[cup] Add `>>` and `<<` operators, `fork()` buildin and `SYS_execve`HEADmaster

4 files changed, 50 insertions, 1 deletions

diff --git a/compiler/builtins.cup b/compiler/builtins.cup
index 8aae3d9..bee24e0 100644
--- a/compiler/builtins.cup
+++ b/compiler/builtins.cup
@@ -18,7 +18,16 @@ fn builtin_create_syscall(name: char*, num_args: int) {
 // Defined in `parser.cup`
 fn constant_push(name: char*, val: int);
 
+const MACOS_SYSCALL_OFFSET = 33554432; // 0x2000000;
 fn builtins_push_posix_constants() {
+    
+    // TODO: make this permanent?
+    if (OS_IS_MACOS) {
+        constant_push("SYS_execve", 59 + MACOS_SYSCALL_OFFSET);
+    } else {
+        constant_push("SYS_execve", 59);
+    }
+
     constant_push("SYS_read", SYS_read);
     constant_push("SYS_write", SYS_write);
     constant_push("SYS_exit", SYS_exit);
@@ -72,6 +81,14 @@ fn initialize_builtins() {
     vector_push(node.d.func.args, variable_new("val", type_new(TYPE_ANY), 0));
     vector_push(all_builtins, node);
 
+    // The return value of fork() is weird on macOS, parent/child is returned in rdx
+    // and the child pid is returned in rax, so we'll make our own wrapper.
+    node = node_new(AST_BUILTIN);
+    node.etyp = type_new(TYPE_INT);
+    node.d.func.name = "fork";
+    node.d.func.args = vector_new_sized(0);
+    vector_push(all_builtins, node);
+
     builtin_create_syscall("syscall0", 0);
     builtin_create_syscall("syscall1", 1);
     builtin_create_syscall("syscall2", 2);
@@ -155,4 +172,17 @@ fn generate_builtins() {
         emit_asm("    syscall\n");
         emit_asm("    ret\n");
     }
+
+    emit_asm("func_fork:\n");
+    emit_asm("    mov rdi, [rsp+8]\n");
+    emit_asm("    mov rax, "); emit_num(SYS_fork); emit_asm("\n");
+    emit_asm("    syscall\n");
+    if (OS_IS_MACOS) {
+        // If rdx is 0, we are in the child, so set rax to 0 to match Linux
+        emit_asm("    cmp rdx, 0\n");
+        emit_asm("    je .L1\n");
+        emit_asm("    mov rax, 0\n");
+        emit_asm(".L1:\n");
+    }
+    emit_asm("    ret\n");
 }
\ No newline at end of file
diff --git a/compiler/codegen.cup b/compiler/codegen.cup
index 67a401c..bf20a8b 100644
--- a/compiler/codegen.cup
+++ b/compiler/codegen.cup
@@ -142,6 +142,21 @@ fn generate_expr_into_rax(node: Node*) {
         emit_asm("    pop rbx\n");
         emit_asm("    sub rax, rbx\n");
 
+    } else if (node.typ == AST_LSHIFT) {
+        generate_expr_into_rax(node.d.binary.rhs);
+        emit_asm("    push rax\n");
+        generate_expr_into_rax(node.d.binary.lhs);
+        emit_asm("    pop rcx\n");
+        emit_asm("    shl rax, cl\n");
+        
+
+    } else if (node.typ == AST_RSHIFT) {
+        generate_expr_into_rax(node.d.binary.rhs);
+        emit_asm("    push rax\n");
+        generate_expr_into_rax(node.d.binary.lhs);
+        emit_asm("    pop rcx\n");
+        emit_asm("    shr rax, cl\n");
+
     } else if (node.typ == AST_DIV) {
         generate_expr_into_rax(node.d.binary.rhs);
         emit_asm("    push rax\n");
diff --git a/compiler/lexer.cup b/compiler/lexer.cup
index 9e967f8..e6d0c0b 100644
--- a/compiler/lexer.cup
+++ b/compiler/lexer.cup
@@ -138,12 +138,16 @@ fn lexer_next(lexer: Lexer*, token: Token*) {
         }
 
         else if (c == '<') {
+            if (lexer_peek_char(lexer, 1) == '<')
+                return lexer_make_token(lexer, token, TOKEN_LSHIFT, 2);
             if (lexer_peek_char(lexer, 1) == '=')
                 return lexer_make_token(lexer, token, TOKEN_LEQ, 2);
             return lexer_make_token(lexer, token, TOKEN_LT, 1);
         }
 
         else if (c == '>') {
+            if (lexer_peek_char(lexer, 1) == '>')
+                return lexer_make_token(lexer, token, TOKEN_RSHIFT, 2);
             if (lexer_peek_char(lexer, 1) == '=')
                 return lexer_make_token(lexer, token, TOKEN_GEQ, 2);
             return lexer_make_token(lexer, token, TOKEN_GT, 1);
diff --git a/compiler/parser.cup b/compiler/parser.cup
index d57885e..d3fa64f 100644
--- a/compiler/parser.cup
+++ b/compiler/parser.cup
@@ -482,7 +482,7 @@ fn parse_additive(lexer: Lexer*): Node* {
 
     let lhs = parse_term(lexer);
     lexer_peek(lexer, &token);
-    while (token.typ == TOKEN_PLUS || token.typ == TOKEN_MINUS) {
+    while (token.typ == TOKEN_PLUS || token.typ == TOKEN_MINUS || token.typ == TOKEN_LSHIFT || token.typ == TOKEN_RSHIFT) {
         lexer_next(lexer, &token);
         let op = node_new(binary_token_to_op(token.typ));
         let rhs = parse_term(lexer);