Add union

chibicc.h

@@ -145,6 +145,7 @@ typedef enum {
   TY_FUNC,
   TY_ARRAY,
   TY_STRUCT,
+  TY_UNION,
 } TypeKind;
 
 struct Type {

parse.c

@@ -27,7 +27,7 @@ struct VarScope {
   Var *var;
 };
 
-// Scope for struct tags
+// Scope for struct or union tags
 typedef struct TagScope TagScope;
 struct TagScope {
   TagScope *next;
@@ -72,6 +72,7 @@ static Node *relational(Token **rest, Token *tok);
 static Node *add(Token **rest, Token *tok);
 static Node *mul(Token **rest, Token *tok);
 static Type *struct_decl(Token **rest, Token *tok);
+static Type *union_decl(Token **rest, Token *tok);
 static Node *postfix(Token **rest, Token *tok);
 static Node *unary(Token **rest, Token *tok);
 static Node *primary(Token **rest, Token *tok);
@@ -225,6 +226,9 @@ static Type *typespec(Token **rest, Token *tok) {
   if (equal(tok, "struct"))
     return struct_decl(rest, tok->next);
 
+  if (equal(tok, "union"))
+    return union_decl(rest, tok->next);
+
   error_tok(tok, "typename expected");
 }
 
@@ -310,7 +314,8 @@ static Node *declaration(Token **rest, Token *tok) {
 
 // Returns true if a given token represents a type.
 static bool is_typename(Token *tok) {
-  return equal(tok, "char") || equal(tok, "int") || equal(tok, "struct");
+  return equal(tok, "char") || equal(tok, "int") || equal(tok, "struct") ||
+         equal(tok, "union");
 }
 
 // stmt = "return" expr ";"
@@ -626,9 +631,9 @@ static void struct_members(Token **rest, Token *tok, Type *ty) {
   ty->members = head.next;
 }
 
-// struct-decl = ident? "{" struct-members
-static Type *struct_decl(Token **rest, Token *tok) {
-  // Read a struct tag.
+// struct-union-decl = ident? ("{" struct-members)?
+static Type *struct_union_decl(Token **rest, Token *tok) {
+  // Read a tag.
   Token *tag = NULL;
   if (tok->kind == TK_IDENT) {
     tag = tok;
@@ -649,6 +654,17 @@ static Type *struct_decl(Token **rest, Token *tok) {
   struct_members(rest, tok->next, ty);
   ty->align = 1;
 
+  // Register the struct type if a name was given.
+  if (tag)
+    push_tag_scope(tag, ty);
+  return ty;
+}
+
+// struct-decl = struct-union-decl
+static Type *struct_decl(Token **rest, Token *tok) {
+  Type *ty = struct_union_decl(rest, tok);
+  ty->kind = TY_STRUCT;
+
   // Assign offsets within the struct to members.
   int offset = 0;
   for (Member *mem = ty->members; mem; mem = mem->next) {
@@ -660,10 +676,24 @@ static Type *struct_decl(Token **rest, Token *tok) {
       ty->align = mem->ty->align;
   }
   ty->size = align_to(offset, ty->align);
+  return ty;
+}
 
-  // Register the struct type if a name was given.
-  if (tag)
-    push_tag_scope(tag, ty);
+// union-decl = struct-union-decl
+static Type *union_decl(Token **rest, Token *tok) {
+  Type *ty = struct_union_decl(rest, tok);
+  ty->kind = TY_UNION;
+
+  // If union, we don't have to assign offsets because they
+  // are already initialized to zero. We need to compute the
+  // alignment and the size though.
+  for (Member *mem = ty->members; mem; mem = mem->next) {
+    if (ty->align < mem->ty->align)
+      ty->align = mem->ty->align;
+    if (ty->size < mem->ty->size)
+      ty->size = mem->ty->size;
+  }
+  ty->size = align_to(ty->size, ty->align);
   return ty;
 }
 
@@ -677,8 +707,8 @@ static Member *get_struct_member(Type *ty, Token *tok) {
 
 static Node *struct_ref(Node *lhs, Token *tok) {
   add_type(lhs);
-  if (lhs->ty->kind != TY_STRUCT)
-    error_tok(lhs->tok, "not a struct");
+  if (lhs->ty->kind != TY_STRUCT && lhs->ty->kind != TY_UNION)
+    error_tok(lhs->tok, "not a struct nor a union");
 
   Node *node = new_unary(ND_MEMBER, lhs, tok);
   node->member = get_struct_member(lhs->ty, tok);

test/union.c

@@ -0,0 +1,12 @@
+#include "test.h"
+
+int main() {
+  ASSERT(8, ({ union { int a; char b[6]; } x; sizeof(x); }));
+  ASSERT(3, ({ union { int a; char b[4]; } x; x.a = 515; x.b[0]; }));
+  ASSERT(2, ({ union { int a; char b[4]; } x; x.a = 515; x.b[1]; }));
+  ASSERT(0, ({ union { int a; char b[4]; } x; x.a = 515; x.b[2]; }));
+  ASSERT(0, ({ union { int a; char b[4]; } x; x.a = 515; x.b[3]; }));
+
+  printf("OK\n");
+  return 0;
+}

tokenize.c

@@ -117,7 +117,7 @@ static int from_hex(char c) {
 static bool is_keyword(Token *tok) {
   static char *kw[] = {
     "return", "if", "else", "for", "while", "int", "sizeof", "char",
-    "struct",
+    "struct", "union",
   };
 
   for (int i = 0; i < sizeof(kw) / sizeof(*kw); i++)