stage1 实现并修复ast2ir的核心功能的正确性

stage1/spl_ast.c

@@ -91,7 +91,7 @@ static spl_ast_node_ref_t new_node(parser_t *p, spl_ast_node_kind_t kind, const
     spl_ast_node_t n = {0};
     n.kind = kind;
     n.resolved_def_id = 0;
-    n.resolved_type_id = 0;
+    n.resolved_lang_tid = 0;
     if (tok) {
         n.dbg.dbg_name = spl_ast_kind_name(kind);
         n.dbg.fname = tok->fname;

stage1/spl_ast.h

@@ -150,8 +150,10 @@ typedef VEC(spl_ast_node_ref_t) spl_ast_node_ref_vec_t;
 struct spl_ast_node {
     spl_ast_node_kind_t kind;
     spl_dbg_node_t dbg;
+
     usize resolved_def_id;
-    usize resolved_type_id;
+    usize resolved_lang_tid;
+    usize expected_lang_tid;
 
     union {
         /* SPL_AST_ROOT */

stage1/spl_ast2ir.c

@@ -8,11 +8,8 @@ static spl_type_id_t to_ir_type(spl_ast2ir_t *a, spl_lang_type_id_t lang) {
 }
 
 static spl_lang_type_id_t node_lang(spl_ast2ir_t *a, spl_ast_node_t *n) {
-    if (!n)
+    (void)a;
-        return 0;
+    return n ? n->resolved_lang_tid : 0;
-    if (n->kind == SPL_AST_EXPR_IDENT)
-        return spl_def_get_lang_id(a->sema->type, n->resolved_def_id);
-    return n->resolved_type_id;
 }
 
 static spl_type_id_t node_ir(spl_ast2ir_t *a, spl_ast_node_t *n) {
@@ -30,27 +27,48 @@ static void emit_stmt(spl_ast2ir_t *a, spl_ast_node_ref_t ref);
 static void emit_decl(spl_ast2ir_t *a, spl_ast_node_ref_t ref);
 static void emit_fn(spl_ast2ir_t *a, spl_ast_node_t *n);
 static void emit_container(spl_ast2ir_t *a, spl_ast_node_ref_t ref);
-static spl_ir_node_ref_t emit_cast(spl_ast2ir_t *a, spl_ast_node_t *n);
 
-static spl_ir_node_ref_t ast2ir_intrin_call(spl_ast2ir_t *a, spl_ast_node_t *n, spl_intrin_t in);
+static spl_ir_node_ref_t emit_cast(spl_ast2ir_t *a, spl_ast_node_t *n);
+static spl_ir_node_ref_t emit_expr_raw(spl_ast2ir_t *a, spl_ast_node_ref_t ref);
+static spl_ir_node_ref_t emit_implicit_cast(spl_ast2ir_t *a, spl_type_id_t ft, spl_type_id_t tt,
+                                            spl_ir_node_ref_t v);
+
+static spl_ir_node_ref_t ast2ir_intrin_call(spl_ast2ir_t *a, spl_ast_node_t *n, spl_intrin_t in,
+                                            spl_ast_node_ref_vec_t *args);
 static spl_ir_node_ref_t ast2ir_intrin_field(spl_ast2ir_t *a, spl_ast_node_t *n, spl_intrin_t in);
 static spl_ir_node_ref_t ast2ir_intrin_len(spl_ast2ir_t *a, spl_ast_node_t *n);
 
-static spl_ir_node_ref_t ast2ir_intrin_call(spl_ast2ir_t *a, spl_ast_node_t *n, spl_intrin_t in) {
+static spl_ir_node_ref_t ast2ir_intrin_call(spl_ast2ir_t *a, spl_ast_node_t *n, spl_intrin_t in,
+                                            spl_ast_node_ref_vec_t *args) {
     switch (in) {
     case SPL_INTRIN_SIZEOF: {
-        spl_ast_node_t *arg = spl_ast_node(a->sema->ast, vec_at(n->call.args, 0));
+        spl_ast_node_t *arg = spl_ast_node(a->sema->ast, vec_at(*args, 0));
         return spl_ir_builder_type_sizeof(a->ir, node_ir(a, arg));
     }
     case SPL_INTRIN_ALIGNOF: {
-        spl_ast_node_t *arg = spl_ast_node(a->sema->ast, vec_at(n->call.args, 0));
+        spl_ast_node_t *arg = spl_ast_node(a->sema->ast, vec_at(*args, 0));
         return spl_ir_builder_type_alignof(a->ir, node_ir(a, arg));
     }
     case SPL_INTRIN_ASSERT: {
-        /* 目前 SEMA 只做类型检查，emit 阶段留空，
+        spl_ir_node_ref_t cond = emit_expr(a, vec_at(*args, 0));
-           @assert 展开需待 control.trap 语义明确后再实现 */
+        if (!cond)
+            return 0;
+
+        spl_lang_type_id_t b = spl_lang_make_bool(a->sema->type);
+        spl_ir_node_ref_t notc = spl_ir_builder_arith_not(a->ir, to_ir_type(a, b), cond);
+        spl_ir_node_ref_t br = spl_ir_builder_control_br(a->ir, notc);
+        spl_ir_builder_block_end(a->ir);
+
+        usize trap_blk = spl_ir_builder_block_begin(a->ir);
+        spl_ir_builder_control_trap(a->ir);
+        spl_ir_builder_block_end(a->ir);
+
+        usize cont_blk = spl_ir_builder_block_begin(a->ir);
+        spl_ir_builder_control_br_fixup(a->ir, br, trap_blk, cont_blk);
         return 0;
     }
+    case SPL_INTRIN_BREAKPOINT:
+        return spl_ir_builder_dbg_breakpoint(a->ir);
     default:
         UNREACHABLE();
     }
@@ -198,7 +216,7 @@ static spl_ir_node_ref_t emit_lvalue(spl_ast2ir_t *a, spl_ast_node_ref_t ref) {
 }
 
 static spl_ir_node_ref_t emit_cast(spl_ast2ir_t *a, spl_ast_node_t *n) {
-    spl_ir_node_ref_t v = emit_expr(a, n->cast.expr);
+    spl_ir_node_ref_t v = emit_expr_raw(a, n->cast.expr);
     if (!v)
         return 0;
     spl_ast_node_t *src = spl_ast_node(a->sema->ast, n->cast.expr);
@@ -238,6 +256,28 @@ static spl_ir_node_ref_t emit_cast(spl_ast2ir_t *a, spl_ast_node_t *n) {
 }
 
 static spl_ir_node_ref_t emit_expr(spl_ast2ir_t *a, spl_ast_node_ref_t ref) {
+    spl_ir_node_ref_t val = emit_expr_raw(a, ref);
+    if (!val)
+        return 0;
+
+    spl_ast_node_t *n = spl_ast_node(a->sema->ast, ref);
+    if (!n || n->expected_lang_tid == 0)
+        return val;
+
+    spl_lang_type_id_t from = n->resolved_lang_tid;
+    spl_lang_type_id_t to = n->expected_lang_tid;
+    if (!from || from == to)
+        return val;
+
+    spl_type_id_t from_ir = to_ir_type(a, from);
+    spl_type_id_t to_ir = to_ir_type(a, to);
+    if (!from_ir || !to_ir || from_ir == to_ir)
+        return val;
+
+    return emit_implicit_cast(a, from_ir, to_ir, val);
+}
+
+static spl_ir_node_ref_t emit_expr_raw(spl_ast2ir_t *a, spl_ast_node_ref_t ref) {
     spl_ast_node_t *n = spl_ast_node(a->sema->ast, ref);
     if (!n)
         return 0;
@@ -396,7 +436,7 @@ static spl_ir_node_ref_t emit_expr(spl_ast2ir_t *a, spl_ast_node_ref_t ref) {
         if (callee && callee->kind == SPL_AST_EXPR_IDENT) {
             spl_intrin_t in = spl_intrin_lookup_call(callee->ident.name);
             if (in != SPL_INTRIN_NONE)
-                return ast2ir_intrin_call(a, n, in);
+                return ast2ir_intrin_call(a, n, in, &n->call.args);
         }
 
         spl_ir_node_ref_t fn = emit_expr(a, n->call.callee);
@@ -573,9 +613,12 @@ static spl_ir_node_ref_t emit_expr(spl_ast2ir_t *a, spl_ast_node_ref_t ref) {
         spl_ir_builder_agg_construct_field(a->ir, node, len);
         return node;
     }
-    case SPL_AST_EXPR_COMPTIME_CALL:
+    case SPL_AST_EXPR_COMPTIME_CALL: {
-        // TODO
+        spl_intrin_t in = spl_intrin_lookup_call(n->comptime_call.name);
+        if (in == SPL_INTRIN_NONE)
             return 0;
+        return ast2ir_intrin_call(a, n, in, &n->comptime_call.args);
+    }
     default:
         TODO();
         return 0;
@@ -959,6 +1002,42 @@ static void emit_container(spl_ast2ir_t *a, spl_ast_node_ref_t ref) {
     }
 }
 
+static spl_ir_node_ref_t emit_implicit_cast(spl_ast2ir_t *a, spl_type_id_t ft, spl_type_id_t tt,
+                                            spl_ir_node_ref_t v) {
+    if (!ft || !tt || ft == tt)
+        return v;
+    spl_type_node_t *fs = spl_type_node(a->sema->type, ft);
+    spl_type_node_t *ts = spl_type_node(a->sema->type, tt);
+    if (!fs || !ts)
+        return v;
+
+    if (fs->kind == SPL_TYPE_INT && ts->kind == SPL_TYPE_INT) {
+        if (fs->int_type.bits == ts->int_type.bits)
+            return spl_ir_builder_cast_bitcast(a->ir, ft, tt, v);
+        if (fs->int_type.bits > ts->int_type.bits)
+            return spl_ir_builder_cast_trunc(a->ir, ft, tt, v);
+        if (fs->int_type.is_signed)
+            return spl_ir_builder_cast_sext(a->ir, ft, tt, v);
+        return spl_ir_builder_cast_zext(a->ir, ft, tt, v);
+    }
+    if (fs->kind == SPL_TYPE_INT && ts->kind == SPL_TYPE_FLOAT)
+        return spl_ir_builder_case_int2float(a->ir, ft, tt, v);
+    if (fs->kind == SPL_TYPE_FLOAT && ts->kind == SPL_TYPE_INT)
+        return spl_ir_builder_case_float2int(a->ir, ft, tt, v);
+    if (fs->kind == SPL_TYPE_FLOAT && ts->kind == SPL_TYPE_FLOAT) {
+        if (fs->float_type.bits > ts->float_type.bits)
+            return spl_ir_builder_cast_ftrunc(a->ir, ft, tt, v);
+        return spl_ir_builder_cast_fext(a->ir, ft, tt, v);
+    }
+    if (fs->kind == SPL_TYPE_BOOL && ts->kind == SPL_TYPE_INT)
+        return spl_ir_builder_cast_bool2int(a->ir, ft, tt, v);
+    if (fs->kind == SPL_TYPE_PTR && ts->kind == SPL_TYPE_INT)
+        return spl_ir_builder_cast_ptr2int(a->ir, ft, tt, v);
+    if (fs->kind == SPL_TYPE_INT && ts->kind == SPL_TYPE_PTR)
+        return spl_ir_builder_cast_int2ptr(a->ir, ft, tt, v);
+    return spl_ir_builder_cast_bitcast(a->ir, ft, tt, v);
+}
+
 static usize spl_int_map_hash(usize k) { return MAP_HASH_INT(k); }
 static int spl_int_map_cmp(usize a, usize b) { return MAP_CMP_INT(a, b); }
 

stage1/spl_intrin.c

@@ -8,6 +8,8 @@ static const spl_intrin_info_t g_intrin_table[] = {
                             SPL_INTRIN_RESULT_USIZE},
     [SPL_INTRIN_ASSERT] = {"@assert", SPL_INTRIN_ASSERT, SPL_INTRIN_FORM_CALL, 1, 2,
                            SPL_INTRIN_RESULT_UNIT},
+    [SPL_INTRIN_BREAKPOINT] = {"@breakpoint", SPL_INTRIN_BREAKPOINT, SPL_INTRIN_FORM_CALL, 0, 0,
+                               SPL_INTRIN_RESULT_UNIT},
     [SPL_INTRIN_LEN] = {"len", SPL_INTRIN_LEN, SPL_INTRIN_FORM_FIELD, 0, 0,
                         SPL_INTRIN_RESULT_USIZE},
 };

stage1/spl_intrin.h

@@ -9,6 +9,7 @@ typedef enum {
     SPL_INTRIN_SIZEOF,
     SPL_INTRIN_ALIGNOF,
     SPL_INTRIN_ASSERT,
+    SPL_INTRIN_BREAKPOINT,
     SPL_INTRIN_LEN,
 } spl_intrin_t;
 

stage1/spl_ir.c

@@ -315,12 +315,16 @@ static spl_ir_node_ref_t builder_alloc_node(spl_ir_builder_t *b, spl_ir_kind_t k
     spl_ir_node_t *n = spl_ir_node(&b->ir, b->current_fn, ref);
     n->kind = kind;
     n->dbg = b->dbg;
-    if (b->current_block < vec_size(b->blocks)) {
+
-        vec_at(b->blocks, b->current_block) = ref;
+    if (b->current_block >= vec_size(b->blocks)) {
-    } else {
         Panic("unknown bblock");
         return 0;
     }
+
+    spl_ir_node_ref_t *slot = &vec_at(b->blocks, b->current_block);
+    if (*slot == 0)
+        *slot = ref;
+
     return ref;
 }
 
@@ -378,7 +382,7 @@ void spl_ir_builder_fn_end(spl_ir_builder_t *b) {
             target = vec_at(b->blocks, fx->block);
         }
         if (target == 0) {
-            Panic("unknown reloc target bblock");
+            Panic("reloc to empty/unknown bblock #%zu", fx->block);
         }
         switch (fx->slot) {
         case SPL_IR_FIX_BR_TRUE:
@@ -411,6 +415,10 @@ void spl_ir_builder_block_end(spl_ir_builder_t *b) {
     if (!b->terminated) {
         Panic("bblock must be end of terminated");
     }
+    if (b->current_block != 0 && b->current_block < vec_size(b->blocks) &&
+        vec_at(b->blocks, b->current_block) == 0) {
+        Panic("bblock #%zu is empty (no entry node emitted)", b->current_block);
+    }
 }
 
 #define X(name, kind)                                                                              \
@@ -769,7 +777,9 @@ spl_ir_node_ref_t spl_ir_builder_control_ret(spl_ir_builder_t *b, spl_type_id_t
 }
 
 spl_ir_node_ref_t spl_ir_builder_control_unreachable(spl_ir_builder_t *b) {
-    return builder_alloc_node(b, SPL_IR_CONTROL_UNREACHABLE);
+    spl_ir_node_ref_t ref = builder_alloc_node(b, SPL_IR_CONTROL_UNREACHABLE);
+    b->terminated = true;
+    return ref;
 }
 
 spl_ir_node_ref_t spl_ir_builder_control_trap(spl_ir_builder_t *b) {