refactor(ast2ir): 移除废弃的ABI依赖并优化类型转换处理

移除了对scc_abi包的依赖,将相关头文件从libs/abi移动到libs/ast2ir目录下。
重构了基本类型解析功能,将parse_base_type函数提取为独立的
scc_ast2ir_parse_base_type实现,并支持有符号/无符号类型区分。

feat(ast2ir): 实现整数常量表达式求值器

新增了完整的整数常量表达式求值功能,支持C11标准中的常量表达式规则,
包括字面量、标识符、sizeof/_Alignof、一元/二元运算、条件表达式和
类型转换等操作。该功能用于数组大小和枚举值的编译期计算验证。

refactor(ast2ir): 完善类型提升和算术转换机制

改进了整数提升和寻常算术转换的实现,修复了移位操作的符号处理问题,
添加了无符号比较操作的支持,增强了类型安全检查,统一了错误处理流程。

fix(ast2ir): 修复赋值表达式返回值和数组大小计算问题

修正了赋值表达式的返回值处理,确保返回右侧值而不是存储指令引用。
使用新的常量表达式求值器替代原有的硬编码数组大小计算,提高了
数组声明的正确性。
This commit is contained in:
zzy
2026-05-31 17:30:22 +08:00
parent c4467b8420
commit d2eafa9dc6
27 changed files with 2579 additions and 218 deletions

View File

@@ -1,9 +0,0 @@
[package]
name = "scc_abi"
version = "0.1.0"
authors = []
description = ""
# dependencies = []
# features = {}
# default_features = []

View File

@@ -7,7 +7,6 @@ description = ""
dependencies = [
{ name = "scc_ast", path = "../ast" },
{ name = "scc_hir", path = "../ir/hir" },
{ name = "scc_abi", path = "../abi" },
]
# features = {}
# default_features = []

View File

@@ -38,6 +38,10 @@ void scc_ast2ir_stmt(scc_ast2ir_ctx_t *ctx, const scc_ast_stmt_t *stmt);
scc_hir_type_ref_t scc_ast2ir_type(scc_ast2ir_ctx_t *ctx,
const scc_ast_qual_type_t *ast_type);
scc_hir_type_ref_t
scc_ast2ir_parse_base_type(scc_ast2ir_ctx_t *ctx,
const scc_ast_qual_type_t *ast_type);
// ====== 类型提升Type Promotion接口 ======
/**
@@ -47,16 +51,17 @@ scc_hir_type_ref_t scc_ast2ir_type(scc_ast2ir_ctx_t *ctx,
* @return 提升后的类型的引用,若无需提升则返回原始类型
*/
scc_hir_type_ref_t scc_ast2ir_integer_promotion(scc_ast2ir_ctx_t *ctx,
scc_hir_type_ref_t type_ref);
scc_hir_type_ref_t type_ref);
/**
* @brief 寻常算术转换C11 6.3.1.8
* 对二元算术操作的两个操作数类型找到公共类型
* @return 公共类型的引用
*/
scc_hir_type_ref_t scc_ast2ir_usual_arithmetic_conversion(
scc_ast2ir_ctx_t *ctx, scc_hir_type_ref_t t1_ref,
scc_hir_type_ref_t t2_ref);
scc_hir_type_ref_t
scc_ast2ir_usual_arithmetic_conversion(scc_ast2ir_ctx_t *ctx,
scc_hir_type_ref_t t1_ref,
scc_hir_type_ref_t t2_ref);
/**
* @brief 插入类型转换指令
@@ -64,7 +69,25 @@ scc_hir_type_ref_t scc_ast2ir_usual_arithmetic_conversion(
* @return 转换后的值引用
*/
scc_hir_value_ref_t scc_ast2ir_emit_conversion(scc_ast2ir_ctx_t *ctx,
scc_hir_value_ref_t value,
scc_hir_type_ref_t target_type);
scc_hir_value_ref_t value,
scc_hir_type_ref_t target_type);
// ====== 常量表达式求值 ======
/**
* @brief 对整数常量表达式求值C11 6.6
*
* 递归计算 AST 表达式,得到一个编译期可确定的整数值。
* 支持整数字面量、字符字面量、枚举常量、sizeof / _Alignof、
* 一元运算(+ - ~ !)、二元运算(算术/位/移位/关系/逻辑)、
* 条件表达式(?:)、强制类型转换。
*
* @param ctx ast2ir 上下文
* @param result 输出参数,存放求值结果
* @param expr 待求值的 AST 表达式
* @return 成功返回 true若表达式不是合法的常量表达式则返回 false
*/
cbool scc_ast2ir_eval_constant_int(scc_ast2ir_ctx_t *ctx, scc_ap_t *result,
const scc_ast_expr_t *expr);
#endif /* __SCC_AST2IR_H__ */

View File

@@ -24,29 +24,6 @@ static int fixed_param_count(const scc_ast_canon_type_t *canon) {
return n;
}
static scc_hir_type_ref_t parse_base_type(scc_ast2ir_ctx_t *ctx,
const scc_ast_qual_type_t *ast_type) {
scc_abi_type_layout_t layout;
// 映射内置类型
ctx->abi->compute_type_layout(ctx->abi, (void *)ast_type, &layout);
switch (layout.size) {
case 0:
return scc_hir_builder_type_void(&ctx->builder);
case 1:
return scc_hir_builder_type_i8(&ctx->builder);
case 2:
return scc_hir_builder_type_i16(&ctx->builder);
case 4:
return scc_hir_builder_type_i32(&ctx->builder);
case 8:
return scc_hir_builder_type_i64(&ctx->builder);
break;
default:
break;
}
return SCC_HIR_REF_nullptr;
}
static inline bool scc_hir_type_is_signed(scc_hir_type_tag_t tag) {
switch (tag) {
case SCC_HIR_TYPE_i8:
@@ -69,6 +46,7 @@ static inline bool scc_hir_type_is_signed(scc_hir_type_tag_t tag) {
// 判断是否为算术类型(整数或浮点)
static cbool is_arithmetic_type(const scc_hir_type_t *type) {
Assert(type != nullptr);
switch (type->tag) {
case SCC_HIR_TYPE_i8:
case SCC_HIR_TYPE_i16:
@@ -235,7 +213,7 @@ scc_hir_type_ref_t scc_ast2ir_type(scc_ast2ir_ctx_t *ctx,
switch (ast_type->base.type) {
case SCC_AST_TYPE_BUILTIN: {
return parse_base_type(ctx, ast_type);
return scc_ast2ir_parse_base_type(ctx, ast_type);
}
case SCC_AST_TYPE_POINTER: {
scc_hir_type_init(&ir_type, SCC_HIR_TYPE_PTR);
@@ -253,16 +231,11 @@ scc_hir_type_ref_t scc_ast2ir_type(scc_ast2ir_ctx_t *ctx,
ir_type.data.array.base = element_type;
ir_type.data.array.len = 0;
if (scc_ast_canon_type(ast_type)->array.size) {
// TODO constant expression
if (scc_ast_canon_type(ast_type)->array.size->base.type !=
SCC_AST_EXPR_INT_LITERAL) {
Panic("TODO: array size expression");
}
scc_ap_t value;
scc_ap_from_cstr(
&value, scc_ast_canon_type(ast_type)->array.size->literal.lexme,
10);
// FIXME hack
if (!scc_ast2ir_eval_constant_int(
ctx, &value, scc_ast_canon_type(ast_type)->array.size)) {
Panic("array size is not a constant expression");
}
ir_type.data.array.len = value.data.digit;
}
} break;
@@ -359,7 +332,7 @@ scc_hir_type_ref_t scc_ast2ir_type(scc_ast2ir_ctx_t *ctx,
.base.type = SCC_AST_TYPE_BUILTIN,
.type = int_canon_type,
};
return parse_base_type(ctx, &int_type);
return scc_ast2ir_parse_base_type(ctx, &int_type);
case SCC_AST_TYPE_TYPEDEF:
// TODO maybe using cache
return scc_ast2ir_type(
@@ -569,7 +542,8 @@ scc_hir_value_ref_t scc_ast2ir_expr(scc_ast2ir_ctx_t *ctx,
rhs = scc_ast2ir_emit_conversion(
ctx, rhs, lhs_ptr_type->data.pointer.base);
}
return scc_hir_builder_store(&ctx->builder, lhs, rhs);
scc_hir_builder_store(&ctx->builder, lhs, rhs);
return rhs;
}
if (expr->binary.op == SCC_AST_OP_LOGICAL_AND ||
@@ -577,30 +551,36 @@ scc_hir_value_ref_t scc_ast2ir_expr(scc_ast2ir_ctx_t *ctx,
// TODO 类型提升
return scc_ast2ir_logical_expr(ctx, expr);
}
rhs = scc_ast2ir_expr(ctx, expr->binary.rhs, false);
lhs = scc_ast2ir_expr(ctx, expr->binary.lhs, false);
if (expr->binary.op == SCC_AST_OP_COMMA) {
lhs = scc_ast2ir_expr(ctx, expr->binary.lhs, false);
rhs = scc_ast2ir_expr(ctx, expr->binary.rhs, false);
} else {
rhs = scc_ast2ir_expr(ctx, expr->binary.rhs, false);
lhs = scc_ast2ir_expr(ctx, expr->binary.lhs, false);
}
// 类型提升对算术类型二元操作应用寻常算术转换C11 6.3.1.8
// 移位操作符不适用(仅对每个操作数单独整数提升)
cbool is_shift = (expr->binary.op == SCC_AST_OP_LEFT_SHIFT ||
expr->binary.op == SCC_AST_OP_RIGHT_SHIFT);
scc_hir_type_ref_t common_type = 0;
if (!is_shift && expr->binary.op != SCC_AST_OP_COMMA) {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t lhs_type_ref =
scc_hir_module_get_value(module, lhs)->type;
scc_hir_type_ref_t rhs_type_ref =
scc_hir_module_get_value(module, rhs)->type;
Assert(lhs_type_ref != 0 && rhs_type_ref != 0);
scc_hir_type_t *lhs_type =
scc_hir_module_get_type(module, lhs_type_ref);
scc_hir_type_t *rhs_type =
scc_hir_module_get_type(module, rhs_type_ref);
if (is_arithmetic_type(lhs_type) && is_arithmetic_type(rhs_type)) {
scc_hir_type_ref_t common =
scc_ast2ir_usual_arithmetic_conversion(ctx, lhs_type_ref,
rhs_type_ref);
if (common != 0) {
lhs = scc_ast2ir_emit_conversion(ctx, lhs, common);
rhs = scc_ast2ir_emit_conversion(ctx, rhs, common);
common_type = scc_ast2ir_usual_arithmetic_conversion(
ctx, lhs_type_ref, rhs_type_ref);
if (common_type != 0) {
lhs = scc_ast2ir_emit_conversion(ctx, lhs, common_type);
rhs = scc_ast2ir_emit_conversion(ctx, rhs, common_type);
}
}
} else if (is_shift) {
@@ -628,16 +608,42 @@ scc_hir_value_ref_t scc_ast2ir_expr(scc_ast2ir_ctx_t *ctx,
case SCC_AST_OP_BITWISE_XOR: op = SCC_HIR_OP_XOR; break;
case SCC_AST_OP_LEFT_SHIFT: op = SCC_HIR_OP_SHL; break;
case SCC_AST_OP_RIGHT_SHIFT: {
op = SCC_HIR_OP_SHR;
// FIXME op = SCC_HIR_OP_SAR;
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t type_ref = scc_hir_module_get_value(module, lhs)->type;
scc_hir_type_t *type = scc_hir_module_get_type(module, type_ref);
op = scc_hir_type_is_signed(type->tag) ? SCC_HIR_OP_SAR : SCC_HIR_OP_SHR;
break;
}
case SCC_AST_OP_EQUAL: op = SCC_HIR_OP_EQ; break;
case SCC_AST_OP_NOT_EQUAL: op = SCC_HIR_OP_NEQ; break;
case SCC_AST_OP_LESS: op = SCC_HIR_OP_LT; break;
case SCC_AST_OP_LESS_EQUAL: op = SCC_HIR_OP_LE; break;
case SCC_AST_OP_GREATER: op = SCC_HIR_OP_GT; break;
case SCC_AST_OP_GREATER_EQUAL: op = SCC_HIR_OP_GE; break;
case SCC_AST_OP_LESS: {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t tr = common_type ? common_type : scc_hir_module_get_value(module, lhs)->type;
scc_hir_type_t *t = scc_hir_module_get_type(module, tr);
op = scc_hir_type_is_signed(t->tag) ? SCC_HIR_OP_LT : SCC_HIR_OP_ULT;
break;
}
case SCC_AST_OP_LESS_EQUAL: {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t tr = common_type ? common_type : scc_hir_module_get_value(module, lhs)->type;
scc_hir_type_t *t = scc_hir_module_get_type(module, tr);
op = scc_hir_type_is_signed(t->tag) ? SCC_HIR_OP_LE : SCC_HIR_OP_ULE;
break;
}
case SCC_AST_OP_GREATER: {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t tr = common_type ? common_type : scc_hir_module_get_value(module, lhs)->type;
scc_hir_type_t *t = scc_hir_module_get_type(module, tr);
op = scc_hir_type_is_signed(t->tag) ? SCC_HIR_OP_GT : SCC_HIR_OP_UGT;
break;
}
case SCC_AST_OP_GREATER_EQUAL: {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t tr = common_type ? common_type : scc_hir_module_get_value(module, lhs)->type;
scc_hir_type_t *t = scc_hir_module_get_type(module, tr);
op = scc_hir_type_is_signed(t->tag) ? SCC_HIR_OP_GE : SCC_HIR_OP_UGE;
break;
}
case SCC_AST_OP_COMMA: {
// 逗号运算符:计算左表达式,丢弃结果,返回右表达式
return rhs;
@@ -854,28 +860,24 @@ scc_hir_value_ref_t scc_ast2ir_expr(scc_ast2ir_ctx_t *ctx,
scc_hir_value_ref_vec_t args;
scc_vec_init(args);
// 获取函数类型信息用于参数转换。注意:不能长期持有指向 types 向量
// 内部的指针(如 ft->data.function.params因为递归的
// scc_ast2ir_expr 可能触发 scc_hir_module_add_type 导致 realloc。
// 改为通过 func_type_ref稳定的索引每次需要时重新获取指针。
// 获取函数类型信息用于参数转换。
scc_hir_func_ref_t callee_func =
(scc_hir_func_ref_t)(usize)scc_hashtable_get(
&ctx->symtab, expr->call.callee->identifier._target->name);
scc_hir_type_ref_t func_type_ref = SCC_HIR_REF_nullptr;
cbool is_variadic = false;
int fixed_count = 0;
if (callee_func != SCC_HIR_REF_nullptr) {
scc_hir_func_t *hir_func = scc_hir_module_get_func(
scc_ast2ir_mir_module(ctx), callee_func);
scc_hir_func_meta_t *meta = SCC_HIR_FUNC_META(hir_func);
is_variadic = meta->is_variadic;
func_type_ref = meta->type;
if (func_type_ref != SCC_HIR_REF_nullptr) {
const scc_hir_type_t *ft = scc_hir_module_get_type(
scc_ast2ir_mir_module(ctx), func_type_ref);
if (ft != nullptr && ft->tag == SCC_HIR_TYPE_FUNC)
fixed_count = (int)ft->data.function.params.size;
}
Assert(callee_func != SCC_HIR_REF_nullptr);
scc_hir_func_t *hir_func =
scc_hir_module_get_func(scc_ast2ir_mir_module(ctx), callee_func);
scc_hir_func_meta_t *meta = SCC_HIR_FUNC_META(hir_func);
is_variadic = meta->is_variadic;
func_type_ref = meta->type;
if (func_type_ref != SCC_HIR_REF_nullptr) {
const scc_hir_type_t *ft = scc_hir_module_get_type(
scc_ast2ir_mir_module(ctx), func_type_ref);
if (ft != nullptr && ft->tag == SCC_HIR_TYPE_FUNC)
fixed_count = (int)ft->data.function.params.size;
}
scc_vec_foreach(expr->call.args, i) {
@@ -897,8 +899,11 @@ scc_hir_value_ref_t scc_ast2ir_expr(scc_ast2ir_ctx_t *ctx,
// 可变参数只应用默认参数提升C11 6.5.2.2
// 整数提升char/short → int
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_ref_t arg_type_ref =
scc_hir_module_get_value(module, arg_node)->type;
scc_hir_value_t *arg_value =
scc_hir_module_get_value(module, arg_node);
Assert(arg_value != nullptr);
scc_hir_type_ref_t arg_type_ref = arg_value->type;
Assert(arg_type_ref != 0);
scc_hir_type_tag_t arg_tag =
scc_hir_module_get_type(module, arg_type_ref)->tag;
@@ -1136,7 +1141,7 @@ scc_hir_value_ref_t scc_ast2ir_expr(scc_ast2ir_ctx_t *ctx,
// FIXME maybe using some array to int;
scc_hir_type_ref_t type_ref = scc_hir_builder_type_i32(&ctx->builder);
scc_ap_t value;
scc_ap_from_cstr(&value, expr->literal.lexme, 10);
scc_ap_from_cstr(&value, expr->literal.lexme, 0);
return scc_hir_builder_integer(&ctx->builder, type_ref, &value);
}
// SCC_AST_EXPR_FLOAT_LITERAL, // 浮点字面量
@@ -1655,24 +1660,23 @@ void scc_ast2ir_decl(scc_ast2ir_ctx_t *ctx, const scc_ast_decl_t *decl,
} break;
case SCC_AST_DECL_ENUM: {
scc_ap_t val;
int idx = 0;
scc_ap_set_int(&val, idx);
scc_ap_set_int(&val, 0);
scc_vec_foreach(decl->record.fields, i) {
scc_ast_decl_t *item = scc_vec_at(decl->record.fields, i);
Assert(item->base.type == SCC_AST_DECL_VAR);
if (item->var.init) {
Assert(item->var.init->base.type == SCC_AST_EXPR_INT_LITERAL);
scc_ap_from_cstr(&val, item->var.init->literal.lexme, 10);
if (!scc_ast2ir_eval_constant_int(ctx, &val, item->var.init)) {
Panic("enum value is not a constant expression");
}
}
scc_hir_value_ref_t item_val_ref = scc_hir_builder_integer(
&ctx->builder, scc_hir_builder_type_i32(&ctx->builder), &val);
scc_hashtable_set(&ctx->ast2ir_cache, item,
(void *)(usize)item_val_ref);
// FIXME hack ap
idx = val.data.digit;
idx += 1;
// scc_ap_add 1 ();
scc_ap_set_int(&val, idx);
/* next = val + 1 */
scc_ap_t one;
scc_ap_set_int(&one, 1);
scc_ap_add(&val, &val, &one);
}
} break;
case SCC_AST_DECL_TYPEDEF:

View File

@@ -0,0 +1,381 @@
#include <scc_ast2ir.h>
#include <scc_ast_def.h>
#include <scc_ast_utils.h>
/* ---------- internal helpers ---------- */
static cbool is_builtin_signed(scc_ast_builtin_type_t t) {
switch (t) {
case SCC_AST_BUILTIN_TYPE_BOOL:
case SCC_AST_BUILTIN_TYPE_CHAR: /* implementation-defined, treat as signed */
case SCC_AST_BUILTIN_TYPE_SHORT:
case SCC_AST_BUILTIN_TYPE_INT:
case SCC_AST_BUILTIN_TYPE_LONG:
case SCC_AST_BUILTIN_TYPE_LONG_LONG:
case SCC_AST_BUILTIN_TYPE_SIGNED_CHAR:
case SCC_AST_BUILTIN_TYPE_SIGNED_SHORT:
case SCC_AST_BUILTIN_TYPE_SIGNED_INT:
case SCC_AST_BUILTIN_TYPE_SIGNED_LONG:
case SCC_AST_BUILTIN_TYPE_SIGNED_LONG_LONG:
return true;
default:
return false;
}
}
static cbool is_builtin_unsigned(scc_ast_builtin_type_t t) {
switch (t) {
case SCC_AST_BUILTIN_TYPE_UNSIGNED_CHAR:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_SHORT:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_INT:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_LONG:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_LONG_LONG:
return true;
default:
return false;
}
}
/* Check whether an AST qual-type is an integer type (not pointer/float/struct) */
static cbool is_ast_integer_type(const scc_ast_qual_type_t *type) {
if (type->base.type != SCC_AST_TYPE_BUILTIN)
return false;
scc_ast_builtin_type_t bt = scc_ast_canon_type(type)->builtin.type;
return bt >= SCC_AST_BUILTIN_TYPE_BOOL &&
bt <= SCC_AST_BUILTIN_TYPE_UNSIGNED_LONG_LONG;
}
/*
* Truncate / sign-extend an AP value to match a target integer type width.
* Uses the ABI calculator to determine the type's size in bytes.
*/
static cbool apply_integer_truncation(scc_ast2ir_ctx_t *ctx, scc_ap_t *val,
const scc_ast_qual_type_t *target_type) {
if (!is_ast_integer_type(target_type))
return false;
scc_abi_type_layout_t layout;
ctx->abi->compute_type_layout(ctx->abi, (void *)target_type, &layout);
int bits = layout.size * 8;
if (bits >= (int)sizeof(scc_ap_digit) * 8)
return true; /* no truncation needed */
/* Extract magnitude */
uint64_t mag = val->data.digit;
uint64_t mask = ((uint64_t)1 << bits) - 1;
uint64_t truncated = mag & mask;
/* For signed types, sign-extend if the MSB is set */
scc_ast_builtin_type_t bt = scc_ast_canon_type(target_type)->builtin.type;
if (is_builtin_signed(bt) && (truncated >> (bits - 1)))
truncated |= ~mask; /* set all bits above bits-1 */
val->data.digit = truncated;
val->len = (val->len < 0 && truncated == 0) ? 1 : val->len;
return true;
}
/* ---------- unary expression ---------- */
static cbool eval_unary(scc_ast2ir_ctx_t *ctx, scc_ap_t *result,
const scc_ast_expr_t *expr) {
scc_ap_t val;
scc_ap_init(&val);
if (!scc_ast2ir_eval_constant_int(ctx, &val, expr->unary.operand))
return false;
switch (expr->unary.op) {
case SCC_AST_OP_UNARY_PLUS:
result->capacity = -1;
result->len = val.len;
result->data.digit = val.data.digit;
return true;
case SCC_AST_OP_UNARY_MINUS:
scc_ap_neg(result, &val);
return true;
case SCC_AST_OP_BITWISE_NOT:
scc_ap_not(result, &val);
return true;
case SCC_AST_OP_LOGICAL_NOT:
scc_ap_set_int(result, scc_ap_is_zero(&val));
return true;
default:
return false;
}
}
/* ---------- binary expression ---------- */
static cbool eval_binary(scc_ast2ir_ctx_t *ctx, scc_ap_t *result,
const scc_ast_expr_t *expr) {
scc_ap_t lhs_val, rhs_val;
scc_ap_init(&lhs_val);
scc_ap_init(&rhs_val);
if (!scc_ast2ir_eval_constant_int(ctx, &lhs_val, expr->binary.lhs))
return false;
if (!scc_ast2ir_eval_constant_int(ctx, &rhs_val, expr->binary.rhs))
return false;
switch (expr->binary.op) {
/* arithmetic */
case SCC_AST_OP_ADD:
scc_ap_add(result, &lhs_val, &rhs_val);
return true;
case SCC_AST_OP_SUB:
scc_ap_sub(result, &lhs_val, &rhs_val);
return true;
case SCC_AST_OP_MUL:
scc_ap_mul(result, &lhs_val, &rhs_val);
return true;
case SCC_AST_OP_DIV:
scc_ap_div(result, &lhs_val, &rhs_val);
return true;
case SCC_AST_OP_MOD:
scc_ap_mod(result, &lhs_val, &rhs_val);
return true;
/* bitwise */
case SCC_AST_OP_BITWISE_AND:
scc_ap_and(result, &lhs_val, &rhs_val);
return true;
case SCC_AST_OP_BITWISE_OR:
scc_ap_or(result, &lhs_val, &rhs_val);
return true;
case SCC_AST_OP_BITWISE_XOR:
scc_ap_xor(result, &lhs_val, &rhs_val);
return true;
/* shift */
case SCC_AST_OP_LEFT_SHIFT:
scc_ap_shl(result, &lhs_val, (unsigned)rhs_val.data.digit);
return true;
case SCC_AST_OP_RIGHT_SHIFT:
scc_ap_shr(result, &lhs_val, (unsigned)rhs_val.data.digit);
return true;
/* relational */
case SCC_AST_OP_EQUAL:
scc_ap_set_int(result, scc_ap_cmp(&lhs_val, &rhs_val) == 0);
return true;
case SCC_AST_OP_NOT_EQUAL:
scc_ap_set_int(result, scc_ap_cmp(&lhs_val, &rhs_val) != 0);
return true;
case SCC_AST_OP_LESS:
scc_ap_set_int(result, scc_ap_cmp(&lhs_val, &rhs_val) < 0);
return true;
case SCC_AST_OP_GREATER:
scc_ap_set_int(result, scc_ap_cmp(&lhs_val, &rhs_val) > 0);
return true;
case SCC_AST_OP_LESS_EQUAL:
scc_ap_set_int(result, scc_ap_cmp(&lhs_val, &rhs_val) <= 0);
return true;
case SCC_AST_OP_GREATER_EQUAL:
scc_ap_set_int(result, scc_ap_cmp(&lhs_val, &rhs_val) >= 0);
return true;
/* logical */
case SCC_AST_OP_LOGICAL_AND:
if (scc_ap_is_zero(&lhs_val)) {
scc_ap_set_int(result, 0);
} else {
scc_ap_set_int(result, !scc_ap_is_zero(&rhs_val));
}
return true;
case SCC_AST_OP_LOGICAL_OR:
if (!scc_ap_is_zero(&lhs_val)) {
scc_ap_set_int(result, 1);
} else {
scc_ap_set_int(result, !scc_ap_is_zero(&rhs_val));
}
return true;
default:
return false;
}
}
/* ---------- identifier (enum constant lookup) ---------- */
static cbool eval_identifier(scc_ast2ir_ctx_t *ctx, scc_ap_t *result,
const scc_ast_expr_t *expr) {
if (expr->identifier._target == nullptr)
return false;
/* Try the ast2ir_cache — enum values are stored there as HIR integer
* constants during scc_ast2ir_decl(SCC_AST_DECL_ENUM) */
scc_hir_value_ref_t cached = (scc_hir_value_ref_t)(usize)scc_hashtable_get(
&ctx->ast2ir_cache, expr->identifier._target);
if (cached != 0) {
scc_hir_value_t *hir_val =
scc_hir_module_get_value(scc_ast2ir_mir_module(ctx), cached);
if (hir_val != nullptr && hir_val->tag == SCC_HIR_VALUE_TAG_INTEGER) {
/* shallow copy — digit mode has no heap allocation */
result->capacity = -1;
result->len = hir_val->data.integer.len;
result->data.digit = hir_val->data.integer.data.digit;
return true;
}
}
/* If not cached yet, evaluate the var init recursively.
* This handles enum-internal forward references within a single enum decl:
* enum { A = 1, B = A + 1 }; — when evaluating B, A may not be cached
* yet, but we can look at A's init expression. */
if (expr->identifier._target->base.type == SCC_AST_DECL_VAR &&
expr->identifier._target->var.init != nullptr) {
return scc_ast2ir_eval_constant_int(ctx, result,
expr->identifier._target->var.init);
}
return false;
}
/* ---------- character literal evaluation ---------- */
static int eval_char_literal(const scc_ast_expr_t *expr) {
const char *lexme = expr->literal.lexme;
Assert(lexme[0] == '\'');
if (lexme[1] != '\\')
return (unsigned char)lexme[1];
switch (lexme[2]) {
case 'a':
return '\a';
case 'b':
return '\b';
case 'f':
return '\f';
case 'n':
return '\n';
case 'r':
return '\r';
case 't':
return '\t';
case 'v':
return '\v';
case '\\':
return '\\';
case '\'':
return '\'';
case '"':
return '"';
case '0':
return '\0';
case 'x':
/* hex escape: \xNN — parse up to 2 hex digits */
{
int val = 0;
int j = 3;
while (1) {
char c = lexme[j];
unsigned d;
if (c >= '0' && c <= '9')
d = c - '0';
else if (c >= 'a' && c <= 'f')
d = c - 'a' + 10;
else if (c >= 'A' && c <= 'F')
d = c - 'A' + 10;
else
break;
val = val * 16 + d;
j++;
}
return val;
}
default:
/* octal: \nnn */
if (lexme[2] >= '0' && lexme[2] <= '7') {
int val = 0;
int j = 2;
while (j < 5 && lexme[j] >= '0' && lexme[j] <= '7') {
val = val * 8 + (lexme[j] - '0');
j++;
}
return val;
}
return (unsigned char)lexme[2];
}
}
/* ===================================================================
* Public API
* =================================================================== */
cbool scc_ast2ir_eval_constant_int(scc_ast2ir_ctx_t *ctx, scc_ap_t *result,
const scc_ast_expr_t *expr) {
if (ctx == nullptr || result == nullptr || expr == nullptr)
return false;
scc_ap_init(result);
switch (expr->base.type) {
case SCC_AST_EXPR_INT_LITERAL:
scc_ap_from_cstr(result, expr->literal.lexme, 0);
return true;
case SCC_AST_EXPR_CHAR_LITERAL:
scc_ap_set_int(result, eval_char_literal(expr));
return true;
case SCC_AST_EXPR_BINARY:
return eval_binary(ctx, result, expr);
case SCC_AST_EXPR_UNARY:
return eval_unary(ctx, result, expr);
case SCC_AST_EXPR_COND: {
scc_ap_t cond_val;
scc_ap_init(&cond_val);
if (!scc_ast2ir_eval_constant_int(ctx, &cond_val, expr->cond.cond))
return false;
return scc_ast2ir_eval_constant_int(ctx, result,
scc_ap_is_zero(&cond_val)
? expr->cond.else_expr
: expr->cond.then_expr);
}
case SCC_AST_EXPR_CAST: {
if (!scc_ast2ir_eval_constant_int(ctx, result, expr->cast.expr))
return false;
/* Truncate / sign-extend to the target type width */
if (!apply_integer_truncation(ctx, result, expr->cast.type))
return false;
return true;
}
case SCC_AST_EXPR_SIZE_OF: {
scc_abi_type_layout_t layout;
if (expr->attr_of.type) {
ctx->abi->compute_type_layout(ctx->abi, (void *)expr->attr_of.type,
&layout);
} else if (expr->attr_of.expr) {
/* sizeof expression: try to get the type through the ABI */
/* For constant expressions, sema should have annotated the type */
/* TODO: handle sizeof(expr) where expr is not a type */
return false;
} else {
return false;
}
scc_ap_set_int(result, layout.size);
return true;
}
case SCC_AST_EXPR_ALIGN_OF: {
scc_abi_type_layout_t layout;
if (expr->attr_of.type) {
ctx->abi->compute_type_layout(ctx->abi, (void *)expr->attr_of.type,
&layout);
} else if (expr->attr_of.expr) {
return false;
} else {
return false;
}
scc_ap_set_int(result, layout.alignment);
return true;
}
case SCC_AST_EXPR_IDENTIFIER:
return eval_identifier(ctx, result, expr);
default:
return false;
}
}

View File

@@ -179,9 +179,10 @@ scc_hir_type_ref_t scc_ast2ir_integer_promotion(scc_ast2ir_ctx_t *ctx,
* d. 如果有符号类型可以表示无符号类型的所有值,无符号转换为有符号
* e. 否则,两个都转换为有符号类型的无符号版本
*/
scc_hir_type_ref_t scc_ast2ir_usual_arithmetic_conversion(
scc_ast2ir_ctx_t *ctx, scc_hir_type_ref_t t1_ref,
scc_hir_type_ref_t t2_ref) {
scc_hir_type_ref_t
scc_ast2ir_usual_arithmetic_conversion(scc_ast2ir_ctx_t *ctx,
scc_hir_type_ref_t t1_ref,
scc_hir_type_ref_t t2_ref) {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_t *t1 = scc_hir_module_get_type(module, t1_ref);
scc_hir_type_t *t2 = scc_hir_module_get_type(module, t2_ref);
@@ -200,13 +201,9 @@ scc_hir_type_ref_t scc_ast2ir_usual_arithmetic_conversion(
scc_hir_type_ref_t p2_ref = scc_ast2ir_integer_promotion(ctx, t2_ref);
scc_hir_type_tag_t tag1 =
(p1_ref != 0)
? scc_hir_module_get_type(module, p1_ref)->tag
: t1->tag;
(p1_ref != 0) ? scc_hir_module_get_type(module, p1_ref)->tag : t1->tag;
scc_hir_type_tag_t tag2 =
(p2_ref != 0)
? scc_hir_module_get_type(module, p2_ref)->tag
: t2->tag;
(p2_ref != 0) ? scc_hir_module_get_type(module, p2_ref)->tag : t2->tag;
// 如果提升后相同,直接返回
if (tag1 == tag2)
@@ -252,6 +249,7 @@ scc_hir_type_ref_t scc_ast2ir_usual_arithmetic_conversion(
* - 返回 -1 表示无需转换(同类型同大小)
*/
static int determine_conv_kind(scc_hir_type_tag_t src, scc_hir_type_tag_t dst) {
Assert(SCC_HIR_TYPE_VALID(src) && SCC_HIR_TYPE_VALID(dst));
if (src == dst)
return -1;
@@ -302,10 +300,8 @@ scc_hir_value_ref_t scc_ast2ir_emit_conversion(scc_ast2ir_ctx_t *ctx,
scc_hir_value_ref_t value,
scc_hir_type_ref_t target_type) {
scc_hir_module_t *module = scc_ast2ir_mir_module(ctx);
scc_hir_type_t *src_type =
scc_hir_module_get_type_by_value(module, value);
scc_hir_type_t *dst_type =
scc_hir_module_get_type(module, target_type);
scc_hir_type_t *src_type = scc_hir_module_get_type_by_value(module, value);
scc_hir_type_t *dst_type = scc_hir_module_get_type(module, target_type);
if (src_type == nullptr || dst_type == nullptr)
return value;

View File

@@ -0,0 +1,76 @@
#include <scc_ast2ir.h>
scc_hir_type_ref_t
scc_ast2ir_parse_base_type(scc_ast2ir_ctx_t *ctx,
const scc_ast_qual_type_t *ast_type) {
scc_abi_type_layout_t layout;
// 映射内置类型
ctx->abi->compute_type_layout(ctx->abi, (void *)ast_type, &layout);
switch (scc_ast_canon_type(ast_type)->builtin.type) {
// FIXME it not a void
case SCC_AST_BUILTIN_TYPE_VA_LIST:
case SCC_AST_BUILTIN_TYPE_VOID:
return scc_hir_builder_type_void(&ctx->builder);
case SCC_AST_BUILTIN_TYPE_BOOL:
case SCC_AST_BUILTIN_TYPE_SIGNED_CHAR:
case SCC_AST_BUILTIN_TYPE_SHORT:
case SCC_AST_BUILTIN_TYPE_SIGNED_SHORT:
case SCC_AST_BUILTIN_TYPE_INT:
case SCC_AST_BUILTIN_TYPE_SIGNED_INT:
case SCC_AST_BUILTIN_TYPE_LONG:
case SCC_AST_BUILTIN_TYPE_SIGNED_LONG:
case SCC_AST_BUILTIN_TYPE_LONG_LONG:
case SCC_AST_BUILTIN_TYPE_SIGNED_LONG_LONG:
switch (layout.size) {
case 1:
return scc_hir_builder_type_i8(&ctx->builder);
case 2:
return scc_hir_builder_type_i16(&ctx->builder);
case 4:
return scc_hir_builder_type_i32(&ctx->builder);
case 8:
return scc_hir_builder_type_i64(&ctx->builder);
break;
default:
break;
}
break;
case SCC_AST_BUILTIN_TYPE_CHAR:
// TODO CHAR TO BE UNSIGNED CHAR
case SCC_AST_BUILTIN_TYPE_UNSIGNED_CHAR:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_SHORT:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_INT:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_LONG:
case SCC_AST_BUILTIN_TYPE_UNSIGNED_LONG_LONG:
switch (layout.size) {
case 1:
return scc_hir_builder_type_u8(&ctx->builder);
case 2:
return scc_hir_builder_type_u16(&ctx->builder);
case 4:
return scc_hir_builder_type_u32(&ctx->builder);
case 8:
return scc_hir_builder_type_u64(&ctx->builder);
break;
default:
break;
}
break;
case SCC_AST_BUILTIN_TYPE_FLOAT:
case SCC_AST_BUILTIN_TYPE_DOUBLE:
switch (layout.size) {
case 4:
return scc_hir_builder_type_f32(&ctx->builder);
case 8:
return scc_hir_builder_type_f64(&ctx->builder);
break;
default:
break;
}
default:
Panic("Unsupported AST type: %d",
scc_ast_canon_type(ast_type)->builtin.type);
break;
}
return SCC_HIR_REF_nullptr;
}

View File

@@ -48,7 +48,9 @@ typedef enum scc_hir_type_tag {
SCC_HIR_TYPE_STRUCT,
SCC_HIR_TYPE_UNION,
SCC_HIR_TYPE_VECTOR, // TODO SIMD
SCC_HIR_TYPE_COUNT,
} scc_hir_type_tag_t;
#define SCC_HIR_TYPE_VALID(tag) ((tag) < SCC_HIR_TYPE_COUNT && (tag) >= 0)
struct scc_hir_type {
scc_hir_type_tag_t tag;
@@ -134,6 +136,14 @@ typedef enum {
SCC_HIR_OP_SHR,
/// Shift right arithmetic.
SCC_HIR_OP_SAR,
/// Unsigned greater than.
SCC_HIR_OP_UGT,
/// Unsigned less than.
SCC_HIR_OP_ULT,
/// Unsigned greater than or equal to.
SCC_HIR_OP_UGE,
/// Unsigned less than or equal to.
SCC_HIR_OP_ULE,
} scc_hir_op_type_t;
typedef enum {

View File

@@ -49,7 +49,9 @@ static const char *get_op_str(scc_hir_op_type_t op) {
[SCC_HIR_OP_AND] = "&", [SCC_HIR_OP_OR] = "|",
[SCC_HIR_OP_XOR] = "^", [SCC_HIR_OP_NOT] = "~",
[SCC_HIR_OP_SHL] = "<<", [SCC_HIR_OP_SHR] = ">>",
[SCC_HIR_OP_SAR] = ">>a",
[SCC_HIR_OP_SAR] = ">>a", [SCC_HIR_OP_ULT] = "u<",
[SCC_HIR_OP_ULE] = "u<=", [SCC_HIR_OP_UGT] = "u>",
[SCC_HIR_OP_UGE] = "u>=",
};
if (op >= 0 && (usize)op < sizeof(ops) / sizeof(ops[0]) && ops[op])
return ops[op];

View File

@@ -174,7 +174,6 @@ static scc_lir_cond_t map_cmp_cond(scc_hir_op_type_t op, cbool is_float) {
Panic("invalid float cmp");
}
} else {
// 默认为有符号比较 (无符号需额外处理类型)
switch (op) {
case SCC_HIR_OP_EQ:
return SCC_LIR_COND_EQ;
@@ -188,6 +187,14 @@ static scc_lir_cond_t map_cmp_cond(scc_hir_op_type_t op, cbool is_float) {
return SCC_LIR_COND_SGT;
case SCC_HIR_OP_GE:
return SCC_LIR_COND_SGE;
case SCC_HIR_OP_ULT:
return SCC_LIR_COND_ULT;
case SCC_HIR_OP_ULE:
return SCC_LIR_COND_ULE;
case SCC_HIR_OP_UGT:
return SCC_LIR_COND_UGT;
case SCC_HIR_OP_UGE:
return SCC_LIR_COND_UGE;
default:
Panic("invalid int cmp");
}
@@ -196,6 +203,54 @@ static scc_lir_cond_t map_cmp_cond(scc_hir_op_type_t op, cbool is_float) {
}
static scc_lir_op_t map_binop(scc_hir_op_type_t op, cbool is_float) {
switch (op) {
case SCC_HIR_OP_EMPTY:
return SCC_LIR_NOP;
/// Not equal to.
case SCC_HIR_OP_NEQ:
/// Equal to.
case SCC_HIR_OP_EQ:
/// Greater than.
case SCC_HIR_OP_GT:
/// Less than.
case SCC_HIR_OP_LT:
/// Greater than or equal to.
case SCC_HIR_OP_GE:
/// Less than or equal to.
case SCC_HIR_OP_LE:
/// Unsigned greater than.
case SCC_HIR_OP_UGT:
/// Unsigned less than.
case SCC_HIR_OP_ULT:
/// Unsigned greater than or equal to.
case SCC_HIR_OP_UGE:
/// Unsigned less than or equal to.
case SCC_HIR_OP_ULE:
return SCC_LIR_CMP;
/// Bitwise AND.
case SCC_HIR_OP_AND:
return SCC_LIR_AND;
/// Bitwise OR.
case SCC_HIR_OP_OR:
return SCC_LIR_OR;
/// Bitwise XOR.
case SCC_HIR_OP_XOR:
return SCC_LIR_XOR;
/// Bitwise NOT.
case SCC_HIR_OP_NOT:
return SCC_LIR_NOT;
/// Shift left logical.
case SCC_HIR_OP_SHL:
return SCC_LIR_SHL;
/// Shift right logical.
case SCC_HIR_OP_SHR:
return SCC_LIR_SHR;
/// Shift right arithmetic.
case SCC_HIR_OP_SAR:
return SCC_LIR_SAR;
default:
break;
}
if (is_float) {
switch (op) {
case SCC_HIR_OP_ADD:
@@ -206,6 +261,9 @@ static scc_lir_op_t map_binop(scc_hir_op_type_t op, cbool is_float) {
return SCC_LIR_FMUL;
case SCC_HIR_OP_DIV:
return SCC_LIR_FDIV;
case SCC_HIR_OP_MOD:
TODO();
return SCC_LIR_FNEG;
default:
Panic("unsupported float binop");
}
@@ -221,20 +279,8 @@ static scc_lir_op_t map_binop(scc_hir_op_type_t op, cbool is_float) {
return SCC_LIR_DIV_S;
case SCC_HIR_OP_MOD:
return SCC_LIR_REM_S;
case SCC_HIR_OP_AND:
return SCC_LIR_AND;
case SCC_HIR_OP_OR:
return SCC_LIR_OR;
case SCC_HIR_OP_XOR:
return SCC_LIR_XOR;
case SCC_HIR_OP_SHL:
return SCC_LIR_SHL;
case SCC_HIR_OP_SHR:
return SCC_LIR_SHR;
case SCC_HIR_OP_SAR:
return SCC_LIR_SAR;
default:
return SCC_LIR_CMP;
Panic("unsupported binop");
}
}
return SCC_LIR_NOP;
@@ -277,7 +323,10 @@ static void translate_hir_value(ir2lir_ctx_t *ctx, scc_hir_value_t *value,
switch (value->tag) {
case SCC_HIR_VALUE_TAG_OP: {
scc_lir_val_t lhs = ir_value_to_lir_operand(ctx, value->data.op.lhs);
scc_lir_val_t rhs = ir_value_to_lir_operand(ctx, value->data.op.rhs);
scc_lir_val_t rhs = SCC_LIR_NONE();
if (value->data.op.rhs != SCC_HIR_REF_nullptr) {
rhs = ir_value_to_lir_operand(ctx, value->data.op.rhs);
}
scc_lir_op_t op = map_binop(value->data.op.op, is_float);
scc_lir_instr_t instr = {
@@ -308,9 +357,18 @@ static void translate_hir_value(ir2lir_ctx_t *ctx, scc_hir_value_t *value,
ir_value_to_lir_operand(ctx, value->data.store.value);
scc_lir_val_t addr =
ir_value_to_lir_operand(ctx, value->data.store.target);
ty = scc_hir_module_get_type_by_value(ctx->hir_module,
value->data.store.value);
ir_type_to_lir_size_ext(ty, &size, &ext);
/* 修复: 使用目标指针的元素类型确定 store 宽度而非源值类型 */
scc_hir_type_t *ptr_type = scc_hir_module_get_type_by_value(
ctx->hir_module, value->data.store.target);
if (ptr_type && ptr_type->tag == SCC_HIR_TYPE_PTR) {
scc_hir_type_t *elem_type = scc_hir_module_get_type(
ctx->hir_module, ptr_type->data.pointer.base);
ir_type_to_lir_size_ext(elem_type, &size, &ext);
} else {
ty = scc_hir_module_get_type_by_value(ctx->hir_module,
value->data.store.value);
ir_type_to_lir_size_ext(ty, &size, &ext);
}
scc_lir_instr_t instr = {.op = SCC_LIR_STORE,
.ext = ext,
.size = size,
@@ -446,10 +504,9 @@ static void translate_hir_value(ir2lir_ctx_t *ctx, scc_hir_value_t *value,
} else {
// 计算源类型宽度
scc_hir_type_t *src_type = scc_hir_module_get_type(
ctx->hir_module,
scc_hir_module_get_value(ctx->hir_module,
value->data.conv.operand)
->type);
ctx->hir_module, scc_hir_module_get_value(
ctx->hir_module, value->data.conv.operand)
->type);
int from_size = scc_hir_module_type_size(ctx->hir_module, src_type);
scc_lir_instr_t instr = {.op = SCC_LIR_EXTEND,
.size = size,

View File

@@ -3,6 +3,15 @@
#include "../core_pass/scc_reg_alloc.h"
/* 默认位掩码策略R8/R9 被排除ABI 参数寄存器),剩余 6 个通用寄存器 */
void scc_reg_alloc_fill_arch_x86(scc_reg_alloc_op_t *ops);
/*
* 顺序分配策略:光标递增,永不回收。
* 池中包含 R8-R15共 8 个寄存器),每分配一个光标前进一格。
* release_reg 为空操作;寄存器耗尽后 acquire_reg 返回 -1
* 由分配器回退为纯栈操作。
*/
void scc_reg_alloc_fill_seq_x86(scc_reg_alloc_op_t *ops);
#endif /* __SCC_X86_REG_ALLOC_H__ */

View File

@@ -148,6 +148,17 @@ static scc_x86_operand_value_t new_vreg_temp(scc_x86_64_isel_t *isel,
return scc_x86_op_vreg(scc_mir_alloc_vreg(isel->func), size);
}
/* 确保操作数为寄存器: IMM → 加载到临时虚拟寄存器 */
static scc_x86_operand_value_t ensure_reg(scc_x86_64_isel_t *isel,
scc_x86_operand_value_t op) {
if (op.kind == SCC_X86_OPR_IMM) {
scc_x86_operand_value_t tmp = new_vreg_temp(isel, op.size);
scc_x86_emit_move(isel, tmp, op);
return tmp;
}
return op;
}
static scc_x86_iform_t cond_to_setcc(scc_lir_cond_t cond) {
switch (cond) {
case SCC_LIR_COND_EQ:
@@ -343,9 +354,11 @@ static void sel_mir(scc_x86_64_isel_t *isel, const scc_lir_instr_t *instr) {
break;
/* ---- 一元运算 ---- */
case SCC_LIR_NEG:
emit_copy_if_needed(isel, dst, src0);
add_instr_1(isel, SCC_X86_IFORM_NEG_GPRV, dst);
break;
case SCC_LIR_NOT:
emit_copy_if_needed(isel, dst, src0);
add_instr_1(isel, SCC_X86_IFORM_NOT_GPRV, dst);
break;
/* ---- 算术/逻辑二元运算 ---- */
@@ -416,11 +429,18 @@ static void sel_mir(scc_x86_64_isel_t *isel, const scc_lir_instr_t *instr) {
scc_x86_operand_value_t rax = scc_x86_op_preg(SCC_X86_REG_RAX, size);
scc_x86_operand_value_t rdx = scc_x86_op_preg(SCC_X86_REG_RDX, size);
scc_x86_emit_move(isel, rax, src0);
if (instr->op == SCC_LIR_DIV_S || instr->op == SCC_LIR_REM_S) {
if (size < 8) {
// TODO 可能需要在lir进行size对齐
scc_x86_operand_value_t rax64 =
scc_x86_op_preg(SCC_X86_REG_RAX, 8);
add_instr_2(isel, SCC_X86_IFORM_MOVSXD_GPRV_GPRZ, rax64, src0);
} else {
scc_x86_emit_move(isel, rax, src0);
}
add_instr_0(isel, SCC_X86_IFORM_CQO);
} else {
scc_x86_emit_move(isel, rax, src0);
scc_x86_operand_value_t zero = scc_x86_op_imm(0, size);
scc_x86_emit_move(isel, rdx, zero);
}
@@ -429,7 +449,12 @@ static void sel_mir(scc_x86_64_isel_t *isel, const scc_lir_instr_t *instr) {
(instr->op == SCC_LIR_DIV_S || instr->op == SCC_LIR_REM_S)
? SCC_X86_IFORM_IDIV_GPRV
: SCC_X86_IFORM_DIV_GPRV;
add_instr_1(isel, div_if, src1);
scc_x86_operand_value_t divisor = src1;
if (src1.kind == SCC_X86_OPR_IMM) {
divisor = new_vreg_temp(isel, size);
scc_x86_emit_move(isel, divisor, src1);
}
add_instr_1(isel, div_if, divisor);
if (instr->op == SCC_LIR_REM_S || instr->op == SCC_LIR_REM_U)
scc_x86_emit_move(isel, dst, rdx);
@@ -440,6 +465,11 @@ static void sel_mir(scc_x86_64_isel_t *isel, const scc_lir_instr_t *instr) {
/* ---- 比较指令 ---- */
case SCC_LIR_CMP: {
Assert(src0.size == src1.size);
if (src0.kind == SCC_X86_OPR_IMM) {
scc_x86_operand_value_t tmp_reg = new_vreg_temp(isel, src0.size);
scc_x86_emit_move(isel, tmp_reg, src0);
src0 = tmp_reg;
}
if (scc_x86_op_is_vreg(&src0) && src1.kind == SCC_X86_OPR_IMM)
// SCC_X86_IFORM_CMP_GPR8_IMMB_82R7 历史遗留指令在amd64中已不再使用
add_instr_2(isel,
@@ -511,6 +541,7 @@ static void sel_mir(scc_x86_64_isel_t *isel, const scc_lir_instr_t *instr) {
int from_size = instr->metadata.extend.from_size;
scc_x86_operand_value_t ext_src =
scc_x86_lir_val_to_mir_op(isel, &instr->arg0, from_size);
ext_src = ensure_reg(isel, ext_src);
if (instr->ext == SCC_LIR_EXT_ZEXT) {
if (from_size == 4) {

View File

@@ -42,12 +42,27 @@ scc_mir_x86_mov_mem_imm(scc_x86_operand_value_t op0,
: SCC_X86_IFORM_MOV_MEMV_IMMZ;
}
static inline u8 get_op_size(scc_x86_operand_value_t op, cbool get_value) {
if (get_value) {
return op.size;
}
if (op.kind == SCC_X86_OPR_MEM) {
return 8;
}
if (op.kind == SCC_X86_OPR_RELOC && op.reloc.kind == SCC_X86_OPR_MEM) {
return 8;
}
return op.size;
}
void scc_x86_emit_move_to_vec(scc_mir_x86_instr_vec_t *vec,
scc_x86_operand_value_t dst,
scc_x86_operand_value_t src) {
if (dst.size != src.size) {
LOG_WARN("Mismatched register sizes for move %d != %d", dst.size,
src.size);
u8 dst_size = get_op_size(dst, false);
u8 src_size = get_op_size(src, false);
if (dst_size != src_size) {
LOG_FATAL("Mismatched register sizes for move %d != %d", dst_size,
src_size);
}
scc_mir_x86_instr_t ins;
if (dst.kind == SCC_X86_OPR_REG) {
@@ -60,7 +75,7 @@ void scc_x86_emit_move_to_vec(scc_mir_x86_instr_vec_t *vec,
} else if (src.kind == SCC_X86_OPR_MEM ||
(src.kind == SCC_X86_OPR_RELOC &&
src.reloc.target == SCC_X86_RELOC_TARGET_SYMBOL)) {
Assert(dst.size == 8);
Assert(dst_size == 8);
scc_mir_x86_instr_2(&ins, SCC_X86_IFORM_LEA_GPRV_AGEN, dst, src,
scc_pos_create());
} else {
@@ -85,8 +100,11 @@ void scc_x86_emit_move_to_vec(scc_mir_x86_instr_vec_t *vec,
void scc_x86_emit_load_to_vec(scc_mir_x86_instr_vec_t *vec,
scc_x86_operand_value_t dst,
scc_x86_operand_value_t src_addr) {
if (dst.size != src_addr.size) {
LOG_WARN("Mismatched sizes for load %d != %d", dst.size, src_addr.size);
u8 dst_size = get_op_size(dst, false);
u8 src_addr_size = get_op_size(src_addr, true);
if (dst_size != src_addr_size) {
LOG_FATAL("Mismatched sizes for load %d != %d", dst_size,
src_addr_size);
}
Assert(dst.kind == SCC_X86_OPR_REG);
scc_x86_operand_value_t mem_op;
@@ -99,9 +117,9 @@ void scc_x86_emit_load_to_vec(scc_mir_x86_instr_vec_t *vec,
.index = SCC_X86_REG_INVALID,
.scale = 1,
.disp.displacement = 0,
.disp.displacement_bits = src_addr.size * 8,
.disp.displacement_bits = src_addr_size * 8,
},
.size = src_addr.size,
.size = src_addr_size,
};
} else if (src_addr.kind == SCC_X86_OPR_MEM) {
mem_op = src_addr;
@@ -117,9 +135,11 @@ void scc_x86_emit_load_to_vec(scc_mir_x86_instr_vec_t *vec,
void scc_x86_emit_store_to_vec(scc_mir_x86_instr_vec_t *vec,
scc_x86_operand_value_t dst_addr,
scc_x86_operand_value_t src) {
if (dst_addr.size != src.size) {
LOG_WARN("Mismatched sizes for store %d != %d", dst_addr.size,
src.size);
u8 dst_addr_size = get_op_size(dst_addr, true);
u8 src_size = get_op_size(src, false);
if (dst_addr_size != src_size) {
LOG_FATAL("Mismatched sizes for store %d != %d", dst_addr_size,
src_size);
}
scc_x86_operand_value_t mem_op;
if (dst_addr.kind == SCC_X86_OPR_REG) {
@@ -131,9 +151,9 @@ void scc_x86_emit_store_to_vec(scc_mir_x86_instr_vec_t *vec,
.index = SCC_X86_REG_INVALID,
.scale = 1,
.disp.displacement = 0,
.disp.displacement_bits = dst_addr.size * 8,
.disp.displacement_bits = dst_addr_size * 8,
},
.size = dst_addr.size,
.size = dst_addr_size,
};
} else if (dst_addr.kind == SCC_X86_OPR_MEM) {
mem_op = dst_addr;

View File

@@ -73,16 +73,25 @@ static void get_implicit_regs(void *ctx, int opcode, const int **uses,
}
}
/* ========== 寄存器池 ========== */
/* ========== 默认寄存器池(位掩码策略) ========== */
static const int reg_pool[] = {
SCC_X86_REG_R8, SCC_X86_REG_R9, SCC_X86_REG_R10, SCC_X86_REG_R11,
SCC_X86_REG_R12, SCC_X86_REG_R13, SCC_X86_REG_R14, SCC_X86_REG_R15};
/*
* R8/R9 被排除在通用寄存器池之外,因为它们被 Win64 ABI 用作参数传递寄存器
* RCX=arg0, RDX=arg1, R8=arg2, R9=arg3
* 如果把它们放在池中,当 lower_call 发出 MOV R8, vreg 时,
* 分配器可能把 R8 分配给源 vreg导致冗余的 MOV R8, R8。
*
* 可用寄存器: R10, R11, R12, R13, R14, R15
*/
static const int reg_pool[] = {SCC_X86_REG_R10, SCC_X86_REG_R11,
SCC_X86_REG_R12, SCC_X86_REG_R13,
SCC_X86_REG_R14, SCC_X86_REG_R15};
#define REG_POOL_SIZE ((int)SCC_ARRLEN(reg_pool))
static uint32_t reg_mask = 0;
static int acquire_reg(void *ctx) {
(void)ctx;
for (int i = 0; i < SCC_ARRLEN(reg_pool); i++)
for (int i = 0; i < REG_POOL_SIZE; i++)
if (!(reg_mask & (1u << i))) {
reg_mask |= (1u << i);
return reg_pool[i];
@@ -92,7 +101,7 @@ static int acquire_reg(void *ctx) {
static void release_reg(void *ctx, int preg) {
(void)ctx;
for (int i = 0; i < SCC_ARRLEN(reg_pool); i++)
for (int i = 0; i < REG_POOL_SIZE; i++)
if (reg_pool[i] == preg) {
reg_mask &= ~(1u << i);
return;
@@ -101,7 +110,7 @@ static void release_reg(void *ctx, int preg) {
static void mark_reg_used(void *ctx, int preg) {
(void)ctx;
for (int i = 0; i < SCC_ARRLEN(reg_pool); i++)
for (int i = 0; i < REG_POOL_SIZE; i++)
if (reg_pool[i] == preg) {
reg_mask |= (1u << i);
return;
@@ -113,6 +122,57 @@ static void clean_mark_regs(void *ctx) {
reg_mask = 0;
}
/* ========== 顺序分配策略(光标,永不回收) ========== */
/*
* 顺序分配策略按固定顺序分配寄存器R8 → R9 → ... → R15
* 一旦分配就永不回收release_reg 为空操作)。
* 当所有 8 个寄存器耗尽后acquire_reg 返回 -1
* 分配器将 vreg 直接替换为栈槽spill/reload 由内存操作数隐式完成)。
*
* 包含 R8/R9 是安全的,因为顺序分配避免了冲突:
* - 如果 R8 已被分配,参数 MOV R8, vreg 中的 vreg 会分配下一个可用寄存器(如
* R10 然后通过 MOV R8, R10 将值移入 R8。
* - 如果 R8 尚未分配,参数 MOV R8, vreg 的 vreg 会分配到 R8
* 产生 MOV R8, R8空操作这在功能上是正确的。
*/
static const int seq_pool[] = {
SCC_X86_REG_R8, SCC_X86_REG_R9, SCC_X86_REG_R10, SCC_X86_REG_R11,
SCC_X86_REG_R12, SCC_X86_REG_R13, SCC_X86_REG_R14, SCC_X86_REG_R15};
#define SEQ_POOL_SIZE ((int)SCC_ARRLEN(seq_pool))
static int seq_cursor = 0;
static int seq_acquire_reg(void *ctx) {
(void)ctx;
if (seq_cursor < SEQ_POOL_SIZE)
return seq_pool[seq_cursor++];
return -1; /* 寄存器耗尽 */
}
static void seq_release_reg(void *ctx, int preg) {
(void)ctx;
(void)preg;
/* 永不回收:释放为空操作 */
}
static void seq_mark_reg_used(void *ctx, int preg) {
(void)ctx;
for (int i = seq_cursor; i < SEQ_POOL_SIZE; i++)
if (seq_pool[i] == preg) {
/* 将光标移到该寄存器之后(燃烧) */
if (i + 1 > seq_cursor)
seq_cursor = i + 1;
return;
}
}
static void seq_clean_mark_regs(void *ctx) {
(void)ctx;
seq_cursor = 0;
}
/* ========== 指令迭代器(两种策略共享) ========== */
static void x86_alloc_iter_begin(scc_reg_alloc_iter_t *iter) {
iter->op_idx = 0;
iter->op_sub_idx = 0;
@@ -128,8 +188,8 @@ static cbool x86_alloc_iter_next(scc_reg_alloc_iter_t *iter, int *out_vreg,
while (iter->op_idx < num_ops) {
const scc_x86_operand_value_t *op =
&ins->x86_instr.operands[iter->op_idx];
scc_reg_op_access_t base_access = SCC_REG_ALLOC_OP_ACCESS_READWRITE;
// base_access = get_operand_access(opcode, iter->op_idx);
scc_reg_op_access_t base_access =
get_operand_access(opcode, iter->op_idx);
*out_size = op->size;
if (op->kind == SCC_X86_OPR_REG) {
@@ -223,6 +283,8 @@ static void x86_alloc_iter_end(scc_reg_alloc_iter_t *iter) {
(void)iter; // 无需清理
}
/* ========== 回调表填充 ========== */
void scc_reg_alloc_fill_arch_x86(scc_reg_alloc_op_t *ops) {
ops->acquire_reg = acquire_reg;
ops->release_reg = release_reg;
@@ -240,3 +302,21 @@ void scc_reg_alloc_fill_arch_x86(scc_reg_alloc_op_t *ops) {
ops->emit_reload = emit_reload;
ops->emit_copy = emit_copy;
}
void scc_reg_alloc_fill_seq_x86(scc_reg_alloc_op_t *ops) {
ops->acquire_reg = seq_acquire_reg;
ops->release_reg = seq_release_reg;
ops->mark_reg_used = seq_mark_reg_used;
ops->clean_mark_regs = seq_clean_mark_regs;
ops->alloc_iter_begin = x86_alloc_iter_begin;
ops->alloc_iter_next = x86_alloc_iter_next;
ops->alloc_iter_replace_preg = x86_alloc_iter_replace_preg;
ops->alloc_iter_replace_slot = x86_alloc_iter_replace_slot;
ops->alloc_iter_end = x86_alloc_iter_end;
ops->get_implicit_regs = get_implicit_regs;
ops->emit_spill = emit_spill;
ops->emit_reload = emit_reload;
ops->emit_copy = emit_copy;
}

View File

@@ -49,16 +49,21 @@ static void alloc_instr(scc_reg_alloc_ctx_t *ctx,
if (mapping == 0)
slot = scc_mir_vreg_map2slot(ctx->func, vreg, size, 8);
preg = ops->acquire_reg(ctx);
scc_vec_push(allocated, preg);
if (preg >= 0) {
scc_vec_push(allocated, preg);
if (access == SCC_REG_ALLOC_OP_ACCESS_READ ||
access == SCC_REG_ALLOC_OP_ACCESS_READWRITE)
ops->emit_reload(&before, preg, slot, size);
if (access == SCC_REG_ALLOC_OP_ACCESS_WRITE ||
access == SCC_REG_ALLOC_OP_ACCESS_READWRITE)
ops->emit_spill(&after, preg, slot, size);
if (access == SCC_REG_ALLOC_OP_ACCESS_READ ||
access == SCC_REG_ALLOC_OP_ACCESS_READWRITE)
ops->emit_reload(&before, preg, slot, size);
if (access == SCC_REG_ALLOC_OP_ACCESS_WRITE ||
access == SCC_REG_ALLOC_OP_ACCESS_READWRITE)
ops->emit_spill(&after, preg, slot, size);
ops->alloc_iter_replace_preg(&iter, preg, size);
ops->alloc_iter_replace_preg(&iter, preg, size);
} else {
/* 寄存器耗尽顺序策略vreg 直接使用栈槽 */
ops->alloc_iter_replace_slot(&iter, slot, size);
}
}
ops->alloc_iter_end(&iter);

View File

@@ -171,6 +171,10 @@ void scc_win_pc_x64_prolog_epilog_init(scc_prolog_epilog_t *ctx) {
ctx->need_epilog = need_epilog;
}
/*
windows x64 calling convention
https://learn.microsoft.com/zh-cn/cpp/build/x64-calling-convention?view=msvc-180#varargs
*/
static void lower_call(void *userdata, const scc_lir_instr_t *instr) {
scc_x86_64_isel_t *isel = userdata;

View File

@@ -1,9 +1,9 @@
[package]
name = "ap"
version = "0.1.0"
authors = []
name = "ap"
version = "0.1.0"
authors = []
description = ""
# dependencies = []
dependencies = [{ name = "core", path = "../scc_core" }]
# features = {}
# default_features = []

View File

@@ -4,7 +4,7 @@
* @brief Arbitrary Precision Library
*
*/
#ifdef __SCC__
#ifndef __NO_SCC_RUNTIME__
#include <scc_core.h>
#define SCC_AP_DIGIT u64
#define SCC_AP_PANIC Panic
@@ -49,21 +49,21 @@ typedef struct {
} scc_ap_t;
static inline void scc_ap_init(scc_ap_t *ap) {
ap->len = 0;
ap->capacity = -1;
ap->len = 1;
ap->data.digit = 0;
}
static inline void scc_ap_set_int(scc_ap_t *ap, int val) {
ap->capacity = -1;
SCC_AP_ASSERT(sizeof(scc_ap_digit) >= sizeof(int));
if (val < 0) {
ap->len = -1;
ap->data.digit = (scc_ap_digit)(-(int64_t)val);
} else {
ap->len = 1;
ap->data.digit = (scc_ap_digit)val;
}
SCC_AP_ASSERT(sizeof(scc_ap_digit) >= sizeof(int));
ap->capacity = -1;
ap->data.digit = val;
}
void scc_ap_drop(scc_ap_t *ap);
@@ -77,13 +77,20 @@ void scc_ap_mul(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b);
void scc_ap_div(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b);
void scc_ap_mod(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b);
/**
* @brief equal
*
* @param a
* @param b
* @return int
*/
void scc_ap_shl(scc_ap_t *to, const scc_ap_t *from, unsigned shift);
void scc_ap_shr(scc_ap_t *to, const scc_ap_t *from, unsigned shift);
void scc_ap_lshr(scc_ap_t *to, const scc_ap_t *from, unsigned shift);
void scc_ap_and(scc_ap_t *to, const scc_ap_t *a, const scc_ap_t *b);
void scc_ap_or(scc_ap_t *to, const scc_ap_t *a, const scc_ap_t *b);
void scc_ap_xor(scc_ap_t *to, const scc_ap_t *a, const scc_ap_t *b);
void scc_ap_not(scc_ap_t *to, const scc_ap_t *from);
void scc_ap_neg(scc_ap_t *to, const scc_ap_t *from);
int scc_ap_cmp(const scc_ap_t *a, const scc_ap_t *b);
int scc_ap_is_zero(const scc_ap_t *ap);
int scc_ap_eql(const scc_ap_t *a, const scc_ap_t *b);
typedef void (*ap_dump_fn)(const char ch, void *user_data);

View File

@@ -1,48 +1,394 @@
#include <ap.h>
static void *scc_ap_alloc(size_t size) {
void *p = SCC_AP_MALLOC(size);
SCC_AP_ASSERT(p != NULL);
return p;
/*
* Fast-path (hack) implementation: all values stored in single-digit mode.
* data.digit holds the absolute magnitude as u64, len sign indicates sign.
* Arithmetic is done via isize with explicit overflow checks — anything
* out of isize range triggers SCC_AP_PANIC.
*/
/* ---------- internal helpers ---------- */
/* Convert from sign-magnitude (digit mode) to isize */
static isize ap_to_isize(const scc_ap_t *ap) {
SCC_AP_ASSERT(ap != nullptr);
SCC_AP_ASSERT(ap->capacity == -1);
uint64_t mag = ap->data.digit;
if (ap->len < 0) {
if (mag == 0)
return 0;
if (mag == (uint64_t)INTPTR_MAX + 1)
return INTPTR_MIN;
SCC_AP_ASSERT(mag <= (uint64_t)INTPTR_MAX);
return -(isize)(int64_t)mag;
}
SCC_AP_ASSERT(mag <= (uint64_t)INTPTR_MAX);
return (isize)(int64_t)mag;
}
/* 重新分配内存 */
static void *scc_ap_realloc_data(void *old, size_t new_size) {
void *p = SCC_AP_REALLOC(old, new_size);
SCC_AP_ASSERT(p != NULL);
return p;
/* Store isize value into sign-magnitude (digit mode) */
static void ap_from_isize(scc_ap_t *ap, isize val) {
ap->capacity = -1;
if (val < 0) {
ap->len = -1;
ap->data.digit = (scc_ap_digit)(-(uint64_t)(int64_t)val);
} else {
ap->len = 1;
ap->data.digit = (scc_ap_digit)val;
}
}
/* Checked addition */
static isize ap_add_isize(isize a, isize b) {
if (b > 0 && a > INTPTR_MAX - b)
SCC_AP_PANIC("scc_ap_add: positive overflow");
if (b < 0 && a < INTPTR_MIN - b)
SCC_AP_PANIC("scc_ap_add: negative overflow");
return a + b;
}
/* Checked subtraction: a - b */
static isize ap_sub_isize(isize a, isize b) {
/* a - b = a + (-b); -INTPTR_MIN overflows, handle specially */
if (b == INTPTR_MIN) {
if (a < 0)
SCC_AP_PANIC("scc_ap_sub: overflow (a - INTPTR_MIN, a < 0)");
SCC_AP_PANIC("scc_ap_sub: overflow (a - INTPTR_MIN, a >= 0)");
}
return ap_add_isize(a, -b);
}
/* Checked multiplication */
static isize ap_mul_isize(isize a, isize b) {
if (a == 0 || b == 0)
return 0;
if (a == 1)
return b;
if (b == 1)
return a;
/* Determine sign of result */
int sign = ((a < 0) == (b < 0)) ? 1 : -1;
/* Absolute values as uint64 */
uint64_t ua = (a == INTPTR_MIN)
? (uint64_t)INTPTR_MAX + 1
: (uint64_t)(a < 0 ? -(int64_t)a : (int64_t)a);
uint64_t ub = (b == INTPTR_MIN)
? (uint64_t)INTPTR_MAX + 1
: (uint64_t)(b < 0 ? -(int64_t)b : (int64_t)b);
/* Overflow threshold */
uint64_t limit =
(sign > 0) ? (uint64_t)INTPTR_MAX : (uint64_t)INTPTR_MAX + 1;
if (ua != 0 && ub > limit / ua)
SCC_AP_PANIC("scc_ap_mul: overflow");
isize result = (isize)(ua * ub);
return (sign > 0) ? result : -result;
}
/* Checked division */
static isize ap_div_isize(isize a, isize b) {
if (b == 0)
SCC_AP_PANIC("scc_ap_div: division by zero");
if (a == INTPTR_MIN && b == -1)
SCC_AP_PANIC("scc_ap_div: overflow (INTPTR_MIN / -1)");
return a / b;
}
/* Checked modulo */
static isize ap_mod_isize(isize a, isize b) {
if (b == 0)
SCC_AP_PANIC("scc_ap_mod: modulo by zero");
if (a == INTPTR_MIN && b == -1)
SCC_AP_PANIC("scc_ap_mod: overflow (INTPTR_MIN %% -1)");
return a % b;
}
/* ---------- public API ---------- */
void scc_ap_drop(scc_ap_t *ap) {
SCC_AP_ASSERT(ap);
// SCC_AP_FREE(ap->data);
SCC_AP_ASSERT(ap != nullptr);
/* In digit mode (capacity == -1), there's no heap allocation to free */
scc_ap_init(ap);
}
void scc_ap_with_bits(scc_ap_t *ap, int bits) {}
void scc_ap_from_string(scc_ap_t *ap, const char *str, int len, int base) {
SCC_AP_ASSERT(base == 10);
SCC_AP_ASSERT(ap && str);
if (len == 0) {
scc_ap_set_int(ap, 0);
}
// FIXME
int int_lit = 0;
for (int i = 0; i < len; i += 1) {
int_lit = int_lit * 10 + (str[i] - '0');
}
scc_ap_set_int(ap, int_lit);
void scc_ap_with_bits(scc_ap_t *ap, int bits) {
SCC_AP_ASSERT(ap != nullptr);
if (bits > (int)(sizeof(scc_ap_digit) * 8))
SCC_AP_PANIC(
"scc_ap_with_bits: request %d bits but digit only has %zu bits",
bits, sizeof(scc_ap_digit) * 8);
(void)ap;
/* Fast path: always stays in digit mode */
}
void scc_ap_set_digit(scc_ap_t *ap, scc_ap_digit digit) {}
void scc_ap_from_string(scc_ap_t *ap, const char *str, int len, int base) {
SCC_AP_ASSERT(ap != nullptr && str != nullptr);
void scc_ap_add_impl(scc_ap_t *to, const scc_ap_t *from_a,
const scc_ap_t *from_b) {}
/* len < 0 means null-terminated */
if (len < 0) {
len = 0;
while (str[len] != '\0')
len++;
}
if (len == 0) {
ap_from_isize(ap, 0);
return;
}
void scc_ap_add(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {}
int i = 0;
int sign = 1;
void scc_ap_sub(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {}
/* Optional leading sign */
if (str[i] == '-') {
sign = -1;
i++;
} else if (str[i] == '+') {
i++;
}
typedef void (*ap_dump_fn)(const char ch, void *user_data);
int scc_ap_dump(scc_ap_t *ap, ap_dump_fn dump_fn, void *user_data) { return 0; }
if (i >= len) {
ap_from_isize(ap, 0);
return;
}
/* ---------- base auto-detection ---------- */
/* hex: 0x / 0X */
if ((base == 0 || base == 16) &&
(i + 2 < len && str[i] == '0' &&
(str[i + 1] == 'x' || str[i + 1] == 'X'))) {
base = 16;
i += 2;
}
/* binary: 0b / 0B */
if ((base == 0 || base == 2) &&
(i + 2 < len && str[i] == '0' &&
(str[i + 1] == 'b' || str[i + 1] == 'B'))) {
base = 2;
i += 2;
}
if (base == 0) {
if (i < len && str[i] == '0') {
base = 8;
i++;
} else {
base = 10;
}
}
/* ---------- parse digits ---------- */
uint64_t val = 0;
for (; i < len; i++) {
char ch = str[i];
unsigned d;
if (ch >= '0' && ch <= '9')
d = (unsigned)(ch - '0');
else if (ch >= 'a' && ch <= 'f')
d = (unsigned)(ch - 'a' + 10);
else if (ch >= 'A' && ch <= 'F')
d = (unsigned)(ch - 'A' + 10);
else
break;
if ((int)d >= base)
break;
/* overflow check: val * base + d <= UINT64_MAX */
if (val > (UINT64_MAX - d) / (unsigned)base)
SCC_AP_PANIC("scc_ap_from_string: overflow");
val = val * (unsigned)base + d;
}
ap->capacity = -1;
ap->data.digit = val;
ap->len = (sign < 0 && val > 0) ? -1 : 1;
}
void scc_ap_set_digit(scc_ap_t *ap, scc_ap_digit digit) {
SCC_AP_ASSERT(ap != nullptr);
if (digit > (scc_ap_digit)INTPTR_MAX)
SCC_AP_PANIC("scc_ap_set_digit: value %llu exceeds isize range",
(unsigned long long)digit);
ap->capacity = -1;
ap->data.digit = digit;
ap->len = 1;
}
void scc_ap_add(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {
SCC_AP_ASSERT(to != nullptr && from_a != nullptr && from_b != nullptr);
isize a = ap_to_isize(from_a);
isize b = ap_to_isize(from_b);
ap_from_isize(to, ap_add_isize(a, b));
}
void scc_ap_sub(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {
SCC_AP_ASSERT(to != nullptr && from_a != nullptr && from_b != nullptr);
isize a = ap_to_isize(from_a);
isize b = ap_to_isize(from_b);
ap_from_isize(to, ap_sub_isize(a, b));
}
void scc_ap_mul(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {
SCC_AP_ASSERT(to != nullptr && from_a != nullptr && from_b != nullptr);
isize a = ap_to_isize(from_a);
isize b = ap_to_isize(from_b);
ap_from_isize(to, ap_mul_isize(a, b));
}
void scc_ap_div(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {
SCC_AP_ASSERT(to != nullptr && from_a != nullptr && from_b != nullptr);
isize a = ap_to_isize(from_a);
isize b = ap_to_isize(from_b);
ap_from_isize(to, ap_div_isize(a, b));
}
void scc_ap_mod(scc_ap_t *to, const scc_ap_t *from_a, const scc_ap_t *from_b) {
SCC_AP_ASSERT(to != nullptr && from_a != nullptr && from_b != nullptr);
isize a = ap_to_isize(from_a);
isize b = ap_to_isize(from_b);
ap_from_isize(to, ap_mod_isize(a, b));
}
void scc_ap_shl(scc_ap_t *to, const scc_ap_t *from, unsigned shift) {
SCC_AP_ASSERT(to != nullptr && from != nullptr);
SCC_AP_ASSERT(from->capacity == -1);
if (shift >= sizeof(scc_ap_digit) * 8)
SCC_AP_PANIC("scc_ap_shl: shift %u exceeds digit width (%zu bits)",
shift, sizeof(scc_ap_digit) * 8);
/* left shift = multiply by 2^shift */
isize val = ap_to_isize(from);
isize result = ap_mul_isize(val, (isize)1 << shift);
ap_from_isize(to, result);
}
void scc_ap_shr(scc_ap_t *to, const scc_ap_t *from, unsigned shift) {
SCC_AP_ASSERT(to != nullptr && from != nullptr);
SCC_AP_ASSERT(from->capacity == -1);
if (shift >= sizeof(scc_ap_digit) * 8)
SCC_AP_PANIC("scc_ap_shr: shift %u exceeds digit width (%zu bits)",
shift, sizeof(scc_ap_digit) * 8);
/* arithmetic right shift — uses >> on isize (arithmetic on GCC/MSVC/Clang) */
isize val = ap_to_isize(from);
ap_from_isize(to, val >> shift);
}
void scc_ap_lshr(scc_ap_t *to, const scc_ap_t *from, unsigned shift) {
SCC_AP_ASSERT(to != nullptr && from != nullptr);
SCC_AP_ASSERT(from->capacity == -1);
if (shift >= sizeof(scc_ap_digit) * 8)
SCC_AP_PANIC("scc_ap_lshr: shift %u exceeds digit width (%zu bits)",
shift, sizeof(scc_ap_digit) * 8);
/*
* logical right shift: reinterpret the signed value as uint64 bit pattern,
* shift right, store as non-negative magnitude.
*/
uint64_t bits;
if (from->len < 0) {
/* negative → two's complement bit pattern */
bits = -(uint64_t)(int64_t)from->data.digit;
} else {
bits = from->data.digit;
}
to->capacity = -1;
to->data.digit = bits >> shift;
to->len = 1;
}
/* ---------- bitwise ---------- */
void scc_ap_and(scc_ap_t *to, const scc_ap_t *a, const scc_ap_t *b) {
SCC_AP_ASSERT(to != nullptr && a != nullptr && b != nullptr);
ap_from_isize(to, ap_to_isize(a) & ap_to_isize(b));
}
void scc_ap_or(scc_ap_t *to, const scc_ap_t *a, const scc_ap_t *b) {
SCC_AP_ASSERT(to != nullptr && a != nullptr && b != nullptr);
ap_from_isize(to, ap_to_isize(a) | ap_to_isize(b));
}
void scc_ap_xor(scc_ap_t *to, const scc_ap_t *a, const scc_ap_t *b) {
SCC_AP_ASSERT(to != nullptr && a != nullptr && b != nullptr);
ap_from_isize(to, ap_to_isize(a) ^ ap_to_isize(b));
}
void scc_ap_not(scc_ap_t *to, const scc_ap_t *from) {
SCC_AP_ASSERT(to != nullptr && from != nullptr);
ap_from_isize(to, ~ap_to_isize(from));
}
/* ---------- unary arithmetic ---------- */
void scc_ap_neg(scc_ap_t *to, const scc_ap_t *from) {
SCC_AP_ASSERT(to != nullptr && from != nullptr);
isize val = ap_to_isize(from);
if (val == INTPTR_MIN)
SCC_AP_PANIC("scc_ap_neg: overflow (INTPTR_MIN)");
ap_from_isize(to, -val);
}
/* ---------- comparison ---------- */
int scc_ap_cmp(const scc_ap_t *a, const scc_ap_t *b) {
SCC_AP_ASSERT(a != nullptr && b != nullptr);
isize va = ap_to_isize(a);
isize vb = ap_to_isize(b);
if (va < vb) return -1;
if (va > vb) return 1;
return 0;
}
int scc_ap_is_zero(const scc_ap_t *ap) {
SCC_AP_ASSERT(ap != nullptr);
return ap->data.digit == 0;
}
int scc_ap_eql(const scc_ap_t *a, const scc_ap_t *b) {
SCC_AP_ASSERT(a != nullptr && b != nullptr);
SCC_AP_ASSERT(a->capacity == -1 && b->capacity == -1);
if (a->len != b->len) {
/* -0 == +0 */
if (a->data.digit == 0 && b->data.digit == 0)
return 1;
return 0;
}
return a->data.digit == b->data.digit;
}
int scc_ap_dump(scc_ap_t *ap, ap_dump_fn dump_fn, void *user_data) {
SCC_AP_ASSERT(ap != nullptr && dump_fn != nullptr);
SCC_AP_ASSERT(ap->capacity == -1);
isize val = ap_to_isize(ap);
if (val == 0) {
dump_fn('0', user_data);
return 0;
}
/* Buffer large enough for INT64_MIN "-9223372036854775808" */
char buf[24];
int pos = sizeof(buf);
if (val < 0) {
dump_fn('-', user_data);
/* Work with positive magnitude to avoid overflow when printing */
uint64_t mag = ap->data.digit;
while (mag > 0) {
buf[--pos] = (char)('0' + (mag % 10));
mag /= 10;
}
} else {
uint64_t mag = (uint64_t)val;
while (mag > 0) {
buf[--pos] = (char)('0' + (mag % 10));
mag /= 10;
}
}
for (int i = pos; i < (int)sizeof(buf); i++)
dump_fn(buf[i], user_data);
return 0;
}

1028
runtime/ap/tests/test_ap.c Normal file

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,184 @@
#include <ap.h>
#include <stdio.h>
#include <string.h>
#include <utest/acutest.h>
/*
* Panic / overflow tests — each test calls SCC_AP_PANIC which aborts the
* process. Run this file separately from the normal tests.
*/
/* ---------- helper: attempt an operation and report result ---------- */
#define TRY_PANIC(call, label) \
do { \
fprintf(stderr, " " label " ... "); \
call; \
fprintf(stderr, "FAIL (no panic)\n"); \
TEST_CHECK(0 && "expected panic"); \
} while (0)
void test_add_overflow_positive(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "9223372036854775807", -1, 10);
scc_ap_set_int(&b, 1);
TRY_PANIC(scc_ap_add(&r, &a, &b), "INTPTR_MAX + 1");
}
void test_add_overflow_negative(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "-9223372036854775808", -1, 10);
scc_ap_set_int(&b, -1);
TRY_PANIC(scc_ap_add(&r, &a, &b), "INTPTR_MIN + (-1)");
}
void test_mul_overflow(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "4611686018427387904", -1, 10); /* INT64_MAX/2 + 1 */
scc_ap_set_int(&b, 2);
TRY_PANIC(scc_ap_mul(&r, &a, &b), "(INT64_MAX/2+1) * 2");
}
void test_div_by_zero(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_set_int(&a, 1);
scc_ap_set_int(&b, 0);
TRY_PANIC(scc_ap_div(&r, &a, &b), "1 / 0");
}
void test_mod_by_zero(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_set_int(&a, 1);
scc_ap_set_int(&b, 0);
TRY_PANIC(scc_ap_mod(&r, &a, &b), "1 % 0");
}
void test_from_string_overflow(void) {
scc_ap_t ap;
scc_ap_init(&ap);
TRY_PANIC(scc_ap_from_string(&ap, "18446744073709551615", -1, 10),
"UINT64_MAX as string");
}
void test_set_digit_overflow(void) {
scc_ap_t ap;
scc_ap_init(&ap);
TRY_PANIC(scc_ap_set_digit(&ap, (scc_ap_digit)18446744073709551615ULL),
"set_digit(UINT64_MAX)");
}
void test_sub_underflow(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "-9223372036854775808", -1, 10);
scc_ap_set_int(&b, 1);
TRY_PANIC(scc_ap_sub(&r, &a, &b), "INTPTR_MIN - 1");
}
void test_div_overflow(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "-9223372036854775808", -1, 10);
scc_ap_set_int(&b, -1);
TRY_PANIC(scc_ap_div(&r, &a, &b), "INTPTR_MIN / -1");
}
void test_mod_overflow(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "-9223372036854775808", -1, 10);
scc_ap_set_int(&b, -1);
TRY_PANIC(scc_ap_mod(&r, &a, &b), "INTPTR_MIN % -1");
}
void test_mul_overflow_neg_limit(void) {
scc_ap_t a, b, r;
scc_ap_init(&a);
scc_ap_init(&b);
scc_ap_init(&r);
scc_ap_from_string(&a, "-9223372036854775808", -1, 10);
scc_ap_set_int(&b, 2);
TRY_PANIC(scc_ap_mul(&r, &a, &b), "INTPTR_MIN * 2");
}
void test_with_bits_overflow(void) {
scc_ap_t ap;
scc_ap_init(&ap);
TRY_PANIC(scc_ap_with_bits(&ap, 65), "with_bits(65)");
}
/* ---------- shift overflow tests ---------- */
void test_shl_overflow(void) {
scc_ap_t a, r;
scc_ap_init(&a);
scc_ap_init(&r);
scc_ap_from_string(&a, "4611686018427387904", -1, 10); /* INT64_MAX/2+1 */
TRY_PANIC(scc_ap_shl(&r, &a, 1), "(INT64_MAX/2+1) << 1");
}
void test_shl_shift_ge_64(void) {
scc_ap_t a, r;
scc_ap_init(&a);
scc_ap_init(&r);
scc_ap_set_int(&a, 1);
TRY_PANIC(scc_ap_shl(&r, &a, 64), "1 << 64");
}
void test_shr_shift_ge_64(void) {
scc_ap_t a, r;
scc_ap_init(&a);
scc_ap_init(&r);
scc_ap_set_int(&a, 1);
TRY_PANIC(scc_ap_shr(&r, &a, 64), "1 >> 64");
}
void test_lshr_shift_ge_64(void) {
scc_ap_t a, r;
scc_ap_init(&a);
scc_ap_init(&r);
scc_ap_set_int(&a, 1);
TRY_PANIC(scc_ap_lshr(&r, &a, 64), "1 lshr 64");
}
TEST_LIST = {
{"test_add_overflow_positive", test_add_overflow_positive},
{"test_add_overflow_negative", test_add_overflow_negative},
{"test_mul_overflow", test_mul_overflow},
{"test_mul_overflow_neg_limit", test_mul_overflow_neg_limit},
{"test_div_by_zero", test_div_by_zero},
{"test_div_overflow", test_div_overflow},
{"test_mod_by_zero", test_mod_by_zero},
{"test_mod_overflow", test_mod_overflow},
{"test_sub_underflow", test_sub_underflow},
{"test_from_string_overflow", test_from_string_overflow},
{"test_set_digit_overflow", test_set_digit_overflow},
{"test_with_bits_overflow", test_with_bits_overflow},
/* shift overflow */
{"test_shl_overflow", test_shl_overflow},
{"test_shl_shift_ge_64", test_shl_shift_ge_64},
{"test_shr_shift_ge_64", test_shr_shift_ge_64},
{"test_lshr_shift_ge_64", test_lshr_shift_ge_64},
{nullptr, nullptr},
};

View File

@@ -12,6 +12,9 @@ typedef struct {
cbool dump_relocs;
cbool dump_hex;
cbool dump_all;
const char *get_section_name;
int get_section_idx;
cbool has_get_section_idx;
int verbose;
} sccfdump_config_t;
@@ -188,6 +191,8 @@ int main(int argc, const char **argv, const char **envp) {
HINT_HEX,
HINT_VERBOSE,
HINT_ALL,
HINT_GET_SECTION,
HINT_GET_SECTION_IDX,
};
static const char *hints_en[] = {
[HINT_PROG_NAME] = "sccfdump",
@@ -201,6 +206,8 @@ int main(int argc, const char **argv, const char **envp) {
[HINT_HEX] = "Dump section hex content",
[HINT_VERBOSE] = "Verbose output (use with -x)",
[HINT_ALL] = "Dump header, sections, symbols and relocs",
[HINT_GET_SECTION] = "Extract raw section data by name (e.g. .text)",
[HINT_GET_SECTION_IDX] = "Extract raw section data by index",
};
static const char *hints_zh[] = {
[HINT_PROG_NAME] = "sccfdump",
@@ -214,6 +221,8 @@ int main(int argc, const char **argv, const char **envp) {
[HINT_HEX] = "转存段的十六进制内容",
[HINT_VERBOSE] = "详细输出(与-x配合使用)",
[HINT_ALL] = "转存文件头、段、符号和重定位",
[HINT_GET_SECTION] = "按名称提取段原始数据(如 .text)",
[HINT_GET_SECTION_IDX] = "按索引提取段原始数据",
};
scc_argparse_lang_t lang = SCC_ARGPARSE_LANG_EN;
@@ -228,6 +237,7 @@ int main(int argc, const char **argv, const char **envp) {
sccfdump_config_t config;
scc_memset(&config, 0, sizeof(config));
config.dump_all = true;
config.get_section_idx = -1;
scc_argparse_t argparse;
scc_argparse_init(&argparse, hints[HINT_PROG_NAME], hints[HINT_DESCRIPTION]);
@@ -275,6 +285,16 @@ int main(int argc, const char **argv, const char **envp) {
scc_argparse_spec_setup_count(&opt_verbose.spec, &config.verbose);
scc_argparse_cmd_add_opt(root, &opt_verbose);
scc_argparse_opt_t opt_get_section;
scc_argparse_opt_init(&opt_get_section, 0, "get-section", hints[HINT_GET_SECTION]);
scc_argparse_spec_setup_string(&opt_get_section.spec, &config.get_section_name);
scc_argparse_cmd_add_opt(root, &opt_get_section);
scc_argparse_opt_t opt_get_section_idx;
scc_argparse_opt_init(&opt_get_section_idx, 0, "get-section-idx", hints[HINT_GET_SECTION_IDX]);
scc_argparse_spec_setup_int(&opt_get_section_idx.spec, &config.get_section_idx);
scc_argparse_cmd_add_opt(root, &opt_get_section_idx);
scc_argparse_arg_t arg_file;
scc_argparse_arg_init(&arg_file, "file", hints[HINT_FILE]);
scc_argparse_spec_setup_string(&arg_file.spec, &config.filepath);
@@ -318,6 +338,60 @@ int main(int argc, const char **argv, const char **envp) {
sccf_parse(&sccf, &buf, 0);
u8 *base = scc_vec_unsafe_get_data(buf);
/* --get-section / --get-section-idx: extract raw data to a file/stdout */
if (config.get_section_name || config.get_section_idx >= 0) {
usize idx = (usize)-1;
if (config.get_section_idx >= 0) {
idx = (usize)config.get_section_idx;
} else {
const sccf_header_t *hdr = (const sccf_header_t *)base;
for (usize i = 0; i < (usize)hdr->sect_header_num; i++) {
const sccf_sect_header_t *sh = sccf_sect_header(base, i);
if (sh && scc_memcmp(sh->name,
config.get_section_name, 8) == 0) {
idx = i;
break;
}
}
if (idx == (usize)-1) {
LOG_ERROR("section not found: %s", config.get_section_name);
scc_vec_free(buf);
return 1;
}
}
const sccf_sect_header_t *sh = sccf_sect_header(base, idx);
if (!sh) {
LOG_ERROR("invalid section index: %d", config.get_section_idx);
scc_vec_free(buf);
return 1;
}
u8 *sdata = sccf_sect_data(base, idx);
if (!sdata || sh->size == 0) {
scc_vec_free(buf);
return 0;
}
if (config.output && scc_strcmp(config.output, "-") == 0) {
scc_fwrite(scc_stdout, sdata, sh->size);
} else if (config.output) {
scc_file_t ofp = scc_fopen(config.output, SCC_FILE_WRITE);
if (!ofp) {
LOG_ERROR("cannot open output: %s", config.output);
scc_vec_free(buf);
return 1;
}
scc_fwrite(ofp, sdata, sh->size);
scc_fclose(ofp);
} else {
scc_fwrite(scc_stdout, sdata, sh->size);
}
scc_vec_free(buf);
return 0;
}
/* --hex without -o: also go to stdout */
if (config.dump_all) {
dump_header(&sccf.header);
dump_sections(base);

View File

@@ -7,6 +7,7 @@ executes the resulting binary, and validates the process return code or stdout.
from __future__ import annotations
import argparse
import difflib
import logging
import os
import subprocess
@@ -23,7 +24,8 @@ from typing import Sequence
# Configuration
# ---------------------------------------------------------------------------
WORKSPACE = Path(__file__).resolve().parent
CC_PATH = WORKSPACE / "../../build/dev/scc"
CC_ROOT = WORKSPACE / "../.."
CC_PATH = CC_ROOT / "build/dev/scc"
CONFIG_PATH = WORKSPACE / "expect.toml"
DEFAULT_TIMEOUT = 1 # seconds
@@ -99,8 +101,15 @@ class Runner:
text=True,
timeout=self.timeout,
)
except subprocess.TimeoutExpired:
return False, "Compilation timed out"
except subprocess.TimeoutExpired as e:
# e.stdout, e.stderr 为已捕获的部分输出(字符串)
captured_stderr = e.stderr.strip() if e.stderr else ""
captured_stdout = e.stdout.strip() if e.stdout else ""
err_msg = f"""Compilation timed out.
stderr: {captured_stderr or '(no output)'}
stdout: {captured_stdout or '(no output)'}
cmd: {" ".join(cmd)}"""
return False, err_msg
except OSError as exc:
return False, f"Failed to invoke compiler: {exc}"
@@ -127,6 +136,26 @@ class Runner:
return proc.returncode, proc.stdout, proc.stderr.strip()
def _print_diff(self, expected: str, actual: str) -> None:
"""Print unified diff between expected and actual stdout."""
expected_lines = expected.splitlines()
actual_lines = actual.splitlines()
diff = difflib.unified_diff(
expected_lines, actual_lines,
fromfile='expected', tofile='actual',
lineterm=''
)
# diff = difflib.ndiff(
# expected_lines, actual_lines,
# )
diff_lines = list(diff)
if not diff_lines:
logger.error(" Strings differ but no diff generated?")
return
logger.error(" Diff (expected vs actual):")
for line in diff_lines:
logger.error(" %s", line)
def run_one(self, test: TestCase) -> bool:
"""Run a single test case. Returns True if passed."""
logger.info("Testing %s", test.source)
@@ -157,9 +186,14 @@ class Runner:
if passed:
logger.info(" PASSED")
else:
logger.error(
" FAILED: expected %r, got %r", test.expected, actual
)
if test.test_type == "stdout":
logger.error(" FAILED: stdout mismatch")
self._print_diff(test.expected, actual)
else:
logger.error(
" FAILED: expected return code %r, got %r",
test.expected, actual
)
# 4. Cleanup
self._remove(exe_path)
@@ -326,12 +360,12 @@ def main() -> None:
runner = Runner(
cc=args.cc,
workspace=WORKSPACE,
workspace=CC_ROOT,
timeout=args.timeout,
keep_temps=args.keep_temps,
)
global_timeout = DEFAULT_TIMEOUT * 3
global_timeout = args.timeout * 3
passed, total = runner.run_all(selected, global_timeout)
logger.info("=" * 40)
if global_timeout and total < len(selected):