @@ static Node *postfix() {
             Token *tok = token;
             expect(TK_IDENT);
             add_type(node);
-            if (node->type->kind != TYPE_STRUCT) {
-                error("postfix() failure: struct型ではありません。");
+            if (node->type->kind != TYPE_STRUCT && node->type->kind != TYPE_UNION) {
+                error("postfix() failure: struct or union型ではありません。");
             }
             Var *member = node->type->member;
             while (member) {

src/token.c

@@ -416,6 +416,24 @@ Token *tokenize(char *p) {
             }
         }
 
+        if (strncmp(p, "union", 5) == 0 && !is_alnum(p[5])) {
+            cur = new_token(TK_TYPE, cur, p, 5);
+            cur->type = new_type(TYPE_UNION);
+            p += 5;
+
+            skip_space(&p);
+
+            if (is_alpha(*p)) {
+                char *q = p;
+                str_advanve(&p);
+                cur->type->name = my_strndup(q, p - q);
+                continue;
+            } else {
+                /* TODO: 無名共用体 */
+                error_at(p, "tokenize() failure: unionの型名が存在しません。");
+            }
+        }
+
         if (strncmp(p, "enum", 4) == 0 && !is_alnum(p[4])) {
             cur = new_token(TK_TYPE, cur, p, 4);
             cur->type = new_type(TYPE_ENUM);