787 lines
35 KiB
Plaintext
787 lines
35 KiB
Plaintext
import t, c
|
||
from stdint import *
|
||
import string
|
||
import stdio
|
||
import stdlib
|
||
import viperlib
|
||
import memhub
|
||
from linkedlist import GSList, GSListNode
|
||
from .__types import LLVMType, TypePrint, Array, PrintStructDefinition, get_struct_name, Struct, ParamNode
|
||
from .__function import Function, FunctionPrint, _ll_name_needs_quote, Param, param_get_next, param_get_ty
|
||
|
||
|
||
# ============================================================
|
||
# GlobalVariable: LLVM IR 全局变量
|
||
#
|
||
# 持有名称 + 类型 + 初始化值 + 常量标志 + 链接类型。
|
||
# 继承 GSListNode[GlobalVariable] 获取强类型 Next。
|
||
# ============================================================
|
||
@t.NoVTable
|
||
class GlobalVariable(GSListNode[GlobalVariable]):
|
||
Name: str # 全局变量名（如 ".str" / "my_global"）
|
||
Ty: rowptr[LLVMType] # 变量类型
|
||
Initializer: rowptr[t.CVoid] # 初始化值 Value*（None = 无初始化）
|
||
IsConstant: int # 1=constant, 0=variable
|
||
Linkage: str # 链接类型（"private"/"internal"/"external"/...）
|
||
Section: str # 段名（None = 默认）
|
||
UnnamedAddr: int # 1=unnamed_addr, 2=local_unnamed_addr, 0=none
|
||
Align: int # 对齐字节数（0=默认对齐，如 @c.Attribute(t.attr.aligned(16))）
|
||
|
||
|
||
# ============================================================
|
||
# NamedTypeNode: 命名结构体类型链表节点
|
||
#
|
||
# 持有 LLVMType* 指针（必须是命名 Struct 类型）。
|
||
# 继承 GSListNode[NamedTypeNode] 获取强类型 Next。
|
||
# ============================================================
|
||
@t.NoVTable
|
||
class NamedTypeNode(GSListNode[NamedTypeNode]):
|
||
Ty: rowptr[LLVMType] # 命名结构体类型指针
|
||
IsPacked: int # 1=packed 结构体（输出 <{...}>，无字段填充）
|
||
|
||
|
||
# ============================================================
|
||
# LLVMModule: LLVM IR 模块容器
|
||
#
|
||
# 使用嵌入式 Head/Tail/Count 字段代替 GSList 指针，
|
||
# 避免泛型 GSList 的 __sizeof__ 返回 0 和 vtable 偏移不一致问题。
|
||
# LLVMModulePrint 输出完整 .ll 文件内容。
|
||
# 注意: 类名必须为 LLVMModule 而非 Module，避免与 ast.Module 命名冲突
|
||
# ============================================================
|
||
@t.NoVTable
|
||
class LLVMModule:
|
||
Name: str # 模块名
|
||
FuncHead: rowptr[Function] # 函数链表头
|
||
FuncTail: rowptr[Function] # 函数链表尾
|
||
FuncCount: t.CSizeT # 函数计数
|
||
GlobalHead: rowptr[GlobalVariable] # 全局变量链表头
|
||
GlobalTail: rowptr[GlobalVariable] # 全局变量链表尾
|
||
GlobalCount: t.CSizeT # 全局变量计数
|
||
TargetTriple: str # 目标三元组（如 "x86_64-pc-linux-gnu"）
|
||
DataLayout: str # 数据布局（如 "e-m:e-p270:32:32-...")
|
||
NamedTypeHead: rowptr[NamedTypeNode] # 命名结构体类型链表头
|
||
NamedTypeTail: rowptr[NamedTypeNode] # 命名结构体类型链表尾
|
||
NamedTypeCount: t.CSizeT # 命名结构体类型计数
|
||
# 总计: 12 字段 * 8 = 96 字节（无 vtable）
|
||
|
||
# ============================================================
|
||
# 实例方法（对象化重构：原 module_* 顶层函数改写为方法，
|
||
# 原顶层函数保留为一行委托）
|
||
# ============================================================
|
||
|
||
# ---- 链表操作 ----
|
||
def add_function(self, func: rowptr[Function]):
|
||
"""将函数追加到模块函数链表（O(1)）"""
|
||
if self is None or func is None: return
|
||
func.Next = None
|
||
if self.FuncHead is None:
|
||
self.FuncHead = func
|
||
else:
|
||
self.FuncTail.Next = func
|
||
self.FuncTail = func
|
||
self.FuncCount += 1
|
||
|
||
def add_global(self, gv: rowptr[GlobalVariable]):
|
||
"""将全局变量追加到模块全局变量链表（O(1)）"""
|
||
if self is None or gv is None: return
|
||
gv.Next = None
|
||
if self.GlobalHead is None:
|
||
self.GlobalHead = gv
|
||
else:
|
||
self.GlobalTail.Next = gv
|
||
self.GlobalTail = gv
|
||
self.GlobalCount += 1
|
||
|
||
# ---- 目标属性 ----
|
||
def set_target(self, triple: str):
|
||
"""设置目标三元组"""
|
||
if self is None: return
|
||
self.TargetTriple = triple
|
||
|
||
def set_datalayout(self, layout: str):
|
||
"""设置数据布局"""
|
||
if self is None: return
|
||
self.DataLayout = layout
|
||
|
||
# ---- 命名结构体类型管理 ----
|
||
def add_named_type(self, pool: rowptr[memhub.MemBuddy], ty: rowptr[LLVMType]):
|
||
"""将命名结构体类型添加到模块（用于输出 %"name" = type { ... } 定义行）
|
||
|
||
非命名结构体（Name 为 None 或非 Struct 类型）会被忽略。
|
||
按 Name 去重；同名时 opaque 声明可被完整定义替换（供跨模块值类型 alloca）。
|
||
"""
|
||
if self is None or pool is None or ty is None:
|
||
return
|
||
name: str = get_struct_name(ty)
|
||
if name is None:
|
||
return
|
||
# 去重检查（opaque 声明可被完整定义替换）
|
||
cur: rowptr[NamedTypeNode] = self.NamedTypeHead
|
||
cnt_dedup: int = 0
|
||
while cur is not None:
|
||
# 步数守卫：NamedTypeHead 链表成环（内存损坏/插入 bug）会导致
|
||
# 无限遍历挂起（G3 编译 LinkedNode.prepend 挂起根因）。
|
||
# 正常链表节点数远小于此上限，超限即链表损坏 → 立即报错终止。
|
||
cnt_dedup += 1
|
||
if cnt_dedup > 100000:
|
||
stdio.printf("[ANT-LOOP] NamedTypeHead 链表疑似成环, name=%s, 遍历节点=%d, 强制终止\n",
|
||
name if name is not None else "(null)", cnt_dedup)
|
||
stdlib.exit(1)
|
||
if cur.Ty is not None:
|
||
existing_name: str = get_struct_name(cur.Ty)
|
||
if existing_name is not None:
|
||
if string.strcmp(existing_name, name) == 0:
|
||
# 已有完整定义则无需替换
|
||
if _is_opaque_struct(cur.Ty) == 0:
|
||
return
|
||
# 已有 opaque，新类型也是 opaque 则不重复
|
||
if _is_opaque_struct(ty) != 0:
|
||
return
|
||
# 用完整定义替换 opaque 声明（原地替换 Ty 指针）
|
||
cur.Ty = ty
|
||
return
|
||
cur = cur.Next
|
||
# 创建新节点
|
||
node: rowptr[NamedTypeNode] = pool.alloc(NAMED_TYPE_NODE_SIZE)
|
||
if node is None:
|
||
return
|
||
string.memset(node, 0, NAMED_TYPE_NODE_SIZE)
|
||
node.Ty = ty
|
||
node.IsPacked = 0
|
||
# opaque 类型插入链表头部，确保在引用它的完整定义之前声明。
|
||
# 例如 TypeRegistry{HashTable*, ...} 中，HashTable 的 opaque 声明
|
||
# 必须在 TypeRegistry 定义之前，否则 llc 报 "use of undefined type"。
|
||
# 完整定义追加到链表尾部，保持原有顺序。
|
||
if _is_opaque_struct(ty) != 0:
|
||
node.Next = self.NamedTypeHead
|
||
if self.NamedTypeHead is None:
|
||
self.NamedTypeTail = node
|
||
self.NamedTypeHead = node
|
||
else:
|
||
node.Next = None
|
||
if self.NamedTypeHead is None:
|
||
self.NamedTypeHead = node
|
||
else:
|
||
self.NamedTypeTail.Next = node
|
||
self.NamedTypeTail = node
|
||
self.NamedTypeCount += 1
|
||
|
||
def _is_type_defined_in_module(self, name: str) -> int:
|
||
"""检查模块中是否已定义指定名称的结构体类型，返回 1=已定义 / 0=未定义"""
|
||
if self is None or name is None:
|
||
return 0
|
||
cur: rowptr[NamedTypeNode] = self.NamedTypeHead
|
||
cnt_tdm: int = 0
|
||
while cur is not None:
|
||
# 步数守卫：同上，防 NamedTypeHead 链表成环无限遍历
|
||
cnt_tdm += 1
|
||
if cnt_tdm > 100000:
|
||
stdio.printf("[TDM-LOOP] NamedTypeHead 链表疑似成环, name=%s, 强制终止\n",
|
||
name if name is not None else "(null)")
|
||
stdlib.exit(1)
|
||
if cur.Ty is not None:
|
||
existing_name: str = get_struct_name(cur.Ty)
|
||
if existing_name is not None:
|
||
if string.strcmp(existing_name, name) == 0:
|
||
return 1
|
||
cur = cur.Next
|
||
return 0
|
||
|
||
def ensure_opaque_for_type(self, pool: rowptr[memhub.MemBuddy],
|
||
ty: rowptr[LLVMType]):
|
||
"""为类型中的跨模块结构体引用添加 opaque 声明
|
||
|
||
当类型引用了 Ptr(Struct(name)) 且 name 包含 '.'（SHA1 限定名）时，
|
||
如果该名称未在模块中定义，则添加 opaque 声明。
|
||
"""
|
||
_scan_type_for_cross_module_refs(pool, self, ty)
|
||
|
||
# ---- 打印 ----
|
||
def PrintModule(self, buf: str, size: t.CSizeT,
|
||
pool: rowptr[memhub.MemBuddy], mode: int):
|
||
"""将模块序列化为 .ll 文本写入 buf"""
|
||
if self is None or buf is None or size == 0: return
|
||
|
||
# 头部注释（stub 和 full 模式输出，text 模式不输出）
|
||
if mode == OUTPUT_FULL or mode == OUTPUT_STUB:
|
||
_append_cstr(buf, size, "; ModuleID = '")
|
||
if self.Name is not None:
|
||
_append_cstr(buf, size, self.Name)
|
||
_append_cstr(buf, size, "'\n")
|
||
|
||
# 目标三元组
|
||
if self.TargetTriple is not None:
|
||
_append_cstr(buf, size, "target triple = \"")
|
||
_append_cstr(buf, size, self.TargetTriple)
|
||
_append_cstr(buf, size, "\"\n")
|
||
|
||
# 数据布局
|
||
if self.DataLayout is not None:
|
||
_append_cstr(buf, size, "target datalayout = \"")
|
||
_append_cstr(buf, size, self.DataLayout)
|
||
_append_cstr(buf, size, "\"\n")
|
||
|
||
# 空行
|
||
_append_cstr(buf, size, "\n")
|
||
|
||
# 命名结构体类型定义（stub 和 full 模式输出，text 模式不输出）
|
||
if mode == OUTPUT_FULL or mode == OUTPUT_STUB:
|
||
# 扫描所有函数签名，为跨模块结构体引用添加 opaque 定义
|
||
# 确保 stub.ll 包含被 declare/define 引用的跨模块类型定义
|
||
# 修复：alloc_buf 返回 viperio.Buf，但 Buf 类型未在 NamedTypeHead 中注册
|
||
func_scan: rowptr[Function] = self.FuncHead
|
||
func_scan_steps: t.CSizeT = 0
|
||
while func_scan is not None:
|
||
func_scan_steps += 1
|
||
if func_scan_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: module funcs（FuncCount=%zu，dump 终止）\n",
|
||
self.FuncCount)
|
||
stdlib.exit(1)
|
||
if func_scan.RetTy is not None:
|
||
self.ensure_opaque_for_type(pool, func_scan.RetTy)
|
||
if func_scan.Params is not None:
|
||
# [DBGCHAIN] Function.Params 指针合理性护栏（临时诊断，用后移除）
|
||
fp_v: t.CUInt64T = t.CUInt64T(func_scan.Params)
|
||
if fp_v < 0x10000 or fp_v == t.CUInt64T(0xFFFFFFFFFFFFFFFF) or ((fp_v >> 56) & 0xFF) == 0xFF:
|
||
fnp: str = "(nil)"
|
||
if func_scan.Name is not None:
|
||
fnp = func_scan.Name
|
||
stdio.printf("[DBGCHAIN] Function.Params 野值: func=%s func=%llx steps=%zu Params=%llx FuncCount=%zu\n",
|
||
fnp, t.CUInt64T(func_scan), func_scan_steps, fp_v, t.CUInt64T(self.FuncCount))
|
||
stdio.fflush(None)
|
||
stdlib.exit(1)
|
||
params_scan: rowptr[GSList[Param]] = func_scan.Params
|
||
param_scan: rowptr[Param] = params_scan.Head
|
||
param_scan_steps: t.CSizeT = 0
|
||
while param_scan is not None:
|
||
param_scan_steps += 1
|
||
if param_scan_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表异常: func params（函数名=%s，Params.Count=%zu，param=%llu next=%llu，遍历超 %d 步，终止）\n",
|
||
func_scan.Name, params_scan.Count,
|
||
t.CUInt64T(param_scan), t.CUInt64T(param_scan.Next), IR_DUMP_CHAIN_LIMIT)
|
||
stdlib.exit(1)
|
||
# [DBGCHAIN] param 指针合理性护栏（临时诊断，用后移除）：
|
||
# 指针被写坏成小值/全 F 时先打印归属函数与链表状态，再退出
|
||
pv_scan: t.CUInt64T = t.CUInt64T(param_scan)
|
||
pv_next: t.CUInt64T = 0
|
||
if param_scan.Next is not None:
|
||
pv_next = t.CUInt64T(param_scan.Next)
|
||
if pv_scan < 0x10000 or pv_scan == t.CUInt64T(0xFFFFFFFFFFFFFFFF) or ((pv_scan >> 56) & 0xFF) == 0xFF:
|
||
stdio.printf("[DBGCHAIN] param 指针野值: func=%s steps=%zu param=%llx next=%llx Params.Head=%llx Params.Tail=%llx Params.Count=%zu\n",
|
||
func_scan.Name, param_scan_steps, pv_scan, pv_next,
|
||
t.CUInt64T(params_scan.Head), t.CUInt64T(params_scan.Tail),
|
||
t.CUInt64T(params_scan.Count))
|
||
stdio.fflush(None)
|
||
stdlib.exit(1)
|
||
# 用访问器（__function.vp 完整类型内）绕开 stub 字段访问
|
||
# 陷阱：跨模块 struct 字段访问可能生成错误 GEP/静默跳过，
|
||
# 导致 param_scan 永不更新死循环。
|
||
if param_get_ty(param_scan) is not None:
|
||
self.ensure_opaque_for_type(pool, param_get_ty(param_scan))
|
||
param_scan = param_get_next(param_scan)
|
||
func_scan = func_scan.Next
|
||
nt: rowptr[NamedTypeNode] = self.NamedTypeHead
|
||
# 结构体定义缓冲区：48 字段 × ~50 字节/字段类型名 ≈ 2400 字节，4096 足够
|
||
# 在循环外分配一次并复用，减少 pool 碎片
|
||
ty_buf: str = pool.alloc(4096)
|
||
nt_steps: t.CSizeT = 0
|
||
while nt is not None:
|
||
nt_steps += 1
|
||
if nt_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: module named types（NamedTypeCount=%zu，dump 终止）\n",
|
||
self.NamedTypeCount)
|
||
stdlib.exit(1)
|
||
if nt.Ty is not None:
|
||
if ty_buf is not None:
|
||
ty_buf[0] = '\0'
|
||
PrintStructDefinition(ty_buf, 4096, nt.Ty, pool, nt.IsPacked)
|
||
_append_cstr(buf, size, ty_buf)
|
||
_append_cstr(buf, size, "\n")
|
||
nt = nt.Next
|
||
|
||
if self.NamedTypeHead is not None:
|
||
_append_cstr(buf, size, "\n")
|
||
|
||
# 全局变量
|
||
gv: rowptr[GlobalVariable] = self.GlobalHead
|
||
gv_steps: t.CSizeT = 0
|
||
if mode == OUTPUT_STUB:
|
||
# stub 模式：输出 external global（无初值）
|
||
while gv is not None:
|
||
gv_steps += 1
|
||
if gv_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: module globals stub（GlobalCount=%zu，dump 终止）\n",
|
||
self.GlobalCount)
|
||
stdlib.exit(1)
|
||
_print_global_stub(buf, size, gv, pool)
|
||
_append_cstr(buf, size, "\n")
|
||
gv = gv.Next
|
||
else:
|
||
# full 和 text 模式：输出完整全局变量
|
||
# text 模式跳过无初值的全局变量（纯声明）
|
||
gv_idx: int = 0
|
||
while gv is not None:
|
||
gv_steps += 1
|
||
if gv_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: module globals（GlobalCount=%zu，dump 终止）\n",
|
||
self.GlobalCount)
|
||
stdlib.exit(1)
|
||
if mode == OUTPUT_TEXT and gv.Initializer is None:
|
||
gv = gv.Next
|
||
gv_idx += 1
|
||
continue
|
||
_print_global(buf, size, gv, pool)
|
||
_append_cstr(buf, size, "\n")
|
||
gv = gv.Next
|
||
gv_idx += 1
|
||
|
||
if self.GlobalHead is not None and (mode == OUTPUT_FULL or mode == OUTPUT_STUB):
|
||
_append_cstr(buf, size, "\n")
|
||
elif self.GlobalHead is not None and mode == OUTPUT_TEXT:
|
||
# text 模式下如果输出了全局变量，也加空行
|
||
had_global: int = 0
|
||
gv2: rowptr[GlobalVariable] = self.GlobalHead
|
||
gv2_steps: t.CSizeT = 0
|
||
while gv2 is not None:
|
||
gv2_steps += 1
|
||
if gv2_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: module globals text（GlobalCount=%zu，dump 终止）\n",
|
||
self.GlobalCount)
|
||
stdlib.exit(1)
|
||
if gv2.Initializer is not None:
|
||
had_global = 1
|
||
break
|
||
gv2 = gv2.Next
|
||
if had_global != 0:
|
||
_append_cstr(buf, size, "\n")
|
||
|
||
# 遍历函数
|
||
func: rowptr[Function] = self.FuncHead
|
||
func_idx: int = 0
|
||
func_steps: t.CSizeT = 0
|
||
while func is not None:
|
||
func_steps += 1
|
||
if func_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: module funcs print（FuncCount=%zu，dump 终止）\n",
|
||
self.FuncCount)
|
||
stdlib.exit(1)
|
||
if mode == OUTPUT_STUB:
|
||
# stub 模式：仅输出 declare（IsDeclared==1）
|
||
if func.IsDeclared == 1:
|
||
FunctionPrint(buf, size, func, pool)
|
||
_append_cstr(buf, size, "\n")
|
||
elif mode == OUTPUT_TEXT:
|
||
# text 模式：仅输出 define（IsDeclared==0）
|
||
if func.IsDeclared == 0:
|
||
FunctionPrint(buf, size, func, pool)
|
||
_append_cstr(buf, size, "\n")
|
||
else:
|
||
# full 模式：输出所有函数
|
||
FunctionPrint(buf, size, func, pool)
|
||
_append_cstr(buf, size, "\n")
|
||
func = func.Next
|
||
func_idx += 1
|
||
|
||
|
||
# ============================================================
|
||
# 工厂函数
|
||
# ============================================================
|
||
def new_module(pool: rowptr[memhub.MemBuddy], name: str) -> rowptr[LLVMModule]:
|
||
"""创建一个模块"""
|
||
# 硬编码 96 字节：12 字段 * 8 字节（无 vtable，@t.NoVTable）
|
||
ptr: rowptr[LLVMModule] = pool.alloc(96)
|
||
if ptr is None: return None
|
||
string.memset(ptr, 0, 96)
|
||
ptr.Name = name
|
||
ptr.FuncHead = None
|
||
ptr.FuncTail = None
|
||
ptr.FuncCount = 0
|
||
ptr.GlobalHead = None
|
||
ptr.GlobalTail = None
|
||
ptr.GlobalCount = 0
|
||
ptr.TargetTriple = None
|
||
ptr.DataLayout = None
|
||
ptr.NamedTypeHead = None
|
||
ptr.NamedTypeTail = None
|
||
ptr.NamedTypeCount = 0
|
||
return ptr
|
||
|
||
|
||
# GlobalVariable 硬编码大小: Next(8) + Name(8) + Ty(8) + Initializer(8) + IsConstant(4) + Linkage(8) + Section(8) + UnnamedAddr(4) + Align(4) = 72 字节
|
||
GV_SIZE: t.CDefine = 72
|
||
|
||
# NamedTypeNode 硬编码大小: Next(8) + Ty(8) + IsPacked(8) = 24 字节
|
||
NAMED_TYPE_NODE_SIZE: t.CDefine = 24
|
||
|
||
def new_global_variable(pool: rowptr[memhub.MemBuddy],
|
||
name: str, ty: rowptr[LLVMType]) -> rowptr[GlobalVariable]:
|
||
"""创建一个全局变量"""
|
||
ptr: rowptr[GlobalVariable] = pool.alloc(GV_SIZE)
|
||
if ptr is None: return None
|
||
string.memset(ptr, 0, GV_SIZE)
|
||
ptr.Name = name
|
||
ptr.Ty = ty
|
||
ptr.Initializer = None
|
||
ptr.IsConstant = 0
|
||
ptr.Linkage = "external"
|
||
ptr.Section = None
|
||
ptr.UnnamedAddr = 0
|
||
ptr.Align = 0
|
||
return ptr
|
||
|
||
|
||
# ============================================================
|
||
# 链表操作（嵌入式 Head/Tail/Count，O(1) 追加）
|
||
# ============================================================
|
||
def module_add_function(mod: rowptr[LLVMModule], func: rowptr[Function]):
|
||
mod.add_function(func)
|
||
|
||
|
||
def module_get_func_head(mod: rowptr[LLVMModule]) -> rowptr[Function]:
|
||
"""获取模块函数链表头（跨模块访问器，绕 stub 字段访问陷阱）"""
|
||
if mod is None:
|
||
return None
|
||
return mod.FuncHead
|
||
|
||
|
||
def module_add_global(mod: rowptr[LLVMModule], gv: rowptr[GlobalVariable]):
|
||
mod.add_global(gv)
|
||
|
||
|
||
def module_set_target(mod: rowptr[LLVMModule], triple: str):
|
||
mod.set_target(triple)
|
||
|
||
|
||
def module_set_datalayout(mod: rowptr[LLVMModule], layout: str):
|
||
mod.set_datalayout(layout)
|
||
|
||
|
||
# ============================================================
|
||
# 命名结构体类型管理
|
||
# ============================================================
|
||
|
||
# ============================================================
|
||
# _is_opaque_struct - 检查类型是否是 opaque 结构体
|
||
#
|
||
# opaque 结构体特征: Struct 类型，fcount == -1（哨兵标记，跨模块前向
|
||
# 声明，仅名称无字段定义）。通过 Struct(pool, None, -1, name) 创建，
|
||
# IR 输出为 %"name" = type opaque。注意区分空结构体（fcount==0,
|
||
# fields==None，无字段类）——它是合法完整定义（= type {}，可 alloca）。
|
||
# 用于在 module_add_named_type 中区分 opaque 声明和完整定义，
|
||
# 确保 opaque 声明插入链表头部（在使用之前声明）。
|
||
#
|
||
# Returns: 1=是 opaque / 0=否
|
||
# ============================================================
|
||
def _is_opaque_struct(ty: rowptr[LLVMType]) -> int:
|
||
"""检查类型是否是 opaque 结构体
|
||
|
||
opaque 是跨模块前向声明（仅名称无字段定义），由
|
||
_scan_type_for_cross_module_refs 创建，fcount=-1 哨兵标记。
|
||
空结构体（fcount==0, fields==None，无字段类）是合法完整定义
|
||
（LLVM {} 类型，可 alloca），必须与 opaque 区分——此前混淆导致
|
||
空类结构体被输出为 "= type opaque"，alloca 报 Cannot allocate
|
||
unsized type。
|
||
"""
|
||
if ty is None:
|
||
return 0
|
||
match ty:
|
||
case LLVMType.Struct(fields, fcount, name):
|
||
# opaque: fcount == -1（哨兵）
|
||
if fcount < 0:
|
||
return 1
|
||
return 0
|
||
case _:
|
||
return 0
|
||
|
||
|
||
def module_add_named_type(mod: rowptr[LLVMModule], pool: rowptr[memhub.MemBuddy],
|
||
ty: rowptr[LLVMType]):
|
||
mod.add_named_type(pool, ty)
|
||
|
||
|
||
def module_set_named_type_packed(mod: rowptr[LLVMModule], pool: rowptr[memhub.MemBuddy],
|
||
ty: rowptr[LLVMType], is_packed: int):
|
||
"""按名称设置命名结构体的 packed 标记（输出 %"name" = type <{...}>）。
|
||
|
||
编译器解析 @c.Attribute(t.attr.packed) 后调用，使 llc 按 packed 布局
|
||
（无字段填充）计算字段偏移/GEP。ty 为命名 Struct 类型（按 get_struct_name
|
||
定位已注册节点）；未找到则静默跳过（opaque/跨模块 stub 场景无布局需求）。
|
||
"""
|
||
if mod is None or ty is None:
|
||
return
|
||
name: str = get_struct_name(ty)
|
||
if name is None:
|
||
return
|
||
cur: rowptr[NamedTypeNode] = mod.NamedTypeHead
|
||
cnt_p: int = 0
|
||
while cur is not None:
|
||
cnt_p += 1
|
||
if cnt_p > 100000:
|
||
return
|
||
if cur.Ty is not None:
|
||
existing_name: str = get_struct_name(cur.Ty)
|
||
if existing_name is not None and string.strcmp(existing_name, name) == 0:
|
||
cur.IsPacked = is_packed
|
||
return
|
||
cur = cur.Next
|
||
|
||
|
||
# ============================================================
|
||
# module_ensure_opaque_for_type - 为类型中的跨模块结构体引用添加 opaque 声明
|
||
#
|
||
# 递归扫描 LLVMType 树，当遇到 Ptr(Struct(name)) 且 name 包含 '.'（SHA1 限定名）时，
|
||
# 如果该名称未在模块中定义，则添加 opaque 声明（%"name" = type opaque）。
|
||
# 这解决了跨模块结构体引用导致的 llc "use of undefined type" 错误。
|
||
# ============================================================
|
||
def _is_type_defined_in_module(mod: rowptr[LLVMModule], name: str) -> int:
|
||
return mod._is_type_defined_in_module(name)
|
||
|
||
|
||
def _scan_type_for_cross_module_refs(pool: rowptr[memhub.MemBuddy],
|
||
mod: rowptr[LLVMModule],
|
||
ty: rowptr[LLVMType],
|
||
depth: int = 0):
|
||
"""递归扫描类型树，为跨模块结构体引用添加 opaque 声明"""
|
||
if ty is None or mod is None or pool is None:
|
||
return
|
||
# 深度守卫：自引用结构体（链表/树节点，如 Node{...; Ptr(Node) next}）
|
||
# 会无限递归栈溢出，与其他递归（TypePrint/DeepCopyType）一致设上限
|
||
if depth >= 100:
|
||
return
|
||
match ty:
|
||
case LLVMType.Ptr(pointee):
|
||
if pointee is not None:
|
||
match pointee:
|
||
case LLVMType.Struct(sfields, sfcount, sname):
|
||
if sname is not None:
|
||
# 检查是否为跨模块引用（名称包含 '.'）
|
||
if string.strchr(sname, 46) is not None:
|
||
# 检查是否已在模块中定义
|
||
if _is_type_defined_in_module(mod, sname) == 0:
|
||
# 创建 opaque 类型并添加到模块
|
||
# fcount=-1 哨兵标记 opaque（区别于空结构体 {}，
|
||
# 后者 fcount==0 是合法完整定义，可 alloca）
|
||
opaque_ty: rowptr[LLVMType] = Struct(pool, None, -1, sname)
|
||
if opaque_ty is not None:
|
||
module_add_named_type(mod, pool, opaque_ty)
|
||
# 递归扫描结构体字段中的跨模块引用
|
||
# 例如 TypeRegistry{HashTable*, MemManager*} 中的 HashTable 引用
|
||
if sfields is not None:
|
||
fld: rowptr[ParamNode] = sfields
|
||
fld_steps: t.CSizeT = 0
|
||
while fld is not None:
|
||
fld_steps += 1
|
||
if fld_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: struct fields（sname=%s，dump 终止）\n",
|
||
sname)
|
||
stdlib.exit(1)
|
||
if fld.Ty is not None:
|
||
_scan_type_for_cross_module_refs(pool, mod, fld.Ty, depth + 1)
|
||
fld = fld.Next
|
||
case _:
|
||
_scan_type_for_cross_module_refs(pool, mod, pointee, depth + 1)
|
||
case LLVMType.Struct(sfields2, sfcount2, sname2):
|
||
# 顶层结构体（非指针）：递归扫描字段
|
||
if sfields2 is not None:
|
||
fld2: rowptr[ParamNode] = sfields2
|
||
fld2_steps: t.CSizeT = 0
|
||
while fld2 is not None:
|
||
fld2_steps += 1
|
||
if fld2_steps > IR_DUMP_CHAIN_LIMIT:
|
||
stdio.printf("[IR-DUMP] 链表成环: struct fields2（sname=%s，dump 终止）\n",
|
||
sname2)
|
||
stdlib.exit(1)
|
||
if fld2.Ty is not None:
|
||
_scan_type_for_cross_module_refs(pool, mod, fld2.Ty, depth + 1)
|
||
fld2 = fld2.Next
|
||
case LLVMType.Array(elem_ty, acount):
|
||
if elem_ty is not None:
|
||
_scan_type_for_cross_module_refs(pool, mod, elem_ty, depth + 1)
|
||
case _:
|
||
pass
|
||
|
||
|
||
def module_ensure_opaque_for_type(mod: rowptr[LLVMModule],
|
||
pool: rowptr[memhub.MemBuddy],
|
||
ty: rowptr[LLVMType]):
|
||
mod.ensure_opaque_for_type(pool, ty)
|
||
|
||
|
||
# ============================================================
|
||
# 全局变量属性设置
|
||
# ============================================================
|
||
def global_set_constant(gv: rowptr[GlobalVariable], is_const: int):
|
||
"""设置全局变量是否为常量"""
|
||
if gv is None: return
|
||
gv.IsConstant = is_const
|
||
|
||
|
||
def global_set_linkage(gv: rowptr[GlobalVariable], linkage: str):
|
||
"""设置链接类型 (private/internal/external/... )"""
|
||
if gv is None: return
|
||
gv.Linkage = linkage
|
||
|
||
|
||
def global_set_unnamed_addr(gv: rowptr[GlobalVariable], level: int):
|
||
"""设置 unnamed_addr 属性 (0=none, 1=unnamed_addr, 2=local_unnamed_addr)"""
|
||
if gv is None: return
|
||
gv.UnnamedAddr = level
|
||
|
||
|
||
def global_set_section(gv: rowptr[GlobalVariable], section: str):
|
||
"""设置段名"""
|
||
if gv is None: return
|
||
gv.Section = section
|
||
|
||
|
||
# ============================================================
|
||
# 输出模式常量
|
||
# ============================================================
|
||
OUTPUT_FULL: t.CDefine = 0 # 完整 IR（stub + text）
|
||
OUTPUT_STUB: t.CDefine = 1 # 仅声明（header + 类型定义 + declare + external global）
|
||
OUTPUT_TEXT: t.CDefine = 2 # 仅代码（define + global 带初值）
|
||
|
||
# dump 哨兵：IR 链表遍历步数上限（成环检测，fail-fast）。
|
||
# 任何一条链（函数/全局/命名类型/结构体字段）的 Next 被越界写破坏成环时，
|
||
# 遍历会无限循环导致 tpv.exe 挂起（builtins.vp dump 卡死根因排查）。
|
||
# 超过上限即打印链表名并 exit(1)，把"卡死"变成"确定性报错定位"
|
||
# （对齐 memhub.self_check 的 max_steps 哲学）。
|
||
IR_DUMP_CHAIN_LIMIT: t.CDefine = 100000
|
||
|
||
|
||
# ============================================================
|
||
# LLVMModulePrint: 将模块序列化为 .ll 文本
|
||
#
|
||
# mode: OUTPUT_FULL=完整, OUTPUT_STUB=仅声明, OUTPUT_TEXT=仅代码
|
||
# ============================================================
|
||
def LLVMModulePrint(buf: str, size: t.CSizeT,
|
||
mod: rowptr[LLVMModule], pool: rowptr[memhub.MemBuddy],
|
||
mode: int):
|
||
mod.PrintModule(buf, size, pool, mode)
|
||
|
||
|
||
# ============================================================
|
||
# 全局变量打印
|
||
# ============================================================
|
||
def _print_global(buf: str, size: t.CSizeT,
|
||
gv: rowptr[GlobalVariable], pool: rowptr[memhub.MemBuddy]):
|
||
"""打印单个全局变量: @<name> = [linkage] [unnamed_addr] [constant] <ty> <init>"""
|
||
if gv is None: return
|
||
|
||
_append_cstr(buf, size, "@")
|
||
if gv.Name is not None:
|
||
if _ll_name_needs_quote(gv.Name) != 0:
|
||
_append_cstr(buf, size, "\"")
|
||
_append_cstr(buf, size, gv.Name)
|
||
_append_cstr(buf, size, "\"")
|
||
else:
|
||
_append_cstr(buf, size, gv.Name)
|
||
_append_cstr(buf, size, " = ")
|
||
|
||
# 链接类型（有初始值时不输出 external，因为 external + 初始值是非法 IR）
|
||
if gv.Linkage is not None:
|
||
if gv.Initializer is not None and string.strcmp(gv.Linkage, "external") == 0:
|
||
pass
|
||
else:
|
||
_append_cstr(buf, size, gv.Linkage)
|
||
_append_cstr(buf, size, " ")
|
||
|
||
# unnamed_addr
|
||
if gv.UnnamedAddr == 1:
|
||
_append_cstr(buf, size, "unnamed_addr ")
|
||
elif gv.UnnamedAddr == 2:
|
||
_append_cstr(buf, size, "local_unnamed_addr ")
|
||
|
||
# constant / global
|
||
if gv.IsConstant == 1:
|
||
_append_cstr(buf, size, "constant ")
|
||
else:
|
||
_append_cstr(buf, size, "global ")
|
||
|
||
# 类型
|
||
ty_buf: str = pool.alloc(512)
|
||
if ty_buf is not None:
|
||
ty_buf[0] = '\0'
|
||
TypePrint(ty_buf, 512, gv.Ty, pool)
|
||
_append_cstr(buf, size, ty_buf)
|
||
|
||
# 初始化值
|
||
if gv.Initializer is not None:
|
||
_append_cstr(buf, size, " ")
|
||
_append_cstr(buf, size, gv.Initializer)
|
||
|
||
# 段名
|
||
if gv.Section is not None:
|
||
_append_cstr(buf, size, ", section \"")
|
||
_append_cstr(buf, size, gv.Section)
|
||
_append_cstr(buf, size, "\"")
|
||
|
||
# 对齐（@c.Attribute(t.attr.aligned(N)) 类的全局实例，如 fpu_context 16 对齐）
|
||
if gv.Align > 0:
|
||
align_buf: str = pool.alloc(32)
|
||
if align_buf is not None:
|
||
viperlib.snprintf(align_buf, 32, ", align %d", gv.Align)
|
||
_append_cstr(buf, size, align_buf)
|
||
|
||
|
||
# ============================================================
|
||
# _print_global_stub - 打印全局变量的声明形式（external global，无初值）
|
||
# ============================================================
|
||
def _print_global_stub(buf: str, size: t.CSizeT,
|
||
gv: rowptr[GlobalVariable], pool: rowptr[memhub.MemBuddy]):
|
||
"""打印全局变量声明: @<name> = external global <ty>"""
|
||
if gv is None: return
|
||
|
||
_append_cstr(buf, size, "@")
|
||
if gv.Name is not None:
|
||
if _ll_name_needs_quote(gv.Name) != 0:
|
||
_append_cstr(buf, size, "\"")
|
||
_append_cstr(buf, size, gv.Name)
|
||
_append_cstr(buf, size, "\"")
|
||
else:
|
||
_append_cstr(buf, size, gv.Name)
|
||
_append_cstr(buf, size, " = external global ")
|
||
|
||
# 类型
|
||
ty_buf: str = pool.alloc(512)
|
||
if ty_buf is not None:
|
||
ty_buf[0] = '\0'
|
||
TypePrint(ty_buf, 512, gv.Ty, pool)
|
||
_append_cstr(buf, size, ty_buf)
|
||
|
||
|
||
# ============================================================
|
||
# 内部辅助
|
||
# ============================================================
|
||
# _append_cstr 写入位置缓存（性能优化：同一缓冲连续追加时 O(1) 定位，
|
||
# 避免每次 strlen(dst) 从头扫描导致的 O(n²) —— IR 序列化热点）。
|
||
# 缓冲切换（dst 不同）或缓冲被重置（dst[0]=0 且 pos>0）时自动回退 strlen。
|
||
_g_last_dst: t.CUInt64T = 0
|
||
_g_last_pos: t.CSizeT = 0
|
||
|
||
def _append_cstr(dst: str, dst_size: t.CSizeT, src: str):
|
||
"""将 C 字符串 src 追加到 dst 末尾"""
|
||
global _g_last_dst, _g_last_pos
|
||
if dst is None or src is None: return
|
||
dlen: t.CSizeT = 0
|
||
if _g_last_dst == t.CUInt64T(dst):
|
||
dlen = _g_last_pos
|
||
# 缓冲被重置（调用方 buf[0]=0 后复用同一地址）→ 回到头部
|
||
if dlen > 0 and dst[0] == '\0':
|
||
dlen = 0
|
||
else:
|
||
dlen = string.strlen(dst)
|
||
_g_last_dst = t.CUInt64T(dst)
|
||
# 无符号下溢防护：dlen 超过缓冲大小时（缓冲已被破坏/悬空）直接返回，
|
||
# 否则 remain = dst_size - dlen 下溢成巨值会把 dst[dlen+i] 写穿到缓冲外
|
||
if dlen >= dst_size:
|
||
return
|
||
slen: t.CSizeT = string.strlen(src)
|
||
remain: t.CSizeT = dst_size - dlen
|
||
if remain <= 0: return
|
||
i: t.CSizeT = 0
|
||
while i < slen and i + 1 < remain:
|
||
dst[dlen + i] = src[i]
|
||
i += 1
|
||
dst[dlen + i] = '\0'
|
||
_g_last_pos = dlen + i |