Files
TransPyC/lib/core/LLVMCG/MemoryOps.py
T
TermiNexus 2922af79f1 fix(LLVMCG): 禁止对 null 指针常量做 load 转 int 字段
_coerce_value 中 target 为 IntType 且 val 为 PointerType 时,若
val 是 null 常量且 pointee.width == target.width,此前直接
builder.load(null) 生成 load i32, i32* null,运行时 ASAN
access-violation。根因场景:Generator.__init__ 里
self.ClassMetaKinds = None(字段注解 t.CInt | t.CPtr,联合类型
退化为 i32 槽位,None 解析为 i32* null)。修复:null 常量赋 int
字段直接返回 0;_store 的 null 分支同步补 IntType 处理。
2026-08-18 12:50:31 +08:00

295 lines
17 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
from __future__ import annotations
import llvmlite.ir as ir
class MemoryOpsMixin:
def _register_local_heap_ptr(self, val: ir.Value, ClassName: str | None = None, VarName: str | None = None) -> None:
if isinstance(val.type, ir.PointerType):
pointee: ir.Type = val.type.pointee
if isinstance(pointee, (ir.LiteralStructType, ir.IdentifiedStructType)):
self._local_heap_ptrs.append((val, 'struct', ClassName))
elif isinstance(pointee, ir.IntType) and pointee.width == 8:
self._local_heap_ptrs.append((val, 'i8', ClassName))
if VarName:
self._var_to_heap_ptr[VarName] = val
def _unregister_local_heap_ptr(self, val: ir.Value) -> None:
self._local_heap_ptrs = [(v, t, cn) for v, t, cn in self._local_heap_ptrs if v is not val]
def _emit_local_heap_frees(self) -> None:
if not self._local_heap_ptrs or not self.builder:
return
deleted_objects: set[ir.Value] = set()
for ptr, ptr_type, ClassName in self._local_heap_ptrs:
if ptr_type == 'i8':
Loaded_ptr: ir.Value = self.builder.load(ptr, name="load_ptr")
null_ptr: ir.Constant = ir.Constant(ir.PointerType(ir.IntType(8)), None)
is_null: ir.Value = self.builder.icmp_unsigned('==', Loaded_ptr, null_ptr, name="is_null")
not_null_block: ir.Block = self.builder.append_basic_block(name="not_null")
next_block: ir.Block = self.builder.append_basic_block(name="next")
self.builder.cbranch(is_null, next_block, not_null_block)
self.builder.position_at_end(not_null_block)
if ClassName and self._has_function(f'{ClassName}.__del__'):
cast_ptr: ir.Value
if isinstance(ptr.type, ir.PointerType) and isinstance(ptr.type.pointee, ir.IntType) and ptr.type.pointee.width == 8:
cast_ptr = self.builder.bitcast(ptr, ir.PointerType(self.structs[ClassName]), name=f"cast_{ClassName}")
else:
cast_ptr = ptr
if cast_ptr not in deleted_objects:
self.builder.call(self._get_function(f'{ClassName}.__del__'), [cast_ptr], name=f"call_{ClassName}.__del__")
deleted_objects.add(cast_ptr)
raw: ir.Value
if ptr_type == 'struct':
raw = self.builder.bitcast(ptr, ir.PointerType(ir.IntType(8)), name="local_free_cast")
elif ptr_type == 'i8':
raw = ptr
else:
continue
free_type: ir.FunctionType = ir.FunctionType(ir.VoidType(), [ir.PointerType(ir.IntType(8))])
free_func: ir.Function = self._get_or_declare_func('free', free_type)
self.builder.call(free_func, [raw])
if ptr_type == 'i8':
self.builder.branch(next_block)
self.builder.position_at_end(next_block)
self._local_heap_ptrs = []
def _RegisterTempPtr(self, val: ir.Value) -> None:
if isinstance(val.type, ir.PointerType):
pointee: ir.Type = val.type.pointee
if isinstance(pointee, (ir.LiteralStructType, ir.IdentifiedStructType)):
for CN, ST in self.structs.items():
if pointee == ST:
self._temp_struct_ptrs.append(val)
return
elif isinstance(pointee, ir.IntType) and pointee.width == 8:
self._temp_struct_ptrs.append(val)
def _UnregisterTempPtr(self, val: ir.Value) -> None:
self._temp_struct_ptrs = [t for t in self._temp_struct_ptrs if t is not val]
def _EmitTempFrees(self) -> None:
if not self._temp_struct_ptrs or not self.builder:
return
for ptr in self._temp_struct_ptrs:
raw: ir.Value
if isinstance(ptr.type, ir.PointerType) and isinstance(ptr.type.pointee, (ir.LiteralStructType, ir.IdentifiedStructType)):
raw = self.builder.bitcast(ptr, ir.PointerType(ir.IntType(8)), name="free_cast")
elif isinstance(ptr.type, ir.PointerType) and isinstance(ptr.type.pointee, ir.IntType) and ptr.type.pointee.width == 8:
raw = ptr
else:
continue
free_type: ir.FunctionType = ir.FunctionType(ir.VoidType(), [ir.PointerType(ir.IntType(8))])
free_func: ir.Function = self._get_or_declare_func('free', free_type)
self.builder.call(free_func, [raw])
self._temp_struct_ptrs = []
def _ZeroConst(self, typ: ir.Type) -> ir.Constant:
if isinstance(typ, ir.PointerType):
return ir.Constant(typ, None)
if isinstance(typ, (ir.LiteralStructType, ir.IdentifiedStructType)):
return ir.Constant(typ, None)
if isinstance(typ, ir.ArrayType):
return ir.Constant(typ, None)
return ir.Constant(typ, 0)
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)
instr: ir.AllocaInstr
if size is not None:
instr = self.builder.alloca(typ, size=size, name=name)
else:
instr = self.builder.alloca(typ, name=name)
if align:
instr.align = align
return instr
def _allocaEntry(self, typ: ir.Type, name: str = '', align: int | None = None) -> ir.AllocaInstr:
if align is None:
align = self._get_align(typ)
saved_block: ir.Block = self.builder.block
entry_block: ir.Block = self.func.entry_basic_block
instr: ir.AllocaInstr
if entry_block.instructions:
first_instr: ir.Instruction = entry_block.instructions[0]
self.builder.position_before(first_instr)
instr = self.builder.alloca(typ, name=name)
if align:
instr.align = align
self.builder.position_at_end(saved_block)
else:
self.builder.position_at_start(entry_block)
instr = self.builder.alloca(typ, name=name)
if align:
instr.align = align
self.builder.position_at_end(saved_block)
return instr
def _load(self, ptr: ir.Value, name: str = '', align: int | None = None) -> ir.LoadInstr:
if align is None:
if isinstance(ptr.type, ir.PointerType):
align = self._get_align(ptr.type.pointee)
else:
align = 0
if isinstance(ptr.type, ir.PointerType):
pointee_type: ir.Type = ptr.type.pointee
if isinstance(pointee_type, ir.IdentifiedStructType) and pointee_type.name in self.typedef_int_widths:
width: int = self.typedef_int_widths[pointee_type.name]
if width > 0:
actual_type: ir.IntType = ir.IntType(width)
bitcast_ptr: ir.Value = self.builder.bitcast(ptr, ir.PointerType(actual_type), name="typedef_bitcast_Load")
return self.builder.load(bitcast_ptr, name=name, align=align)
return self.builder.load(ptr, name=name, align=align)
def _Load_var(self, name: str) -> ir.Value | None:
if name in self._direct_values:
return self._direct_values[name]
if name in self.variables and self.variables[name] is not None:
VarPtr: ir.Value = self.variables[name]
if isinstance(VarPtr, ir.GlobalVariable) and isinstance(VarPtr.type, ir.PointerType) and isinstance(VarPtr.type.pointee, (ir.IdentifiedStructType, ir.LiteralStructType, ir.ArrayType)):
return VarPtr
if isinstance(VarPtr.type, ir.PointerType) and isinstance(VarPtr.type.pointee, (ir.IdentifiedStructType, ir.LiteralStructType, ir.ArrayType)):
return VarPtr
return self._load(VarPtr, name=name)
if name in self.global_vars and name in self.module.globals:
GVar: ir.GlobalVariable = self.module.globals[name]
self.variables[name] = GVar
if isinstance(GVar, ir.Function):
return GVar
if isinstance(GVar.type, ir.PointerType) and isinstance(GVar.type.pointee, (ir.IdentifiedStructType, ir.LiteralStructType, ir.ArrayType)):
return GVar
return self._load(GVar, name=name)
if name in self.module.globals:
GVar = self.module.globals[name]
self.variables[name] = GVar
if isinstance(GVar, ir.Function):
return GVar
if isinstance(GVar.type, ir.PointerType) and isinstance(GVar.type.pointee, (ir.IdentifiedStructType, ir.LiteralStructType, ir.ArrayType)):
return GVar
return self._load(GVar, name=name)
return None
def _coerce_value(self, val: ir.Value, target_type: ir.Type) -> ir.Value | None:
if val.type == target_type:
return val
if isinstance(target_type, ir.IdentifiedStructType) and target_type.name in self.typedef_int_widths:
if isinstance(val.type, ir.IntType):
return val
if isinstance(target_type, ir.IntType) and isinstance(val.type, ir.IntType):
if target_type.width < val.type.width:
return self.builder.trunc(val, target_type, name="trunc_store")
elif val.type.width == 1:
# i1 是布尔值(来自 icmp/fcmp/__contains__ 归一化),无符号语义,
# 必须用 zext 而非 sext,否则 i1 的 1 会被符号扩展为 -1
return self.builder.zext(val, target_type, name="zext_bool_store")
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)
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)
if isinstance(val.type.pointee, ir.IntType) and val.type.pointee.width == target_type.width:
return self.builder.load(val, name="load_ptr_as_int_store")
if target_type.width <= 8 and isinstance(val.type.pointee, ir.IntType):
Loaded: ir.Value = self.builder.load(val, name="load_ptr_byte_store")
if Loaded.type != target_type:
if isinstance(Loaded.type, ir.IntType) and isinstance(target_type, ir.IntType):
if target_type.width < Loaded.type.width:
return self.builder.trunc(Loaded, target_type, name="trunc_ptr_byte_store")
return self.builder.zext(Loaded, target_type, name="zext_ptr_byte_store")
return Loaded
return self.builder.ptrtoint(val, target_type, name="ptr_to_int_store")
if isinstance(target_type, ir.FloatType) and isinstance(val.type, ir.DoubleType):
return self.builder.fptrunc(val, target_type, name="fptrunc_store")
if isinstance(target_type, ir.DoubleType) and isinstance(val.type, ir.FloatType):
return self.builder.fpext(val, target_type, name="fpext_store")
if isinstance(target_type, ir.IntType) and isinstance(val.type, (ir.FloatType, ir.DoubleType)):
return self.builder.fptosi(val, target_type, name="fptosi_store")
if isinstance(target_type, (ir.FloatType, ir.DoubleType)) and isinstance(val.type, ir.IntType):
return self.builder.sitofp(val, target_type, name="sitofp_store")
if isinstance(target_type, ir.ArrayType) and isinstance(val.type, ir.ArrayType):
if target_type.count == val.type.count and target_type.element == val.type.element:
return val
if (isinstance(target_type.element, ir.IntType) and target_type.element.width == 8
and isinstance(val.type.element, ir.IntType) and val.type.element.width == 8):
if target_type.count > val.type.count:
padded: list[ir.Constant] = list(val.constant) if hasattr(val, 'constant') else []
if padded and len(padded) == val.type.count:
padded.extend([ir.Constant(ir.IntType(8), 0)] * (target_type.count - val.type.count))
return ir.Constant(target_type, padded)
elif target_type.count < val.type.count:
trimmed: list[ir.Constant] = list(val.constant) if hasattr(val, 'constant') else []
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):
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):
if val.type.pointee.name == target_type.name:
Loaded: ir.Value = self.builder.load(val, name="Load_struct_for_store")
if Loaded.type == target_type:
return Loaded
return None
def _store(self, val: ir.Value, ptr: ir.Value, align: int | None = None) -> ir.StoreInstr:
if align is None:
if isinstance(ptr.type, ir.PointerType):
align = self._get_align(ptr.type.pointee)
else:
align = 0
if isinstance(ptr.type, ir.PointerType):
target_type: ir.Type = ptr.type.pointee
if isinstance(target_type, ir.IdentifiedStructType) and target_type.name in self.typedef_int_widths:
if isinstance(val.type, ir.IntType):
width: int = self.typedef_int_widths[target_type.name]
if width > 0:
actual_type: ir.IntType = ir.IntType(width)
bitcast_ptr: ir.Value = self.builder.bitcast(ptr, ir.PointerType(actual_type), name="typedef_bitcast")
if val.type.width != actual_type.width:
if val.type.width > actual_type.width:
val = self.builder.trunc(val, actual_type, name="trunc_typedef")
else:
val = self.builder.zext(val, actual_type, name="zext_typedef")
return self.builder.store(val, bitcast_ptr, align=align)
if val.type != target_type:
if isinstance(val, ir.Constant) and val.constant is None:
if isinstance(target_type, ir.PointerType):
val = ir.Constant(target_type, None)
elif isinstance(target_type, ir.IdentifiedStructType):
val = ir.Constant(ir.PointerType(target_type), None)
ptr = self.builder.bitcast(ptr, ir.PointerType(ir.PointerType(target_type)), name="null_struct_ptr_cast")
elif isinstance(target_type, ir.IntType):
# null 指针常量赋给 int 字段 → 0(与 _coerce_value 一致,
# 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)
if coerced is not None:
val = coerced
else:
src_info: str = self._get_node_info()
raise TypeError(f"cannot store {val.type} to {ptr.type}: mismatching types{src_info}")
return self.builder.store(val, ptr, align=align)
def _get_var_ptr(self, name: str) -> ir.Value | None:
if name in self._reg_values:
val: ir.Value = self._reg_values[name]
var: ir.AllocaInstr = self._alloca(val.type, name=name)
self._store(val, var)
self.variables[name] = var
del self._reg_values[name]
return var
if name in self.variables and self.variables[name] is not None:
return self.variables[name]
return None