feat(cast): 指针隐式转换新规则 + None 可隐式转任意类型 + 诊断错位修复(9 文件)

MemoryOps._ClassifyPtrImplicitConvert:同类型放行;层数不同禁;任一侧通用指针(i8*)放行;皆非通用按被指类型(非基本类型禁、基本标量按位宽无损);指针与非基本值互换禁

_IsNoneValue 收敛:AST None 常量 或 指针型 null 常量 → 放行任意目标(指针目标给 null,非指针目标给零值);整数 0 不误命中;短路口覆盖 9 处禁令

诊断:新增 BaseGen._DescribeValueNode(子表达式下沉 + 有符号列号 + 源码片段),9 处禁令统一使用,不再打印外层 Assign 与 4294967295 列号

注:TPC 侧规则③比 TPV 更严(指针→非基本值恒禁),可能与既有按值回读冲突;待实机编译确认后再统一两侧口径
This commit is contained in:
TermiNexus committed 2026-09-28 15:10:39 +08:00
1 parent 35de6321f6
commit d7158bd336
9 files changed
+318 -66

No files matched your search

+49 -24
View File
@@ -218,31 +218,43 @@ class AnnAssignHandle(BaseHandle):
self._StoreWithCoerce(Value, MemberPtr)
break
def _StoreWithCoerce(self, Value: ir.Value, Ptr: ir.Value) -> None:
def _StoreWithCoerce(self, Value: ir.Value, Ptr: ir.Value, Node: ast.AST | None = None) -> None:
Gen: LlvmCodeGenerator = self.Trans.LlvmGen
if isinstance(Ptr.type, ir.PointerType):
TargetType: ir.Type = Ptr.type.pointee
if Value.type != TargetType:
if isinstance(TargetType, ir.IntType) and isinstance(Value.type, ir.IntType):
# None(空值常量)允许隐式转换为任意字段类型:指针→空指针,其余→零值。
if Gen._IsNoneValue(Value, Node):
Value = (ir.Constant(TargetType, None) if isinstance(TargetType, ir.PointerType)
else Gen._ZeroConst(TargetType))
elif isinstance(TargetType, ir.IntType) and isinstance(Value.type, ir.IntType):
if TargetType.width < Value.type.width:
Value = Gen.builder.trunc(Value, TargetType, name="trunc")
else:
Value = Gen.builder.zext(Value, TargetType, name="zext")
elif isinstance(TargetType, ir.PointerType) and isinstance(Value.type, ir.PointerType):
Allowed: bool
Reason: str
Allowed, Reason = Gen._ClassifyPtrImplicitConvert(Value.type, TargetType)
if not Allowed:
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转换已禁止({Reason}):不能把指针类型 {Value.type} 隐式转为 {TargetType};"
f"如需转换请显式 cast(如 t.CPtr(...))"
f"{Gen._DescribeValueNode(Node)}"
)
Value = Gen.builder.bitcast(Value, TargetType, name="ptr_bitcast")
elif isinstance(TargetType, ir.PointerType) and isinstance(Value.type, ir.IntType):
Value = Gen.builder.inttoptr(Value, TargetType, name="int_to_ptr")
elif isinstance(TargetType, ir.IntType) and isinstance(Value.type, ir.PointerType):
if isinstance(Value, ir.Constant) and Value.constant is None:
# null 指针常量赋给整数字段 → 0(C 空指针常量语义)
Value = ir.Constant(TargetType, 0)
else:
# 指针 → 整数:隐式转换一律禁止(与 TransPyV coerce_to_type 同口径)
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:不能把指针类型 {Value.type} 赋给整数类型 {TargetType};"
f"如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
)
# 指针 → 整数:隐式转换一律禁止(与 TransPyV coerce_to_type 同口径)
# (None 已在上方短路为零值,不会落到这里)。
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:不能把指针类型 {Value.type} 赋给整数类型 {TargetType};"
f"如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{Gen._DescribeValueNode(Node)}"
)
Gen._store(Value, Ptr)
def _HandleGenericListLiteralAssign(self, VarName: str, Node: ast.AnnAssign, Gen: LlvmCodeGenerator, list_sub: ast.Subscript) -> None:
@@ -892,29 +904,42 @@ class AnnAssignHandle(BaseHandle):
if InitValue:
try:
if InitValue.type != VarType:
if isinstance(InitValue.type, ir.IntType) and isinstance(VarType, ir.IntType):
if Gen._IsNoneValue(InitValue, Node.value):
# None(空值常量)允许隐式转换为任意变量类型:
# 指针→空指针,其余→零值(整数 0 / 浮点 0.0 / 聚合 zeroinitializer)。
InitValue = (ir.Constant(VarType, None) if isinstance(VarType, ir.PointerType)
else Gen._ZeroConst(VarType))
elif isinstance(InitValue.type, ir.IntType) and isinstance(VarType, ir.IntType):
if VarType.width > InitValue.type.width:
InitValue = Gen.builder.zext(InitValue, VarType, name=f"zext_{VarName}")
elif VarType.width < InitValue.type.width:
InitValue = Gen.builder.trunc(InitValue, VarType, name=f"trunc_{VarName}")
elif isinstance(InitValue.type, ir.PointerType) and isinstance(VarType, ir.PointerType):
Allowed: bool
Reason: str
Allowed, Reason = Gen._ClassifyPtrImplicitConvert(InitValue.type, VarType)
if not Allowed:
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转换已禁止({Reason}):不能用指针类型 {InitValue.type} 初始化 "
f"{VarName}: {VarType};如需转换请显式 cast(如 t.CPtr(...))"
f"{Gen._DescribeValueNode(Node.value)}"
)
InitValue = Gen.builder.bitcast(InitValue, VarType, name=f"cast_{VarName}")
elif isinstance(VarType, ir.PointerType) and isinstance(InitValue.type, ir.IntType):
InitValue = Gen.builder.inttoptr(InitValue, VarType, name=f"int2ptr_{VarName}")
elif isinstance(VarType, ir.IntType) and isinstance(InitValue.type, ir.PointerType):
if isinstance(InitValue, ir.Constant) and InitValue.constant is None:
# null 指针常量初始化整数变量 → 0(C 空指针常量语义)
InitValue = ir.Constant(VarType, 0)
else:
# 指针 → 整数:隐式转换一律禁止(与 TransPyV coerce_to_type 同口径)
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:不能用指针类型 {InitValue.type} 初始化整数变量 "
f"{VarName}: {VarType};如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
)
# 指针 → 整数:隐式转换一律禁止(与 TransPyV coerce_to_type 同口径)
# (None 已在上方短路为零值,不会落到这里)。
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:不能用指针类型 {InitValue.type} 初始化整数变量 "
f"{VarName}: {VarType};如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{Gen._DescribeValueNode(Node.value)}"
)
# 大端局部变量:存储时 bswap
InitValue = Gen._apply_bswap_if_big(InitValue, TypeInfo.ByteOrder, f"bswap_store_{VarName}")
Gen._store(InitValue, var)
Gen._store(InitValue, var, Node=Node.value)
except (TypeError, ValueError) as _e:
# 仅回退 llvmlite 自身的类型/值错误;FailFastError 等一律上抛(不吞)
if _config_mode == "strict":
+37 -24
View File
@@ -365,23 +365,23 @@ class AssignHandle(BaseHandle):
Loaded_differ = False
if Loaded_differ:
try:
Coerced: ir.Value | None = Gen._coerce_value(LoadedValue, TargetType)
Coerced: ir.Value | None = Gen._coerce_value(LoadedValue, TargetType, Node.value)
if Coerced is not None:
LoadedValue = Coerced
except AssertionError:
pass
Gen._store(LoadedValue, VarPtr)
Gen._store(LoadedValue, VarPtr, Node=Node.value)
elif types_differ:
try:
Coerced: ir.Value | None = Gen._coerce_value(Value, TargetType)
Coerced: ir.Value | None = Gen._coerce_value(Value, TargetType, Node.value)
if Coerced is not None:
Gen._store(Coerced, VarPtr)
Gen._store(Coerced, VarPtr, Node=Node.value)
else:
Gen._store(Value, VarPtr)
Gen._store(Value, VarPtr, Node=Node.value)
except AssertionError:
Gen._store(Value, VarPtr)
Gen._store(Value, VarPtr, Node=Node.value)
else:
Gen._store(Value, VarPtr)
Gen._store(Value, VarPtr, Node=Node.value)
if VarName not in Gen.variables:
# 检查是否为模块级全局变量
# main 函数代表模块级作用域,直接修改全局变量
@@ -391,12 +391,12 @@ class AssignHandle(BaseHandle):
IsMainScope: bool = CurrentFuncName == 'main' or CurrentFuncName.endswith('.main')
if VarName in Gen.module.globals and (VarName in Gen.global_vars or IsMainScope):
GVar: ir.GlobalVariable = Gen.module.globals[VarName]
Gen._store(Value, GVar)
Gen._store(Value, GVar, Node=Node.value)
Gen.variables[VarName] = GVar
elif isinstance(Value.type, ir.PointerType) and isinstance(Value.type.pointee, ir.PointerType):
LoadedValue: ir.Value = Gen._load(Value, name=f"Load_{VarName}")
var: ir.AllocaInstr = Gen._allocaEntry(LoadedValue.type, name=VarName)
Gen._store(LoadedValue, var)
Gen._store(LoadedValue, var, Node=Node.value)
Gen.variables[VarName] = var
else:
var: ir.AllocaInstr = Gen._allocaEntry(Value.type, name=VarName)
@@ -523,7 +523,7 @@ class AssignHandle(BaseHandle):
if DerefValue is None:
self._RaiseUnresolvedExpr("解引用写入", Node.value, Gen)
return
self._StoreWithCoerce(DerefValue, Ptr)
self._StoreWithCoerce(DerefValue, Ptr, Node.value)
elif isinstance(Target, (ast.Tuple, ast.List)):
# ast.List 与 ast.Tuple 字段表逐字相同(Nodes.py:119-120 均为
# ('elts', 'ctx')),故 `[a, b] = f()` 与 `a, b = f()` 共用本分支。
@@ -1300,13 +1300,15 @@ class AssignHandle(BaseHandle):
MemberPtr: ir.Value = Gen.builder.gep(LoadedPtr, [ir.Constant(ir.IntType(32), 0), ir.Constant(ir.IntType(32), offset)], name=AttrName)
self._StoreWithCoerce(Value, MemberPtr)
def _StoreArrElemWithCoerceBswap(self, Value: ir.Value, Ptr: ir.Value, arr_byte_order: str) -> None:
def _StoreArrElemWithCoerceBswap(self, Value: ir.Value, Ptr: ir.Value, arr_byte_order: str,
Node: ast.AST | None = None) -> None:
"""对数组元素存储应用 bswap 后再 coerce 存储"""
Gen: "Translator.LlvmGen" = self.Trans.LlvmGen
Value = Gen._apply_bswap_if_big(Value, arr_byte_order, "bswap_arr_store")
self._StoreWithCoerce(Value, Ptr)
self._StoreWithCoerce(Value, Ptr, Node)
def _HandleSubscriptStoreLlvm(self, Target: ast.Subscript, Value: ir.Value) -> None:
def _HandleSubscriptStoreLlvm(self, Target: ast.Subscript, Value: ir.Value,
Node: ast.AST | None = None) -> None:
Gen: "Translator.LlvmGen" = self.Trans.LlvmGen
# pointee-volatile 基址名后缀(与 HandlesExprAttr._HandleSubscriptLlvm 同款);
# 空后缀 ⇒ 下标 GEP 名仍是 "subscript",既有代码 IR 逐字节不变。
@@ -1382,7 +1384,7 @@ class AssignHandle(BaseHandle):
if isinstance(IndexVal.type, ir.IntType) and IndexVal.type.width < 64:
IndexVal = Gen.builder.zext(IndexVal, ir.IntType(64), name="zext_ptr_index_store")
ElemPtr: ir.Value = Gen.builder.gep(ValueVal, [IndexVal], name=f"subscript{_PVSuffix}")
self._StoreArrElemWithCoerceBswap(Value, ElemPtr, arr_byte_order)
self._StoreArrElemWithCoerceBswap(Value, ElemPtr, arr_byte_order, Node)
return
if isinstance(ValueVal.type, ir.PointerType):
pointee: ir.Type = ValueVal.type.pointee
@@ -1531,12 +1533,16 @@ class AssignHandle(BaseHandle):
f"(`rawptr[T] | t.CPtr` 或 `rawptr[rawptr[T]]`)后再赋值;"
f"或把基座显式转为 t.CPtr(无类型指针)后按 8 字节指针语义写入。{SrcInfo}")
def _StoreWithCoerce(self, Value: ir.Value, Ptr: ir.Value) -> None:
def _StoreWithCoerce(self, Value: ir.Value, Ptr: ir.Value, Node: ast.AST | None = None) -> None:
Gen: "Translator.LlvmGen" = self.Trans.LlvmGen
if isinstance(Ptr.type, ir.PointerType):
TargetType: ir.Type = Ptr.type.pointee
if Value.type != TargetType:
if isinstance(TargetType, ir.ArrayType) and isinstance(Value.type, ir.PointerType):
if Gen._IsNoneValue(Value, Node):
# None(空值常量)允许隐式转换为任意目标类型:指针→空指针,其余→零值。
Value = (ir.Constant(TargetType, None) if isinstance(TargetType, ir.PointerType)
else Gen._ZeroConst(TargetType))
elif isinstance(TargetType, ir.ArrayType) and isinstance(Value.type, ir.PointerType):
ElemType: ir.Type = TargetType.element
ElemSize: int = Gen._get_struct_size(ElemType)
for i in range(TargetType.count):
@@ -1557,18 +1563,25 @@ class AssignHandle(BaseHandle):
continue
Gen._store(SrcVal, DstElem)
return
if isinstance(TargetType, ir.PointerType) and isinstance(Value, ir.Constant) and Value.constant is None:
Value = ir.Constant(TargetType, None)
elif isinstance(TargetType, ir.PointerType) and isinstance(Value.type, ir.PointerType):
try:
Value = Gen.builder.bitcast(Value, TargetType, name="null_ptr_bitcast")
except Exception: # 回退:bitcast 失败时设为 null 常量
Value = ir.Constant(TargetType, None)
# 新增指针隐式转换规则①/②:通用指针通配放行;非通用 T*→U* 仅在
# 被指类型无损时放行,非基本/有损/层数不同 fail-fast。
Allowed: bool
Reason: str
Allowed, Reason = Gen._ClassifyPtrImplicitConvert(Value.type, TargetType)
if not Allowed:
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转换已禁止({Reason}):不能把指针类型 {Value.type} 隐式转为 {TargetType};"
f"如需转换请显式 cast(如 t.CPtr(...))"
f"{Gen._DescribeValueNode(Node)}"
)
Value = Gen.builder.bitcast(Value, TargetType, name="ptr_bitcast")
else:
Coerced: ir.Value | None = Gen._coerce_value(Value, TargetType)
Coerced: ir.Value | None = Gen._coerce_value(Value, TargetType, Node)
if Coerced is not None:
Value = Coerced
Gen._store(Value, Ptr)
Gen._store(Value, Ptr, Node=Node)
def _get_int_ptr(self, target: ast.expr) -> ir.Value | None:
+7
View File
@@ -179,6 +179,9 @@ class ExprHandle(BaseHandle):
elem_val: Any = self.HandleExprLlvm(elem, VarType=elem_type)
if elem_val is None:
continue
# None(空值常量)允许隐式转换为任意元素类型 ⇒ 物化为目标元素类型零值。
if Gen._IsNoneValue(elem_val, elem):
elem_val = Gen._ZeroConst(elem_llvm_type)
byte_offset: int = i
if elem_type == 'i32':
@@ -209,10 +212,12 @@ class ExprHandle(BaseHandle):
elem_val = Gen.builder.trunc(elem_val, ir.IntType(8), name=f"trunc_elem_{i}")
elif isinstance(elem_val.type, ir.PointerType):
# 指针 → 整数:隐式转换一律禁止(元素值是指针、元素类型是整数)
# (None 已在上方短路为零值,不会落到这里)。
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:数组字面量第 {i} 个元素是指针类型 {elem_val.type},"
f"但目标元素类型是 i8;如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{Gen._DescribeValueNode(elem)}"
)
elif elem_type == 'i32':
if isinstance(elem_val.type, ir.IntType):
@@ -222,10 +227,12 @@ class ExprHandle(BaseHandle):
elem_val = Gen.builder.trunc(elem_val, ir.IntType(32), name=f"trunc_elem_{i}")
elif isinstance(elem_val.type, ir.PointerType):
# 指针 → 整数:隐式转换一律禁止(ptrtoint 到 i32 还会截断地址)
# (None 已在上方短路为零值,不会落到这里)。
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:数组字面量第 {i} 个元素是指针类型 {elem_val.type},"
f"但目标元素类型是 i32;如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{Gen._DescribeValueNode(elem)}"
)
elif elem_type == 'i16':
if isinstance(elem_val.type, ir.IntType):
+10 -2
View File
@@ -182,7 +182,10 @@ class ExprLambdaHandle(BaseHandle):
if not ThenEndBB.is_terminated:
Gen.builder.position_at_end(ThenEndBB)
if BodyVal.type != result_type:
if isinstance(BodyVal.type, ir.IntType) and isinstance(result_type, ir.IntType):
if Gen._IsNoneValue(BodyVal, Node.body):
# None(空值常量)允许隐式转换为任意结果类型 ⇒ 物化为零值。
BodyVal = Gen._ZeroConst(result_type)
elif isinstance(BodyVal.type, ir.IntType) and isinstance(result_type, ir.IntType):
if BodyVal.type.width < result_type.width:
BodyVal = Gen.builder.zext(BodyVal, result_type, name="ifexp_body_zext")
else:
@@ -205,13 +208,17 @@ class ExprLambdaHandle(BaseHandle):
raise FailFastError(
f"指针隐式转整数已禁止:条件表达式两分支类型不一致({BodyVal.type} / {result_type});"
f"如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{Gen._DescribeValueNode(Node.body)}"
)
Gen.builder.branch(MergeBB)
# 在 ElseBB 末尾添加类型转换(如果需要)和分支
if not ElseEndBB.is_terminated:
Gen.builder.position_at_end(ElseEndBB)
if OrelseVal.type != result_type:
if isinstance(OrelseVal.type, ir.IntType) and isinstance(result_type, ir.IntType):
if Gen._IsNoneValue(OrelseVal, Node.orelse):
# None(空值常量)允许隐式转换为任意结果类型 ⇒ 物化为零值。
OrelseVal = Gen._ZeroConst(result_type)
elif isinstance(OrelseVal.type, ir.IntType) and isinstance(result_type, ir.IntType):
if OrelseVal.type.width < result_type.width:
OrelseVal = Gen.builder.zext(OrelseVal, result_type, name="ifexp_orelse_zext")
else:
@@ -233,6 +240,7 @@ class ExprLambdaHandle(BaseHandle):
raise FailFastError(
f"指针隐式转整数已禁止:条件表达式两分支类型不一致({OrelseVal.type} / {result_type});"
f"如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{Gen._DescribeValueNode(Node.orelse)}"
)
Gen.builder.branch(MergeBB)
Gen.builder.position_at_start(MergeBB)
+5 -5
View File
@@ -78,7 +78,7 @@ class ReturnHandle(BaseHandle):
Gen._closure_return_types[short_name] = (fn_return_type, nonlocal_types)
# 返回闭包结构体指针(bitcast 到 i8*)
closure_ret: ir.Value = Gen.builder.bitcast(closure_ptr, i8_ptr_type, name="closure_ret")
Gen.emit_return(closure_ret)
Gen.emit_return(closure_ret, Node.value)
return
active_finally: tuple[ir.Block, ir.AllocaInstr, ir.AllocaInstr | None] | None = self._find_active_finally(Gen)
if getattr(self.Trans, 'CurrentMultiRetTypes', None) and self.Trans.CurrentMultiRetTypes and Node.value:
@@ -133,7 +133,7 @@ class ReturnHandle(BaseHandle):
Gen._store(ir.Constant(ir.IntType(32), 1), pending_ret_flag)
Gen.builder.branch(FinallyBB)
else:
Gen.emit_return(result)
Gen.emit_return(result, Node.value)
return
if Node.value:
# 本函数返回类型即"使用处宽度":先取回返回类型再翻译值,供 CDefine 常量
@@ -159,7 +159,7 @@ class ReturnHandle(BaseHandle):
Gen._store(ir.Constant(ir.IntType(32), 1), pending_ret_flag)
Gen.builder.branch(FinallyBB)
else:
Gen.emit_return(zero_val)
Gen.emit_return(zero_val, Node.value)
return
if getattr(self.Trans, 'CurrentMultiRetTypes', None) and self.Trans.CurrentMultiRetTypes:
struct_type: ir.Type = Gen.func.ftype.return_type
@@ -197,7 +197,7 @@ class ReturnHandle(BaseHandle):
Gen._store(ir.Constant(ir.IntType(32), 1), pending_ret_flag)
Gen.builder.branch(FinallyBB)
else:
Gen.emit_return(result)
Gen.emit_return(result, Node.value)
return
if active_finally:
FinallyBB: ir.Block = active_finally[0]
@@ -208,7 +208,7 @@ class ReturnHandle(BaseHandle):
Gen._store(ir.Constant(ir.IntType(32), 1), pending_ret_flag)
Gen.builder.branch(FinallyBB)
else:
Gen.emit_return(Val)
Gen.emit_return(Val, Node.value)
else:
if active_finally:
FinallyBB: ir.Block = active_finally[0]
+33
View File
@@ -960,6 +960,39 @@ class BaseGenMixin:
info += f"\n → {source_line}"
return info
def _DescribeValueNode(self, Node: Any = None) -> str:
"""把「真正被转换的值节点」渲染成一段诊断文本,供隐式转换禁令引用。
报文风格照本仓既有「表达式文本 + 位置 + 源码片段」:
表达式 '<unparse>'(<节点形态>,位置 文件:行:列),源码: <该行>
关键:**列号按有符号打印**——未知(无属性 / <0)时打印 '-',绝不把 col_offset
的 -1 当无符号打成 4294967295(此前 TPV 侧诊断错位的根因)。取不到节点时
回退当前语句上下文(_get_node_info 风格),不再固定打印外层 Assign。
"""
if Node is None:
return self._get_node_info()
try:
Expr: str = ast.unparse(Node)
except Exception:
Expr = getattr(Node, 'attr', '') or getattr(Node, 'id', '') or '<expr>'
RawLine: int = getattr(Node, 'lineno', 0) or 0
Col: Any = getattr(Node, 'col_offset', -1)
Line: int = RawLine or self._current_lineno or 0
Loc: str = self._current_source_file or ''
if Line > 0:
Loc += f":{Line}"
Loc += f":{Col}" if isinstance(Col, int) and Col >= 0 else ":-"
Code: str = ''
if Line > 0:
try:
Code = (self._get_source_line(Line) or '').strip()
except Exception:
Code = ''
Text: str = f";出错表达式: 表达式 '{Expr}'({type(Node).__name__} 形态,位置 {Loc})"
if Code:
Text += f",源码: {Code}"
return Text
# ========== const / volatile:判定、登记与报文(两侧同口径) ==========
# 语义唯一来源见 Debugger/type-descriptors.md。TPC 与 TransPyV 必须**同判据、
# 同报文**;任何未消除的差异都列在该文档的「已知差异」一节。
+17 -1
View File
@@ -294,7 +294,7 @@ class ExprGenMixin:
return self.builder.sitofp(val, ftype, name="int2float")
return val
def emit_return(self, value: ir.Value | None = None) -> None:
def emit_return(self, value: ir.Value | None = None, Node: Any = None) -> None:
if not self.builder or self.builder.block.is_terminated:
return
if value is not None:
@@ -327,8 +327,22 @@ class ExprGenMixin:
else:
if self.func and hasattr(self.func, 'ftype'):
ret_type: ir.Type = self.func.ftype.return_type
# None(空值常量)允许隐式转换为任意返回类型:指针目标→空指针,
# 其余目标→目标类型零值(整数 0 / 浮点 0.0 / 聚合 zeroinitializer)。
if not isinstance(ret_type, ir.VoidType) and self._IsNoneValue(value, Node):
value = self._ZeroConst(ret_type)
if ret_type != value.type:
if isinstance(ret_type, ir.PointerType) and isinstance(value.type, ir.PointerType):
Allowed: bool
Reason: str
Allowed, Reason = self._ClassifyPtrImplicitConvert(value.type, ret_type)
if not Allowed:
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转换已禁止({Reason}):函数 {self.func.name} 声明返回 {ret_type},"
f"但 return 的是指针类型 {value.type};如需转换请显式 cast(如 t.CPtr(...))"
f"{self._DescribeValueNode(Node)}"
)
value = self.builder.bitcast(value, ret_type, name="ret_cast")
elif isinstance(value.type, ir.PointerType) and not isinstance(ret_type, ir.PointerType):
if isinstance(ret_type, ir.IntType):
@@ -338,6 +352,7 @@ class ExprGenMixin:
raise FailFastError(
f"指针隐式转整数已禁止:函数 {self.func.name} 声明返回 {ret_type},"
f"但 return 的是指针类型 {value.type};如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{self._DescribeValueNode(Node)}"
)
elif isinstance(value.type.pointee, ret_type.__class__) and not isinstance(value.type.pointee, ir.IntType):
value = self._load(value, name="ret_Load")
@@ -350,6 +365,7 @@ class ExprGenMixin:
raise FailFastError(
f"指针隐式转整数已禁止:函数 {self.func.name} 声明返回 {ret_type},"
f"但 return 的是指针类型 {Loaded.type};如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{self._DescribeValueNode(Node)}"
)
elif isinstance(ret_type, ir.IntType) and isinstance(value.type, ir.IntType):
if value.type.width < ret_type.width:
+10
View File
@@ -165,6 +165,16 @@ class FuncGenMixin:
else:
adjusted.append(elem_ptr)
else:
# 新增指针隐式转换规则①/②:通用指针通配放行;非通用
# T*→U* 仅在无损时放行,非基本/有损/层数不同 fail-fast。
Allowed: bool
Reason: str
Allowed, Reason = self._ClassifyPtrImplicitConvert(arg.type, param.type)
if not Allowed:
raise FailFastError(
f"指针隐式转换已禁止({Reason}):调用 {func.name} 第 {i} 个实参"
f" {arg.type} 不能隐式转为形参 {param.type};如需转换请显式 cast"
f"(如 t.CPtr(...)){self._DescribeValueNode()}")
adjusted.append(self.builder.bitcast(arg, param.type, name=f"cast_arg_{i}"))
else:
if isinstance(arg.type, ir.PointerType) and isinstance(param.type, (ir.IdentifiedStructType, ir.LiteralStructType, ir.ArrayType)):
+150 -10
View File
@@ -1,4 +1,5 @@
from __future__ import annotations
from typing import Any
import llvmlite.ir as ir
@@ -137,6 +138,117 @@ class MemoryOpsMixin:
return ir.Constant(typ, None)
return ir.Constant(typ, 0)
# ============================================================
# 隐式转换公共判据:None(空值常量)/ 指针层级 / 基本类型 / 无损
#
# 语义唯一来源与 TransPyV 对齐(ExprGenBase.IsNoneValue、coerce_to_type)。
# 收敛于此,供赋值 / 返回 / 实参 / 数组元素等所有 coerce 入口复用,避免
# 各处各自演化出相反结论。
# ============================================================
@staticmethod
def _IsNoneValue(val: Any, node: Any = None) -> bool:
"""None 判据:True=空值常量(None),False=否。
None 无语义类型信息,**允许隐式转换为任意类型**:指针目标→空指针,
其余目标→目标类型零值(整数 0 / 浮点 0.0 / 聚合 zeroinitializer)。故
「指针↔整数」「指针↔非指针值」「指针↔指针(非基本/有损/层数不同)」
这类针对**真实类型值**的隐式转换禁令都要对 None 短路放行。
两条判据(任一命中):
① AST 层:node 是 Constant 且 value is None;
② 值层(主判据):已生成好的 **空指针常量**(ir.Constant 且 constant is None
且类型为指针)。整数常量 0 的类型是整数,不会误命中。
"""
if node is not None and type(node).__name__ == 'Constant' and getattr(node, 'value', 1) is None:
return True
return (isinstance(val, ir.Constant) and val.constant is None
and isinstance(val.type, ir.PointerType))
@staticmethod
def _PtrLevel(typ: Any) -> int:
"""指针层级:PointerType 嵌套层数(0=非指针)。"""
n: int = 0
while isinstance(typ, ir.PointerType):
n += 1
typ = typ.pointee
return n
@staticmethod
def _IsGenericPtrType(typ: Any) -> bool:
"""通用指针(t.CPtr / void*):指针链末端为 i8(i8* / i8** / ... 均算)。"""
if not isinstance(typ, ir.PointerType):
return False
p: Any = typ.pointee
while isinstance(p, ir.PointerType):
p = p.pointee
return isinstance(p, ir.IntType) and p.width == 8
@staticmethod
def _IsBasicScalarType(typ: Any) -> bool:
"""基本类型 = 整数/浮点(含布尔 i1);结构体/类/联合/数组/指针一律非基本。"""
return isinstance(typ, (ir.IntType, ir.FloatType, ir.DoubleType))
@staticmethod
def _ScalarConvertLossless(Src: Any, Dst: Any) -> bool:
"""既有整数/浮点规则下的「无损」判定:同族且目标位宽 >= 源位宽。
整数→整数 / 浮点→浮点:按位宽(宽化无损、同宽无损、窄化有损);
整数↔浮点:跨类别,一律视为非无损(有损)。
"""
def _FloatBits(x: Any) -> int:
if isinstance(x, ir.DoubleType):
return 64
if isinstance(x, ir.FloatType):
return 32
return 0
if isinstance(Src, ir.IntType) and isinstance(Dst, ir.IntType):
return Dst.width >= Src.width
sb: int = _FloatBits(Src)
db: int = _FloatBits(Dst)
if sb and db:
return db >= sb
return False
@staticmethod
def _ClassifyPtrImplicitConvert(Src: Any, Dst: Any) -> tuple[bool, str]:
"""隐式转换判据:返回 (Allowed, Reason)。
规则(用户决定,与 TransPyV 同口径):
① 通用指针(t.CPtr/i8*)为通配:与任何 T* **双向**互转放行(须同级)。
② T* → U*(皆非通用):允许 ⇔ 被指类型 T → U 无损;非基本类型
(结构体/类/联合)一律不允许 ⇒ Foo*→Bar* 禁止。
③ 非基本类型 T 与 T* 之间的隐式转换(Foo ↔ Foo*)一律禁止(层数不同)。
④ 完全相同类型不算转换 ⇒ 放行。
Reason 供报文说明「哪一级不允许 + 原因(非基本类型/有损/层数不同)」。
"""
if Src == Dst:
return (True, '同类型(不构成转换)')
s_ptr: bool = isinstance(Src, ir.PointerType)
d_ptr: bool = isinstance(Dst, ir.PointerType)
if s_ptr and d_ptr:
if MemoryOpsMixin._PtrLevel(Src) != MemoryOpsMixin._PtrLevel(Dst):
return (False, '层数不同')
if MemoryOpsMixin._IsGenericPtrType(Src) or MemoryOpsMixin._IsGenericPtrType(Dst):
return (True, '通用指针通配')
sp: Any = Src.pointee
dp: Any = Dst.pointee
if sp == dp:
return (True, '被指类型相同')
if MemoryOpsMixin._IsBasicScalarType(sp) and MemoryOpsMixin._IsBasicScalarType(dp):
if MemoryOpsMixin._ScalarConvertLossless(sp, dp):
return (True, '被指类型无损')
return (False, '被指类型有损')
return (False, '非基本类型(结构体/类/联合)')
if not s_ptr and not d_ptr:
return (True, '非指针间(本判据不裁决)')
# 一方指针、一方非指针:基本标量走既有「指针隐式转整数」禁令;非基本
# 类型即规则③(T ↔ T* 隐式互转)。
other: Any = Dst if s_ptr else Src
if MemoryOpsMixin._IsBasicScalarType(other):
return (False, '指针与标量(须显式 cast)')
return (False, '非基本类型(结构体/类/联合)')
def _alloca(self, typ: ir.Type, name: str = '', align: int | None = None, size: ir.Value | None = None) -> ir.AllocaInstr:
if align is None:
align = self._get_align(typ)
@@ -244,9 +356,16 @@ class MemoryOpsMixin:
return self._load(GVar, name=name)
return None
def _coerce_value(self, val: ir.Value, target_type: ir.Type) -> ir.Value | None:
def _coerce_value(self, val: ir.Value, target_type: ir.Type, Node: Any = None) -> ir.Value | None:
if val.type == target_type:
return val
# None(空值常量)允许隐式转换为任意类型:指针目标→空指针,其余目标→
# 目标类型零值(整数 0 / 浮点 0.0 / 聚合 zeroinitializer)。收敛到
# _IsNoneValue 单一判据,任何「指针↔整数 / 指针↔非指针值」禁令都不拦 None。
if self._IsNoneValue(val, Node):
if isinstance(target_type, ir.PointerType):
return ir.Constant(target_type, None)
return self._ZeroConst(target_type)
if isinstance(target_type, ir.IdentifiedStructType) and target_type.name in self.typedef_int_widths:
if isinstance(val.type, ir.IntType):
return val
@@ -260,26 +379,33 @@ class MemoryOpsMixin:
else:
return self.builder.sext(val, target_type, name="sext_store")
if isinstance(target_type, ir.PointerType) and isinstance(val.type, ir.PointerType):
if isinstance(val, ir.Constant) and val.constant is None:
return ir.Constant(target_type, None)
# 新增指针隐式转换规则①/②:通用指针通配双向放行;非通用 T*→U* 仅在
# 被指类型无损时放行,非基本类型/有损/层数不同一律 fail-fast。
Allowed: bool
Reason: str
Allowed, Reason = self._ClassifyPtrImplicitConvert(val.type, target_type)
if not Allowed:
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转换已禁止({Reason}):不能把指针类型 {val.type} 隐式转为 {target_type};"
f"如需转换请显式 cast(如 t.CPtr(...))"
f"{self._DescribeValueNode(Node)}"
)
return self.builder.bitcast(val, target_type, name="ptr_bitcast_store")
if isinstance(target_type, ir.PointerType) and isinstance(val.type, ir.IntType):
if val.type.width < 64:
val = self.builder.zext(val, ir.IntType(64), name="zext_ptr_store")
return self.builder.inttoptr(val, target_type, name="int_to_ptr_store")
if isinstance(target_type, ir.IntType) and isinstance(val.type, ir.PointerType):
if isinstance(val, ir.Constant) and val.constant is None:
# null 指针常量赋给 int 字段 → 0。绝不能对 null 做 load
# (此前 pointee.width == target.width 时直接 load null,
# 生成 `load i32, i32* null` → ASAN access-violation)
return ir.Constant(target_type, 0)
# 指针 → 整数:隐式转换一律禁止(与 TransPyV coerce_to_type 同口径)。
# 原先三条路径都是静默的:同宽直接 load 指针所指物的位模式、≤8 位 load
# 首字节、其余 ptrtoint 截断地址——把"类型写错"藏到运行期。
# None(空指针常量)已在上方短路为目标零值,不会落到这里。
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转整数已禁止:不能把指针类型 {val.type} 赋给整数类型 {target_type};"
f"如需转换请显式 cast(如 t.CUInt64T(p) / t.CInt(p))"
f"{self._DescribeValueNode(Node)}"
)
if isinstance(target_type, ir.FloatType) and isinstance(val.type, ir.DoubleType):
return self.builder.fptrunc(val, target_type, name="fptrunc_store")
@@ -304,6 +430,20 @@ class MemoryOpsMixin:
if trimmed and len(trimmed) == val.type.count:
return ir.Constant(target_type, trimmed[:target_type.count])
if isinstance(val.type, ir.PointerType) and not isinstance(target_type, ir.PointerType):
# 规则③:非基本类型 T 与 T* 之间的隐式转换(Foo ↔ Foo*)一律禁止——
# 指针→非指针值的隐式 load(结构体/数组按值回读)属类型信息缺失,
# 须显式解引用(c.Deref(p) / *p)或显式强转。基本标量目标已在上方
# 「指针隐式转整数」分支 fail-fast,此处只兜非基本类型。
Allowed3: bool
Reason3: str
Allowed3, Reason3 = self._ClassifyPtrImplicitConvert(val.type, target_type)
if not Allowed3:
from lib.core.LLVMCG.FuncGen import FailFastError
raise FailFastError(
f"指针隐式转换已禁止({Reason3}):不能把指针类型 {val.type} 隐式转为 {target_type};"
f"如需转换请显式 cast(解引用 c.Deref(p) / *p,或显式强转)"
f"{self._DescribeValueNode(Node)}"
)
if val.type.pointee == target_type:
return self.builder.load(val, name="Load_struct_for_store")
if isinstance(val.type.pointee, ir.IdentifiedStructType) and isinstance(target_type, ir.IdentifiedStructType):
@@ -314,7 +454,7 @@ class MemoryOpsMixin:
return None
def _store(self, val: ir.Value, ptr: ir.Value, align: int | None = None,
volatile: bool | None = None) -> ir.StoreInstr:
volatile: bool | None = None, Node: Any = None) -> ir.StoreInstr:
if align is None:
if isinstance(ptr.type, ir.PointerType):
align = self._get_align(ptr.type.pointee)
@@ -350,7 +490,7 @@ class MemoryOpsMixin:
# 防止 `store i32* null, i32*` 类型错乱 IR)
val = ir.Constant(target_type, 0)
else:
coerced: ir.Value | None = self._coerce_value(val, target_type)
coerced: ir.Value | None = self._coerce_value(val, target_type, Node)
if coerced is not None:
val = coerced
else: