[cc] support C99 compound literals

jgarzik · jgarzik · commit 64f8faa2a15b · 2025-12-08T21:08:33.000-05:00
diff --git a/cc/arch/aarch64/codegen.rs b/cc/arch/aarch64/codegen.rs
@@ -168,8 +168,11 @@ impl Aarch64CodeGen {
         // Data section
         self.push_lir(Aarch64Inst::Directive(Directive::Data));
 
-        // Global visibility (if not static)
-        if !is_static {
+        // Check if this is a local symbol (starts with '.')
+        let is_local = name.starts_with('.');
+
+        // Global visibility (if not static and not local)
+        if !is_static && !is_local {
             self.push_lir(Aarch64Inst::Directive(Directive::global(name)));
         }
 
@@ -184,8 +187,12 @@ impl Aarch64CodeGen {
             )));
         }
 
-        // Label
-        self.push_lir(Aarch64Inst::Directive(Directive::global_label(name)));
+        // Label - use local_label for names starting with '.'
+        if is_local {
+            self.push_lir(Aarch64Inst::Directive(Directive::local_label(name)));
+        } else {
+            self.push_lir(Aarch64Inst::Directive(Directive::global_label(name)));
+        }
 
         // Emit initializer
         self.emit_initializer_data(init, size as usize);
diff --git a/cc/arch/x86_64/codegen.rs b/cc/arch/x86_64/codegen.rs
@@ -168,8 +168,11 @@ impl X86_64CodeGen {
         // Data section
         self.push_lir(X86Inst::Directive(Directive::Data));
 
-        // Global visibility (if not static)
-        if !is_static {
+        // Check if this is a local symbol (starts with '.')
+        let is_local = name.starts_with('.');
+
+        // Global visibility (if not static and not local)
+        if !is_static && !is_local {
             self.push_lir(X86Inst::Directive(Directive::global(name)));
         }
 
@@ -182,8 +185,12 @@ impl X86_64CodeGen {
             self.push_lir(X86Inst::Directive(Directive::Align(align.trailing_zeros())));
         }
 
-        // Label
-        self.push_lir(X86Inst::Directive(Directive::global_label(name)));
+        // Label - use local_label for names starting with '.'
+        if is_local {
+            self.push_lir(X86Inst::Directive(Directive::local_label(name)));
+        } else {
+            self.push_lir(X86Inst::Directive(Directive::global_label(name)));
+        }
 
         // Emit initializer
         self.emit_initializer_data(init, size as usize);
diff --git a/cc/doc/BUGS.md b/cc/doc/BUGS.md
@@ -4,25 +4,6 @@ This file documents known bugs where valid C99 code fails to compile.
 
 ## Parse Errors on Valid C99 Code
 
-### Compound Literals
-
-**Status**: Parse error
-
-**Valid C99 Code**:
-```c
-struct Point { int x, y; };
-struct Point p = (struct Point){1, 2};  // Compound literal
-```
-
-**Error**:
-```
-parse error: unexpected token in expression
-```
-
-**C99 Reference**: 6.5.2.5
-
----
-
 ### Array Parameter with `static`
 
 **Status**: Parse error
diff --git a/cc/ir/linearize.rs b/cc/ir/linearize.rs
@@ -94,6 +94,8 @@ pub struct Linearizer<'a> {
     current_func_name: String,
     /// Counter for generating unique static local names
     static_local_counter: u32,
+    /// Counter for generating unique compound literal names (for file-scope compound literals)
+    compound_literal_counter: u32,
     /// Static local variables (local name -> static local info)
     /// This is persistent across function calls (not cleared per function)
     static_locals: HashMap<String, StaticLocalInfo>,
@@ -135,6 +137,7 @@ impl<'a> Linearizer<'a> {
             struct_return_size: 0,
             current_func_name: String::new(),
             static_local_counter: 0,
+            compound_literal_counter: 0,
             static_locals: HashMap::new(),
             current_pos: None,
             target,
@@ -518,7 +521,8 @@ impl<'a> Linearizer<'a> {
     /// - Struct initializers with designated and positional fields
     /// - Address-of expressions (&symbol)
     /// - Nested initializers
-    fn ast_init_to_ir(&self, expr: &Expr, typ: TypeId) -> Initializer {
+    /// - Compound literals (C99 6.5.2.5)
+    fn ast_init_to_ir(&mut self, expr: &Expr, typ: TypeId) -> Initializer {
         match &expr.kind {
             ExprKind::IntLit(v) => Initializer::Int(*v),
             ExprKind::FloatLit(v) => Initializer::Float(*v),
@@ -573,6 +577,33 @@ impl<'a> Linearizer<'a> {
             // Initializer list for arrays/structs
             ExprKind::InitList { elements } => self.ast_init_list_to_ir(elements, typ),
 
+            // Compound literal in initializer context (C99 6.5.2.5)
+            ExprKind::CompoundLiteral {
+                typ: cl_type,
+                elements,
+            } => {
+                // Check if compound literal type matches target type
+                if *cl_type == typ {
+                    // Direct value - treat like InitList
+                    self.ast_init_list_to_ir(elements, typ)
+                } else if self.types.kind(typ) == TypeKind::Pointer {
+                    // Pointer initialization - create anonymous static global
+                    // and return its address
+                    let anon_name = format!(".CL{}", self.compound_literal_counter);
+                    self.compound_literal_counter += 1;
+
+                    // Create the anonymous global
+                    let init = self.ast_init_list_to_ir(elements, *cl_type);
+                    self.module.add_global(&anon_name, *cl_type, init);
+
+                    // Return address of the anonymous global
+                    Initializer::SymAddr(anon_name)
+                } else {
+                    // Type mismatch - try to use the initializer list directly
+                    self.ast_init_list_to_ir(elements, typ)
+                }
+            }
+
             // Identifier - for constant addresses (function pointers, array decay, etc.)
             // or enum constants
             ExprKind::Ident { name } => {
@@ -603,7 +634,7 @@ impl<'a> Linearizer<'a> {
     }
 
     /// Convert an AST initializer list to an IR Initializer
-    fn ast_init_list_to_ir(&self, elements: &[InitElement], typ: TypeId) -> Initializer {
+    fn ast_init_list_to_ir(&mut self, elements: &[InitElement], typ: TypeId) -> Initializer {
         let type_kind = self.types.kind(typ);
         let total_size = (self.types.size_bits(typ) / 8) as usize;
 
@@ -2147,6 +2178,28 @@ impl<'a> Linearizer<'a> {
                 ));
                 addr
             }
+            ExprKind::CompoundLiteral { typ, elements } => {
+                // Compound literal as lvalue: create it and return its address
+                // This is used for &(struct S){...}
+                let sym_id = self.alloc_pseudo();
+                let unique_name = format!(".compound_literal#{}", sym_id.0);
+                let sym = Pseudo::sym(sym_id, unique_name.clone());
+                if let Some(func) = &mut self.current_func {
+                    func.add_pseudo(sym);
+                    func.add_local(&unique_name, sym_id, *typ, false, self.current_bb);
+                }
+                self.linearize_init_list(sym_id, *typ, elements);
+
+                // Return address of the compound literal
+                let result = self.alloc_pseudo();
+                let pseudo = Pseudo::reg(result, result.0);
+                if let Some(func) = &mut self.current_func {
+                    func.add_pseudo(pseudo);
+                }
+                let ptr_type = self.types.pointer_to(*typ);
+                self.emit(Instruction::sym_addr(result, sym_id, ptr_type));
+                result
+            }
             _ => {
                 // Fallback: just evaluate the expression (shouldn't happen for valid lvalues)
                 self.linearize_expr(expr)
@@ -3249,6 +3302,44 @@ impl<'a> Linearizer<'a> {
                 panic!("InitList should be handled in declaration context, not as standalone expression")
             }
 
+            ExprKind::CompoundLiteral { typ, elements } => {
+                // Compound literals have automatic storage at block scope
+                // Create an anonymous local variable, similar to how local variables work
+
+                // Create a symbol pseudo for the compound literal (its address)
+                let sym_id = self.alloc_pseudo();
+                let unique_name = format!(".compound_literal#{}", sym_id.0);
+                let sym = Pseudo::sym(sym_id, unique_name.clone());
+                if let Some(func) = &mut self.current_func {
+                    func.add_pseudo(sym);
+                    // Register as local for proper stack allocation
+                    func.add_local(&unique_name, sym_id, *typ, false, self.current_bb);
+                }
+
+                // Initialize using existing init list machinery
+                self.linearize_init_list(sym_id, *typ, elements);
+
+                // For arrays: return pointer (array-to-pointer decay)
+                // For structs/scalars: load and return the value
+                let result = self.alloc_pseudo();
+                let pseudo = Pseudo::reg(result, result.0);
+                if let Some(func) = &mut self.current_func {
+                    func.add_pseudo(pseudo);
+                }
+
+                if self.types.kind(*typ) == TypeKind::Array {
+                    // Array compound literal - decay to pointer to first element
+                    let elem_type = self.types.base_type(*typ).unwrap_or(self.types.int_id);
+                    let ptr_type = self.types.pointer_to(elem_type);
+                    self.emit(Instruction::sym_addr(result, sym_id, ptr_type));
+                } else {
+                    // Struct/scalar compound literal - load the value
+                    let size = self.types.size_bits(*typ);
+                    self.emit(Instruction::load(result, sym_id, 0, *typ, size));
+                }
+                result
+            }
+
             // ================================================================
             // Variadic function support (va_* builtins)
             // ================================================================
diff --git a/cc/parse/ast.rs b/cc/parse/ast.rs
@@ -244,6 +244,13 @@ pub enum ExprKind {
         expr: Box<Expr>,
     },
 
+    /// Compound literal: (type){ init-list }
+    /// C99 6.5.2.5: Creates unnamed object with automatic/static storage
+    CompoundLiteral {
+        typ: TypeId,
+        elements: Vec<InitElement>,
+    },
+
     /// sizeof type: sizeof(int)
     SizeofType(TypeId),
 
diff --git a/cc/parse/parser.rs b/cc/parse/parser.rs
@@ -1511,12 +1511,19 @@ impl<'a> Parser<'a> {
             result_id = self.types.intern(Type::pointer(result_id));
         }
 
-        // Handle array declarators: int[10], char[20], etc.
+        // Handle array declarators: int[10], char[20], int[], etc.
         while self.is_special(b'[') {
             self.advance();
-            if let Ok(size_expr) = self.parse_conditional_expr() {
+            // Check for empty brackets [] (incomplete array type for compound literals)
+            if self.is_special(b']') {
+                // Empty brackets - create array with size 0 (size to be determined from initializer)
+                result_id = self.types.intern(Type::array(result_id, 0));
+            } else if let Ok(size_expr) = self.parse_conditional_expr() {
                 if let Some(size) = self.eval_const_expr(&size_expr) {
                     result_id = self.types.intern(Type::array(result_id, size as usize));
+                } else {
+                    // Non-constant size (VLA in type name) - create array with size 0
+                    result_id = self.types.intern(Type::array(result_id, 0));
                 }
             }
             if !self.is_special(b']') {
@@ -2214,11 +2221,42 @@ impl<'a> Parser<'a> {
                     let paren_pos = self.current_pos();
                     self.advance();
 
-                    // Try to detect cast (type) - simplified check
+                    // Try to detect cast (type) or compound literal (type){...}
                     if let Some(typ) = self.try_parse_type_name() {
                         self.expect_special(b')')?;
+
+                        // Check for compound literal: (type){ ... }
+                        if self.is_special(b'{') {
+                            let init_list = self.parse_initializer_list()?;
+                            let elements = match init_list.kind {
+                                ExprKind::InitList { elements } => elements,
+                                _ => unreachable!(),
+                            };
+
+                            // For incomplete array types (size 0), determine size from initializer
+                            let final_typ = if self.types.kind(typ) == TypeKind::Array
+                                && self.types.get(typ).array_size == Some(0)
+                            {
+                                // Array size should be determined from initializer element count
+                                let elem_type =
+                                    self.types.base_type(typ).unwrap_or(self.types.int_id);
+                                self.types.intern(Type::array(elem_type, elements.len()))
+                            } else {
+                                typ
+                            };
+
+                            return Ok(Self::typed_expr(
+                                ExprKind::CompoundLiteral {
+                                    typ: final_typ,
+                                    elements,
+                                },
+                                final_typ,
+                                paren_pos,
+                            ));
+                        }
+
+                        // Regular cast expression
                         let expr = self.parse_unary_expr()?;
-                        // Cast expression has the cast type
                         return Ok(Self::typed_expr(
                             ExprKind::Cast {
                                 cast_type: typ,
diff --git a/cc/tests/datatypes/struct_type.rs b/cc/tests/datatypes/struct_type.rs
