feat(gen): 函数生命期摘要表 + 逃逸 walker（M2.3 v1b 步骤1-3）
- StructGenMixin：EscTable/EscCount（追加在类末尾，避免偏移漂移）；register_file_funcs 每文件重置并逐函数算摘要。 - 逃逸 walker：return p / global 赋值 / obj.f 赋值，递归 if/while/for；不进嵌套 def/class；形参访问逐级显式重转（AST 退化）。 - 表内只存 u64 名哈希（djb2-64），不解引用字符串指针，消除槽内坏指针导致的 strcmp 崩溃。 - 表从 PermPool 分配并 memset。 - GenApi：get_param_escapes/get_ret_borrow 查询入口。调用点检查（步骤4）待接。
This commit is contained in:
TermiNexus committed
1 parent
a6e1957768
commit
4ce577d8fb
2 files changed
+202
-3
No files matched your search
@@ -285,6 +285,19 @@ def is_file_func(name: str) -> int:
|
||||
return 0
|
||||
|
||||
|
||||
# ---------------- 生命期摘要查询（M2.3 v1b, delegates to StructGenMixin） ----------------
|
||||
def get_param_escapes(name: str) -> int:
|
||||
"""形参逃逸掩码（第 k 位=第 k 个形参逃逸）；无记录返回 0"""
|
||||
if _g_gen is not None: return _g_gen.StructGen.get_param_escapes(name)
|
||||
return 0
|
||||
|
||||
|
||||
def get_ret_borrow(name: str) -> int:
|
||||
"""返回借用标记（1=该函数直接返回某形参）；无记录返回 0"""
|
||||
if _g_gen is not None: return _g_gen.StructGen.get_ret_borrow(name)
|
||||
return 0
|
||||
|
||||
|
||||
# ---------------- find_struct_by_type (delegates to StructGenMixin.find_struct_by_type) ----------------
|
||||
def find_struct_by_type(struct_ty: mlirlite.LLVMType*) -> GenTypes.StructEntry*:
|
||||
"""Delegate to gen.find_struct_by_type"""
|
||||
|
||||
@@ -101,13 +101,20 @@ novirtual class StructGenMixin:
|
||||
|
||||
# Top-level FunctionDef name table for this file (see register_file_funcs).
|
||||
# A stub-injected bare-name declaration (extern/state/export, no SHA1 prefix) that matches a local declaration would take the name first and make the local declaration silently dropped in forward_declare_functions; the injection side (VpiReader._inject_function_declare) skips these names.
|
||||
FileFuncNames: hashtable.HashTable*
|
||||
|
||||
FileFuncNames: hashtable.HashTable*
|
||||
|
||||
# --- Name | type index (fast lookup path; falls back to a linear scan when the index is untrusted) ---
|
||||
StructNameIdx: sizet* # name -> first registration index (u64 slot: 0 = empty, else index+1)
|
||||
StructTypeIdx: sizet* # Ty pointer value -> first registration index (same layout)
|
||||
StructIdxCap: int # index slot count (power of two, > 2*STRUCT_MAX)
|
||||
StructIdxBuilt: int # indexed entry count (-1 = untrusted; trusted only when == StructCount)
|
||||
StructIdxBuilt: int # indexed entry count (-1 = untrusted; trusted only when == StructCount)
|
||||
|
||||
# 生命期摘要（M2.3 v1b）：本文件函数 → 形参逃逸掩码 / 返回借用标记。
|
||||
# **必须追加在类字段末尾**——新增字段插在中间会改变其后字段的偏移（本仓约定，见
|
||||
# VarEntry/FieldEntry 的 "Appended at the end so existing field offsets are unchanged"）。
|
||||
# 池数组 + 线性查找；每项 3 槽：name(池上 str 地址) | mask(第 k 位=第 k 形参逃逸) | ret_borrow
|
||||
EscTable: sizet*
|
||||
EscCount: int
|
||||
|
||||
def InitStructGen(self) -> int:
|
||||
"""Initialize struct-mixin fields; called inline from subclass constructors."""
|
||||
@@ -123,6 +130,8 @@ novirtual class StructGenMixin:
|
||||
self.FileClassNames = None
|
||||
self.FileModuleSha1 = None
|
||||
self.FileFuncNames = None
|
||||
self.EscTable = None
|
||||
self.EscCount = 0
|
||||
self.StructNameIdx = None
|
||||
self.StructTypeIdx = None
|
||||
self.StructIdxCap = 0
|
||||
@@ -244,6 +253,17 @@ novirtual class StructGenMixin:
|
||||
if ch is None: return
|
||||
if self.FileFuncNames is None: self.FileFuncNames = hashtable.HashTable(pool)
|
||||
if self.FileFuncNames is None: return
|
||||
# 生命期摘要表：每次进入本文件时重置（v1b 只覆盖同文件；跨文件属 v3）
|
||||
# 两个要点（对照 init_struct_table）：
|
||||
# 1) 必须从 **PermPool** 分配——临时池在 Phase1 会被 reset_to(mark) 回收，同块内存
|
||||
# 被复用后我们会握着已变脏的旧指针（ASan 抓不到，表现为槽里读出垃圾 → strcmp 崩）。
|
||||
# 2) 必须显式清零：未写入的槽若不是 0，_EscLookup 的 strcmp 会对脏指针解引用。
|
||||
esc_cap: sizet = 1024
|
||||
esc_pool: memhub.MemBuddy* = CompilerFactory.factory.PermPool
|
||||
if esc_pool is None: esc_pool = pool
|
||||
self.EscTable = sizet*(esc_pool.alloc(esc_cap * 24))
|
||||
if self.EscTable is not None: string.memset(self.EscTable, 0, esc_cap * 24)
|
||||
self.EscCount = 0
|
||||
cn: sizet = len(ch)
|
||||
ci: sizet
|
||||
for ci in range(cn):
|
||||
@@ -259,12 +279,178 @@ novirtual class StructGenMixin:
|
||||
if nm_buf is None: continue
|
||||
string.strcpy(nm_buf, fd.name)
|
||||
self.FileFuncNames.set_int(nm_buf, 1)
|
||||
# 生命期摘要（M2.3 v1b）：表内只存 u64 名哈希——不存字符串指针，
|
||||
# 从根上消除"槽里是坏指针 → strcmp 崩在 0xffff…"这一整类问题。
|
||||
self._RegisterFuncEscapeSummary(esc_pool, fd, fd.name)
|
||||
|
||||
def is_file_func(self, name: str) -> int:
|
||||
"""Return 1 when the bare name is a top-level function declared in this file."""
|
||||
if name is None or self.FileFuncNames is None: return 0
|
||||
return self.FileFuncNames.__contains__(name)
|
||||
|
||||
# ============================================================
|
||||
# 生命期摘要（M2.3 v1b）：本文件函数 → 形参逃逸掩码 / 返回借用标记
|
||||
# 覆盖：`return p`、`global G` 下的 `G = p`、`obj.f = p`（含 AnnAssign）；
|
||||
# 递归进 if/while/for 的体与 else；**不**进嵌套 def/class。
|
||||
# 不做传递闭包(v2)、不做跨模块(v3)。
|
||||
# ============================================================
|
||||
|
||||
def _EscLookup(self, name: str) -> int:
|
||||
"""线性查找摘要项下标；未找到返回 -1"""
|
||||
if name is None or self.EscTable is None: return -1
|
||||
h: u64 = self._NameHash(name)
|
||||
cap: int = 1024
|
||||
i: int = 0
|
||||
while i < self.EscCount and i < cap:
|
||||
# 槽 0 是名哈希（u64）；未写入的槽为 0，与任何非零哈希都不相等，无需判空
|
||||
if u64(self.EscTable[i * 3 + 0]) == h: return i
|
||||
i += 1
|
||||
return -1
|
||||
|
||||
def _NameHash(self, name: str) -> u64:
|
||||
"""djb2-64（小立即数，溢出按 u64 回绕）：把函数名压成表键"""
|
||||
h: u64 = 5381
|
||||
if name is None: return h
|
||||
n: sizet = string.strlen(name)
|
||||
i: sizet = 0
|
||||
while i < n:
|
||||
h = h * 33 + u64(uint8(name[i]))
|
||||
i += 1
|
||||
return h
|
||||
|
||||
def get_param_escapes(self, name: str) -> int:
|
||||
"""形参逃逸掩码：第 k 位=第 k 个形参逃逸；无记录返回 0"""
|
||||
idx: int = self._EscLookup(name)
|
||||
if idx < 0: return 0
|
||||
return int(self.EscTable[idx * 3 + 1])
|
||||
|
||||
def get_ret_borrow(self, name: str) -> int:
|
||||
"""返回借用标记：1=该函数直接返回某形参；无记录返回 0"""
|
||||
idx: int = self._EscLookup(name)
|
||||
if idx < 0: return 0
|
||||
return int(self.EscTable[idx * 3 + 2])
|
||||
|
||||
def _RegisterFuncEscapeSummary(self, pool: memhub.MemBuddy*, fd: ast.FunctionDef*, key: str) -> int:
|
||||
"""扫函数体登记摘要；key 必须是与 EscTable 同池的字符串"""
|
||||
if pool is None or fd is None or key is None or self.EscTable is None: return 1
|
||||
if self._EscLookup(key) >= 0: return 0
|
||||
if self.EscCount >= 1024: return 1
|
||||
gset: hashtable.HashTable* = hashtable.HashTable(pool)
|
||||
if gset is not None: self._CollectGlobalNames(fd.children, gset, 0)
|
||||
box: sizet* = sizet*(pool.alloc(16))
|
||||
if box is None: return 1
|
||||
box[0] = 0
|
||||
box[1] = 0
|
||||
self._ScanEscapes(fd, fd.children, gset, box, 0)
|
||||
slot: int = self.EscCount
|
||||
self.EscTable[slot * 3 + 0] = sizet(self._NameHash(key))
|
||||
self.EscTable[slot * 3 + 1] = box[0]
|
||||
self.EscTable[slot * 3 + 2] = box[1]
|
||||
self.EscCount = slot + 1
|
||||
return 0
|
||||
|
||||
def _CollectGlobalNames(self, stmts: list[ast.AST*]*, gset: hashtable.HashTable*, depth: int) -> int:
|
||||
"""收集本函数（含嵌套块）里 `global` 声明的名字"""
|
||||
if stmts is None or gset is None or depth > 64: return 0
|
||||
n: sizet = len(stmts)
|
||||
for i in range(n):
|
||||
st: ast.AST* = stmts.get(i)
|
||||
if st is None: continue
|
||||
k: int = st.kind()
|
||||
if k == ast.ASTKind.Global:
|
||||
gn: ast.Global* = ast.Global*(st)
|
||||
if gn is not None and gn.names is not None:
|
||||
cg: sizet = len(gn.names)
|
||||
for j in range(cg):
|
||||
nnd: ast.AST* = gn.names.get(j)
|
||||
if nnd is not None and nnd.kind() == ast.ASTKind.Name:
|
||||
nnm: ast.Name* = ast.Name*(nnd)
|
||||
if nnm.id is not None: gset.set_int(nnm.id, 1)
|
||||
elif k == ast.ASTKind.If:
|
||||
ifn: ast.If* = ast.If*(st)
|
||||
self._CollectGlobalNames(ifn.children, gset, depth + 1)
|
||||
self._CollectGlobalNames(ifn.orelse, gset, depth + 1)
|
||||
elif k == ast.ASTKind.While:
|
||||
whn: ast.While* = ast.While*(st)
|
||||
self._CollectGlobalNames(whn.children, gset, depth + 1)
|
||||
self._CollectGlobalNames(whn.orelse, gset, depth + 1)
|
||||
elif k == ast.ASTKind.For:
|
||||
frn: ast.For* = ast.For*(st)
|
||||
self._CollectGlobalNames(frn.children, gset, depth + 1)
|
||||
self._CollectGlobalNames(frn.orelse, gset, depth + 1)
|
||||
return 0
|
||||
|
||||
def _ParamIndex(self, fd: ast.FunctionDef*, value: ast.AST*) -> int:
|
||||
"""值节点是形参名时返回其下标，否则 -1"""
|
||||
if value is None or value.kind() != ast.ASTKind.Name: return -1
|
||||
vnm: ast.Name* = ast.Name*(value)
|
||||
if vnm.id is None: return -1
|
||||
# 形参注解可能退化为 ast.AST*（AST 只有 children，没有 args）→ 使用点逐级显式重转
|
||||
fdef: ast.FunctionDef* = ast.FunctionDef*(fd)
|
||||
if fdef is None: return -1
|
||||
# FunctionDef.args 的成员类型在 TPC 侧只登记到 AST*，二级 .args 需再转一次
|
||||
agr: ast.Arguments* = ast.Arguments*(fdef.args)
|
||||
if agr is None or agr.args is None: return -1
|
||||
pargs: list[ast.AST*]* = agr.args
|
||||
if pargs is None: return -1
|
||||
cnt: sizet = len(pargs)
|
||||
for i in range(cnt):
|
||||
an: ast.AST* = pargs.get(i)
|
||||
if an is None: continue
|
||||
pa: ast.Arg* = ast.Arg*(an)
|
||||
if pa.arg is not None and string.strcmp(pa.arg, vnm.id) == 0: return int(i)
|
||||
return -1
|
||||
|
||||
def _MarkEscapingStore(self, fd: ast.FunctionDef*, target: ast.AST*, value: ast.AST*, gset: hashtable.HashTable*, box: sizet*) -> int:
|
||||
"""`target = 形参`：若 target 是对象字段或已声明的全局，则该形参逃逸"""
|
||||
pi: int = self._ParamIndex(fd, value)
|
||||
if pi < 0 or target is None or box is None: return 0
|
||||
tk: int = target.kind()
|
||||
if tk == ast.ASTKind.Attribute:
|
||||
box[0] = box[0] | (sizet(1) << pi)
|
||||
elif tk == ast.ASTKind.Name:
|
||||
tnm: ast.Name* = ast.Name*(target)
|
||||
if tnm.id is not None and gset is not None and gset.__contains__(tnm.id) != 0:
|
||||
box[0] = box[0] | (sizet(1) << pi)
|
||||
return 0
|
||||
|
||||
def _ScanEscapes(self, fd: ast.FunctionDef*, stmts: list[ast.AST*]*, gset: hashtable.HashTable*, box: sizet*, depth: int) -> int:
|
||||
"""递归扫描：return p / G = p / obj.f = p；只进 if/while/for，不进嵌套 def/class"""
|
||||
if fd is None or stmts is None or box is None or depth > 64: return 0
|
||||
n: sizet = len(stmts)
|
||||
for i in range(n):
|
||||
st: ast.AST* = stmts.get(i)
|
||||
if st is None: continue
|
||||
k: int = st.kind()
|
||||
if k == ast.ASTKind.Return:
|
||||
rt: ast.Return* = ast.Return*(st)
|
||||
pi: int = self._ParamIndex(fd, rt.value)
|
||||
if pi >= 0:
|
||||
box[0] = box[0] | (sizet(1) << pi)
|
||||
box[1] = 1
|
||||
elif k == ast.ASTKind.Assign:
|
||||
asg: ast.Assign* = ast.Assign*(st)
|
||||
if asg.targets is not None:
|
||||
nt: sizet = len(asg.targets)
|
||||
for t in range(nt):
|
||||
self._MarkEscapingStore(fd, asg.targets.get(t), asg.value, gset, box)
|
||||
elif k == ast.ASTKind.AnnAssign:
|
||||
aa: ast.AnnAssign* = ast.AnnAssign*(st)
|
||||
self._MarkEscapingStore(fd, aa.target, aa.value, gset, box)
|
||||
elif k == ast.ASTKind.If:
|
||||
ifn: ast.If* = ast.If*(st)
|
||||
self._ScanEscapes(fd, ifn.children, gset, box, depth + 1)
|
||||
self._ScanEscapes(fd, ifn.orelse, gset, box, depth + 1)
|
||||
elif k == ast.ASTKind.While:
|
||||
whn: ast.While* = ast.While*(st)
|
||||
self._ScanEscapes(fd, whn.children, gset, box, depth + 1)
|
||||
self._ScanEscapes(fd, whn.orelse, gset, box, depth + 1)
|
||||
elif k == ast.ASTKind.For:
|
||||
frn: ast.For* = ast.For*(st)
|
||||
self._ScanEscapes(fd, frn.children, gset, box, depth + 1)
|
||||
self._ScanEscapes(fd, frn.orelse, gset, box, depth + 1)
|
||||
return 0
|
||||
|
||||
def get_file_module_sha1(self) -> str:
|
||||
"""Return the module SHA1 of the file being translated (None if unregistered)."""
|
||||
return self.FileModuleSha1
|
||||
|
||||
Reference in new issue
Block a user