Merge pull request #307 from tobiichi3227/goto-statement-support

Support 'goto' statement

Author: jserv

COMPLIANCE.md

@@ -17,6 +17,7 @@ This document tracks compliance gaps and non-standard behaviors.
 
 ### Control Flow
 - `if`/`else` statements
+- `goto` and label statements
 - `while`, `do-while`, `for` loops
 - `switch`/`case`/`default` statements
 - `break`, `continue`, `return` statements
@@ -96,7 +97,6 @@ This document tracks compliance gaps and non-standard behaviors.
 
 | Feature | Status | Description |
 |---------|--------|-------------|
-| `goto` and labels | Missing | No arbitrary jumps |
 | Designated initializers | Missing | No `.field = value` syntax |
 | Compound literals | Partial | Limited support |
 | Flexible array members | Missing | No `[]` at struct end |
@@ -115,6 +115,7 @@ This document tracks compliance gaps and non-standard behaviors.
 - Escape sequence: `\e` for ESC character
 - `void*` arithmetic (treated as `char*`)
 - `sizeof(void)` returns 0 (should be error)
+- Computed goto
 
 ### Implementation-Specific
 - Array compound literals in scalar context use first element

src/defs.h

@@ -20,6 +20,7 @@
 #define MAX_LOCALS 1600
 #define MAX_FIELDS 64
 #define MAX_TYPES 256
+#define MAX_LABELS 256
 #define MAX_IR_INSTR 80000
 #define MAX_BB_PRED 128
 #define MAX_BB_DOM_SUCC 64
@@ -179,6 +180,7 @@ typedef enum {
     T_break,
     T_default,
     T_continue,
+    T_goto,
     T_const, /* const qualifier */
     /* C pre-processor directives */
     T_cppd_include,
@@ -270,6 +272,7 @@ typedef enum {
     OP_branch,   /* conditional jump */
     OP_jump,     /* unconditional jump */
     OP_func_ret, /* returned value */
+    OP_label,    /* for goto label */
 
     /* function pointer */
     OP_address_of_func, /* resolve function entry */
@@ -567,6 +570,13 @@ struct ref_block {
  * type, parameters) with SSA-related information (e.g., basic blocks, control
  * flow) to support parsing, analysis, optimization, and code generation.
  */
+
+typedef struct {
+    char label_name[MAX_ID_LEN];
+    basic_block_t *bb;
+    bool used;
+} label_t;
+
 struct func {
     /* Syntatic info */
     var_t return_def;

src/globals.c

@@ -1481,6 +1481,14 @@ void dump_bb_insn(func_t *func, basic_block_t *bb, bool *at_func_start)
             printf("br %%%s, %s, %s", rs1->var_name, bb->then_->bb_label_name,
                    bb->else_->bb_label_name);
             break;
+        case OP_jump:
+            print_indent(1);
+            printf("jmp %s", bb->next->bb_label_name);
+            break;
+        case OP_label:
+            print_indent(0);
+            printf("%s:", insn->str);
+            break;
         case OP_push:
             print_indent(1);
             printf("push %%%s", rs1->var_name);

src/lexer.c

@@ -12,7 +12,7 @@
 
 /* Hash table constants */
 #define NUM_DIRECTIVES 11
-#define NUM_KEYWORDS 17
+#define NUM_KEYWORDS 18
 
 /* Token mapping structure for elegant initialization */
 typedef struct {
@@ -85,6 +85,7 @@ void lex_init_keywords()
         {"break", T_break},
         {"default", T_default},
         {"continue", T_continue},
+        {"goto", T_goto},
         {"union", T_union},
         {"const", T_const},
     };
@@ -786,6 +787,8 @@ token_t lex_token_impl(bool aliasing)
                     keyword = T_enum;
             } else if (!memcmp(token_str, "case", 4))
                 keyword = T_case;
+            else if (!memcmp(token_str, "goto", 4))
+                keyword = T_goto;
             break;
 
         case 5: /* 5-letter keywords: while, break, union, const */

src/parser.c

@@ -26,6 +26,12 @@ int break_exit_idx = 0;
 basic_block_t *continue_bb[MAX_NESTING];
 int continue_pos_idx = 0;
 
+/* Label utilities */
+label_t labels[MAX_LABELS];
+int label_idx = 0;
+basic_block_t *backpatch_bb[MAX_LABELS];
+int backpatch_bb_idx = 0;
+
 /* stack of the operands of 3AC */
 var_t *operand_stack[MAX_OPERAND_STACK_SIZE];
 int operand_stack_idx = 0;
@@ -40,6 +46,26 @@ void parse_array_init(var_t *var,
                       basic_block_t **bb,
                       bool emit_code);
 
+
+label_t *find_label(char *name)
+{
+    for (int i = 0; i < label_idx; i++) {
+        if (!strcmp(name, labels[i].label_name))
+            return &labels[i];
+    }
+    return NULL;
+}
+
+void add_label(char *name, basic_block_t *bb)
+{
+    if (label_idx > MAX_LABELS - 1)
+        error("Too many labels in function");
+
+    label_t *l = &labels[label_idx++];
+    strncpy(l->label_name, name, MAX_ID_LEN);
+    l->bb = bb;
+}
+
 char *gen_name_to(char *buf)
 {
     sprintf(buf, ".t%d", global_var_idx++);
@@ -997,6 +1023,61 @@ basic_block_t *handle_while_statement(block_t *parent, basic_block_t *bb)
     return else_;
 }
 
+basic_block_t *handle_goto_statement(block_t *parent, basic_block_t *bb)
+{
+    /* Since a goto splits the current program into two basic blocks and makes
+     * the subsequent basic block unreachable, this causes problems for later
+     * CFG operations. Therefore, we create a fake if that always executes to
+     * wrap the goto, and connect the unreachable basic block to the else
+     * branch. Finally, return this else block.
+     *
+     * after:
+     * a = b + c;
+     * goto label;
+     * c *= d;
+     *
+     * before:
+     * a = b + c;
+     * if (1)
+     *     goto label;
+     * c *= d;
+     */
+
+    char token[MAX_ID_LEN];
+    if (!lex_peek(T_identifier, token))
+        error("Expected identifier after 'goto'");
+
+    lex_expect(T_identifier);
+    lex_expect(T_semicolon);
+
+    basic_block_t *fake_if = bb_create(parent);
+    bb_connect(bb, fake_if, NEXT);
+    var_t *val = require_var(parent);
+    gen_name_to(val->var_name);
+    val->init_val = 1;
+    add_insn(parent, fake_if, OP_load_constant, val, NULL, NULL, 0, NULL);
+    add_insn(parent, fake_if, OP_branch, NULL, val, NULL, 0, NULL);
+
+    basic_block_t *then_ = bb_create(parent);
+    basic_block_t *else_ = bb_create(parent);
+    bb_connect(fake_if, then_, THEN);
+    bb_connect(fake_if, else_, ELSE);
+
+    add_insn(parent, then_, OP_jump, NULL, NULL, NULL, 0, token);
+    label_t *label = find_label(token);
+    if (label) {
+        label->used = true;
+        bb_connect(then_, label->bb, NEXT);
+        return else_;
+    }
+
+    if (backpatch_bb_idx > MAX_LABELS - 1)
+        error("Too many forward-referenced labels");
+
+    backpatch_bb[backpatch_bb_idx++] = then_;
+    return else_;
+}
+
 basic_block_t *handle_struct_variable_decl(block_t *parent,
                                            basic_block_t *bb,
                                            char *token)
@@ -4169,6 +4250,9 @@ basic_block_t *read_body_statement(block_t *parent, basic_block_t *bb)
         return do_while_end;
     }
 
+    if (lex_accept(T_goto))
+        return handle_goto_statement(parent, bb);
+
     /* empty statement */
     if (lex_accept(T_semicolon))
         return bb;
@@ -4753,6 +4837,21 @@ basic_block_t *read_body_statement(block_t *parent, basic_block_t *bb)
         return bb;
     }
 
+    if (lex_peek(T_identifier, token)) {
+        lex_accept(T_identifier);
+        if (lex_accept(T_colon)) {
+            label_t *l = find_label(token);
+            if (l)
+                error("label redefinition");
+
+            basic_block_t *n = bb_create(parent);
+            bb_connect(bb, n, NEXT);
+            add_label(token, n);
+            add_insn(parent, n, OP_label, NULL, NULL, NULL, 0, token);
+            return n;
+        }
+    }
+
     error("Unrecognized statement token");
     return NULL;
 }
@@ -4794,6 +4893,28 @@ void read_func_body(func_t *func)
     basic_block_t *body = read_code_block(func, NULL, NULL, func->bbs);
     if (body)
         bb_connect(body, func->exit, NEXT);
+
+    for (int i = 0; i < backpatch_bb_idx; i++) {
+        basic_block_t *bb = backpatch_bb[i];
+        insn_t *g = bb->insn_list.tail;
+        label_t *label = find_label(g->str);
+        if (!label)
+            error("goto label undefined");
+
+        label->used = true;
+        bb_connect(bb, label->bb, NEXT);
+    }
+
+    for (int i = 0; i < label_idx; i++) {
+        label_t *label = &labels[i];
+        if (label->used)
+            continue;
+
+        printf("Warning: unused label %s\n", label->label_name);
+    }
+
+    backpatch_bb_idx = 0;
+    label_idx = 0;
 }
 
 /* if first token is type */

src/ssa.c

@@ -929,6 +929,82 @@ void unwind_phi(void)
     }
 }
 
+bool is_dominate(basic_block_t *pred, basic_block_t *succ)
+{
+    int i;
+    bool found = false;
+    for (i = 0; i < MAX_BB_DOM_SUCC; i++) {
+        if (!pred->dom_next[i])
+            break;
+        if (pred->dom_next[i] == succ) {
+            found = true;
+            break;
+        }
+        found |= is_dominate(pred->dom_next[i], succ);
+    }
+
+    return found;
+}
+
+/*
+ * For any variable, the basic block that defines it must dominate all the
+ * basic blocks where it is used; otherwise, it is an invalid cross-block
+ * initialization.
+ */
+void bb_check_var_cross_init(func_t *func, basic_block_t *bb)
+{
+    UNUSED(func);
+
+    for (insn_t *insn = bb->insn_list.head; insn; insn = insn->next) {
+        if (insn->opcode != OP_allocat)
+            continue;
+
+        var_t *var = insn->rd;
+        ref_block_t *ref;
+        for (ref = var->ref_block_list.head; ref; ref = ref->next) {
+            if (ref->bb == bb)
+                continue;
+
+            if (!is_dominate(bb, ref->bb))
+                printf("Warning: Variable '%s' cross-initialized\n",
+                       var->var_name);
+        }
+    }
+}
+
+/**
+ * A variable's initialization lives in a basic block that does not dominate
+ * all of its uses, so control flow can reach a use without first passing
+ * through its initialization (i.e., a possibly-uninitialized use).
+ *
+ * For Example:
+ * // Jumps directly to 'label', skipping the declaration below
+ * goto label;
+ * if (1) {
+ *     // This line is never executed when 'goto' is taken
+ *     int x;
+ * label:
+ *     // Uses 'x' after its declaration was bypassed
+ *     x = 5;
+ * }
+ */
+void check_var_cross_init()
+{
+    bb_traversal_args_t *args = arena_alloc_traversal_args();
+    for (func_t *func = FUNC_LIST.head; func; func = func->next) {
+        /* Skip function declarations without bodies */
+        if (!func->bbs)
+            continue;
+
+        args->func = func;
+        args->bb = func->bbs;
+
+        func->visited++;
+        args->postorder_cb = bb_check_var_cross_init;
+        bb_forward_traversal(args);
+    }
+}
+
 #ifdef __SHECC__
 #else
 void bb_dump_connection(FILE *fd,
@@ -1112,6 +1188,12 @@ void bb_dump(FILE *fd, func_t *func, basic_block_t *bb)
                 sprintf(str, "<BRANCH %s<SUB>%d</SUB>>", insn->rs1->var_name,
                         insn->rs1->subscript);
                 break;
+            case OP_jump:
+                sprintf(str, "<JUMP>");
+                break;
+            case OP_label:
+                sprintf(str, "<LABEL>");
+                break;
             case OP_push:
                 sprintf(str, "<PUSH %s<SUB>%d</SUB>>", insn->rs1->var_name,
                         insn->rs1->subscript);
@@ -1281,6 +1363,9 @@ void ssa_build(void)
     build_df();
 
     solve_globals();
+
+    check_var_cross_init();
+
     solve_phi_insertion();
     solve_phi_params();