fix: 修复 Map/Set GC 追踪与 CodeForge 缓存键

This commit is contained in:
2026-06-17 09:38:48 +08:00
parent a074636619
commit 6a99209af8
10 changed files with 482 additions and 126 deletions
+89 -21
View File
@@ -6,8 +6,6 @@ use crate::builtins::set::SetKey;
use crate::native::NativeResult;
use crate::vm::Vm;
const MAP_PROP_INDEX: u8 = 0;
macro_rules! native_try {
($expr:expr) => {
match $expr {
@@ -17,7 +15,9 @@ macro_rules! native_try {
};
}
/// Retrieve the `IndexMap` pointer stored at `MAP_PROP_INDEX` of a Map object.
type MapInner = indexmap::IndexMap<SetKey, JsValue>;
/// Retrieve the `IndexMap` pointer stored in a Map object's native-data slot.
///
/// # Safety contract maintained by callers
///
@@ -27,7 +27,7 @@ macro_rules! native_try {
/// is created in `new_map_inner()` and is never freed until the process exits (intentional
/// leak — lifetime is tied to the epoch). Only one live `*mut` alias exists at a time
/// per Map object because native calls are single-threaded.
fn get_map_inner(vm: &mut Vm, this_val: JsValue) -> Result<*mut indexmap::IndexMap<SetKey, JsValue>, JsValue> {
fn get_map_inner(vm: &mut Vm, this_val: JsValue) -> Result<*mut MapInner, JsValue> {
if !this_val.is_object() {
return Err(crate::builtins::error::create_type_error(vm, "called on non-Map object"));
}
@@ -45,40 +45,108 @@ fn get_map_inner(vm: &mut Vm, this_val: JsValue) -> Result<*mut indexmap::IndexM
"Map.prototype method called on incompatible receiver",
));
}
let inner_val = map_obj.get_prop_at(MAP_PROP_INDEX);
if !inner_val.is_object() {
return Err(crate::builtins::error::create_type_error(vm, "Map internal state invalid"));
}
// SAFETY: inner_val holds the raw pointer written by `alloc_map` as
// `JsValue::object(Box::into_raw(Box::new(IndexMap::new())) as *mut u8)`.
// The pointer is a valid, heap-allocated `Box<IndexMap<SetKey, JsValue>>` and is
// never freed during the Map object's lifetime (see module safety contract above).
// SAFETY: native_data holds the raw pointer written by `alloc_map`.
// The pointer is a valid, heap-allocated `Box<IndexMap<SetKey, JsValue>>`.
// Alignment: IndexMap requires at most 8-byte alignment; the global allocator satisfies this.
let inner_ptr = inner_val.as_js_object_ptr() as *mut indexmap::IndexMap<SetKey, JsValue>;
let inner_ptr = map_obj.native_data() as *mut MapInner;
if inner_ptr.is_null() {
return Err(crate::builtins::error::create_type_error(vm, "Map internal state invalid"));
}
Ok(inner_ptr)
}
/// Allocate a new `IndexMap` on the global heap.
/// The caller is responsible for storing the pointer in the Map `JsObject` at `MAP_PROP_INDEX`
/// via `JsValue::object(ptr as *mut u8)` and ensuring it is never double-freed.
fn new_map_inner() -> *mut indexmap::IndexMap<SetKey, JsValue> {
Box::into_raw(Box::new(indexmap::IndexMap::new()))
fn new_map_inner() -> *mut MapInner {
Box::into_raw(Box::new(MapInner::new()))
}
fn alloc_map(vm: &mut Vm) -> *mut JsObject {
let map_proto = vm.session().builtin_world().map_proto.as_ptr() as *mut JsObject;
let mut obj = JsObject::new_empty(EMPTY_SHAPE_ID, JsValue::from_js_object(map_proto));
obj.set_map(true);
// SAFETY: new_map_inner() returns a valid Box<IndexMap> pointer; stored as a JsValue
// object tag so get_map_inner() can retrieve and cast it back (same provenance).
let inner = new_map_inner();
obj.set_prop_at(MAP_PROP_INDEX, JsValue::object(inner as *mut u8));
obj.set_native_data(inner as *mut u8);
vm.epoch().alloc(obj)
}
pub(crate) fn map_native_edges(obj: &JsObject) -> Vec<JsValue> {
if !obj.is_map() {
return Vec::new();
}
let inner = obj.native_data() as *const MapInner;
if inner.is_null() {
return Vec::new();
}
unsafe {
(*inner)
.iter()
.flat_map(|(key, value)| [key.0, *value])
.filter(|value| value.is_object())
.collect()
}
}
pub(crate) fn clone_map_native_with_rewrite<F>(src: &JsObject, dst: &mut JsObject, mut rewrite: F)
where
F: FnMut(JsValue) -> JsValue,
{
if !src.is_map() {
return;
}
let inner = src.native_data() as *const MapInner;
if inner.is_null() {
dst.set_native_data(std::ptr::null_mut());
return;
}
let mut cloned = MapInner::new();
unsafe {
for (key, value) in (*inner).iter() {
let new_key = if key.0.is_object() { SetKey(rewrite(key.0)) } else { *key };
let new_value = if value.is_object() { rewrite(*value) } else { *value };
cloned.insert(new_key, new_value);
}
}
dst.set_native_data(Box::into_raw(Box::new(cloned)) as *mut u8);
}
pub(crate) fn rewrite_map_native<F>(obj: &mut JsObject, mut rewrite: F)
where
F: FnMut(JsValue) -> JsValue,
{
if !obj.is_map() {
return;
}
let inner = obj.native_data() as *mut MapInner;
if inner.is_null() {
return;
}
unsafe {
let mut rewritten = MapInner::with_capacity((*inner).len());
for (key, value) in (*inner).iter() {
let new_key = if key.0.is_object() { SetKey(rewrite(key.0)) } else { *key };
let new_value = if value.is_object() { rewrite(*value) } else { *value };
rewritten.insert(new_key, new_value);
}
*inner = rewritten;
}
}
pub(crate) fn drop_map_native(obj: &mut JsObject) -> u64 {
if !obj.is_map() {
return 0;
}
let inner = obj.native_data() as *mut MapInner;
if inner.is_null() {
return 0;
}
unsafe {
let boxed = Box::from_raw(inner);
let bytes = std::mem::size_of::<MapInner>() + boxed.capacity() * std::mem::size_of::<(SetKey, JsValue)>();
drop(boxed);
obj.set_native_data(std::ptr::null_mut());
bytes as u64
}
}
pub fn map_constructor(vm: &mut Vm, _args: &[u8]) -> NativeResult {
let map_obj = alloc_map(vm);
NativeResult::Ok(JsValue::from_js_object(map_obj))
+81 -21
View File
@@ -7,8 +7,6 @@ use oxide_types::value::JsValue;
use crate::native::NativeResult;
use crate::vm::Vm;
const SET_PROP_INDEX: u8 = 0;
macro_rules! native_try {
($expr:expr) => {
match $expr {
@@ -59,7 +57,9 @@ impl Hash for SetKey {
}
}
/// Retrieve the `IndexSet` pointer stored at `SET_PROP_INDEX` of a Set object.
type SetInner = indexmap::IndexSet<SetKey>;
/// Retrieve the `IndexSet` pointer stored in a Set object's native-data slot.
///
/// # Safety contract maintained by callers
///
@@ -68,7 +68,7 @@ impl Hash for SetKey {
/// native builtin is executing. The `Box<IndexSet>` is allocated in `new_set_inner()`
/// and is never freed during the Set's lifetime. Only one live `*mut` alias exists per
/// Set object at a time because native calls are single-threaded.
fn get_set_inner(vm: &mut Vm, this_val: JsValue) -> Result<*mut indexmap::IndexSet<SetKey>, JsValue> {
fn get_set_inner(vm: &mut Vm, this_val: JsValue) -> Result<*mut SetInner, JsValue> {
if !this_val.is_object() {
return Err(crate::builtins::error::create_type_error(vm, "called on non-Set object"));
}
@@ -85,40 +85,100 @@ fn get_set_inner(vm: &mut Vm, this_val: JsValue) -> Result<*mut indexmap::IndexS
"Set.prototype.add called on incompatible receiver",
));
}
let inner_val = set_obj.get_prop_at(SET_PROP_INDEX);
if !inner_val.is_object() {
return Err(crate::builtins::error::create_type_error(vm, "Set internal state invalid"));
}
// SAFETY: inner_val holds the raw pointer written by `alloc_set` as
// `JsValue::object(Box::into_raw(Box::new(IndexSet::new())) as *mut u8)`.
// The pointer is a valid, heap-allocated `Box<IndexSet<SetKey>>` and is never freed
// during the Set object's lifetime (see module safety contract above).
// SAFETY: native_data holds the raw pointer written by `alloc_set`.
// The pointer is a valid, heap-allocated `Box<IndexSet<SetKey>>`.
// Alignment: IndexSet requires at most 8-byte alignment; the global allocator satisfies this.
let inner_ptr = inner_val.as_js_object_ptr() as *mut indexmap::IndexSet<SetKey>;
let inner_ptr = set_obj.native_data() as *mut SetInner;
if inner_ptr.is_null() {
return Err(crate::builtins::error::create_type_error(vm, "Set internal state invalid"));
}
Ok(inner_ptr)
}
/// Allocate a new `IndexSet` on the global heap.
/// The caller is responsible for storing the pointer in the Set `JsObject` at `SET_PROP_INDEX`
/// via `JsValue::object(ptr as *mut u8)` and ensuring it is never double-freed.
fn new_set_inner() -> *mut indexmap::IndexSet<SetKey> {
Box::into_raw(Box::new(indexmap::IndexSet::new()))
fn new_set_inner() -> *mut SetInner {
Box::into_raw(Box::new(SetInner::new()))
}
fn alloc_set(vm: &mut Vm) -> *mut JsObject {
let set_proto = vm.session().builtin_world().set_proto.as_ptr() as *mut JsObject;
let mut obj = JsObject::new_empty(EMPTY_SHAPE_ID, JsValue::from_js_object(set_proto));
obj.set_set(true);
// SAFETY: new_set_inner() returns a valid Box<IndexSet> pointer; stored as a JsValue
// object tag so get_set_inner() can retrieve and cast it back (same provenance).
let inner = new_set_inner();
obj.set_prop_at(SET_PROP_INDEX, JsValue::object(inner as *mut u8));
obj.set_native_data(inner as *mut u8);
vm.epoch().alloc(obj)
}
pub(crate) fn set_native_edges(obj: &JsObject) -> Vec<JsValue> {
if !obj.is_set() {
return Vec::new();
}
let inner = obj.native_data() as *const SetInner;
if inner.is_null() {
return Vec::new();
}
unsafe { (*inner).iter().map(|key| key.0).filter(|value| value.is_object()).collect() }
}
pub(crate) fn clone_set_native_with_rewrite<F>(src: &JsObject, dst: &mut JsObject, mut rewrite: F)
where
F: FnMut(JsValue) -> JsValue,
{
if !src.is_set() {
return;
}
let inner = src.native_data() as *const SetInner;
if inner.is_null() {
dst.set_native_data(std::ptr::null_mut());
return;
}
let mut cloned = SetInner::new();
unsafe {
for key in (*inner).iter() {
let new_key = if key.0.is_object() { SetKey(rewrite(key.0)) } else { *key };
cloned.insert(new_key);
}
}
dst.set_native_data(Box::into_raw(Box::new(cloned)) as *mut u8);
}
pub(crate) fn rewrite_set_native<F>(obj: &mut JsObject, mut rewrite: F)
where
F: FnMut(JsValue) -> JsValue,
{
if !obj.is_set() {
return;
}
let inner = obj.native_data() as *mut SetInner;
if inner.is_null() {
return;
}
unsafe {
let mut rewritten = SetInner::with_capacity((*inner).len());
for key in (*inner).iter() {
let new_key = if key.0.is_object() { SetKey(rewrite(key.0)) } else { *key };
rewritten.insert(new_key);
}
*inner = rewritten;
}
}
pub(crate) fn drop_set_native(obj: &mut JsObject) -> u64 {
if !obj.is_set() {
return 0;
}
let inner = obj.native_data() as *mut SetInner;
if inner.is_null() {
return 0;
}
unsafe {
let boxed = Box::from_raw(inner);
let bytes = std::mem::size_of::<SetInner>() + boxed.capacity() * std::mem::size_of::<SetKey>();
drop(boxed);
obj.set_native_data(std::ptr::null_mut());
bytes as u64
}
}
pub fn set_constructor(vm: &mut Vm, _args: &[u8]) -> NativeResult {
let set_obj = alloc_set(vm);
NativeResult::Ok(JsValue::from_js_object(set_obj))
+10
View File
@@ -3,6 +3,7 @@ use std::collections::HashMap;
use oxide_types::object::JsObject;
use oxide_types::value::JsValue;
use crate::builtins::{map, set};
use crate::vm::Vm;
impl Vm {
@@ -41,6 +42,15 @@ impl Vm {
let dst_ref = unsafe { &mut *dst };
dst_ref.rewrite_object_values(|value| self.promote_value_if_epoch_object(value, forwarding));
if src_ref.is_map() {
map::clone_map_native_with_rewrite(src_ref, dst_ref, |value| {
self.promote_value_if_epoch_object(value, forwarding)
});
} else if src_ref.is_set() {
set::clone_set_native_with_rewrite(src_ref, dst_ref, |value| {
self.promote_value_if_epoch_object(value, forwarding)
});
}
dst
}
+96 -1
View File
@@ -5,6 +5,7 @@ use std::time::Instant;
use oxide_types::object::{JsObject, PropMetaEntry};
use oxide_types::value::JsValue;
use crate::builtins::{map, set};
use crate::vm::Vm;
pub struct SessionGc {
@@ -80,6 +81,12 @@ impl SessionGc {
if obj.home_object().is_object() {
edges.push(obj.home_object());
}
if obj.is_map() {
edges.extend(map::map_native_edges(obj));
}
if obj.is_set() {
edges.extend(set::set_native_edges(obj));
}
edges
}
@@ -161,6 +168,9 @@ impl SessionGc {
std::mem::drop(vec);
}
freed_bytes += map::drop_map_native(obj);
freed_bytes += set::drop_set_native(obj);
freed_bytes
}
}
@@ -186,8 +196,15 @@ impl SessionGc {
let is_live = unsafe { (*old_ptr).is_gc_marked() };
if is_live {
survivors += 1;
let clone = unsafe { &*old_ptr }.clone_for_session_epoch();
let old_ref = unsafe { &*old_ptr };
let clone = old_ref.clone_for_session_epoch();
let new_ptr = new_arena.alloc(clone) as *mut JsObject;
let new_ref = unsafe { &mut *new_ptr };
if old_ref.is_map() {
map::clone_map_native_with_rewrite(old_ref, new_ref, |value| value);
} else if old_ref.is_set() {
set::clone_set_native_with_rewrite(old_ref, new_ref, |value| value);
}
forwarding.insert(old_ptr, new_ptr);
freed_bytes += Self::drop_session_object_heap_data(old_ptr);
} else {
@@ -208,6 +225,11 @@ impl SessionGc {
}
value
});
if obj.is_map() {
map::rewrite_map_native(obj, |value| rewrite_forwarded_value(value, &forwarding));
} else if obj.is_set() {
set::rewrite_set_native(obj, |value| rewrite_forwarded_value(value, &forwarding));
}
}
rewrite_vm_roots(vm, &forwarding);
@@ -353,6 +375,8 @@ mod tests {
use oxide_types::object::JsObject;
use super::*;
use crate::builtins::{map, set};
use crate::native::NativeResult;
use crate::vm::{CallFrame, FrameContinuation};
fn plain_object(vm: &mut Vm) -> *mut JsObject {
@@ -480,6 +504,14 @@ mod tests {
Vm::with_kernel_core(core)
}
fn native_ok(result: NativeResult) -> JsValue {
match result {
NativeResult::Ok(value) => value,
NativeResult::Err(err) => panic!("native error: {err}"),
NativeResult::TailCall { .. } => panic!("unexpected native bytecode call"),
}
}
#[test]
fn reset_maybe_collect_collects_after_threshold() {
let mut vm = vm_with_low_threshold();
@@ -536,4 +568,67 @@ mod tests {
assert!(!std::ptr::eq(new_value.as_js_object_ptr(), old_ptr));
assert_eq!(unsafe { (*new_value.as_js_object_ptr()).get_prop_at(0) }, JsValue::int(42));
}
#[test]
fn map_native_storage_is_not_a_normal_object_edge() {
let mut vm = Vm::new();
let map_value = native_ok(map::map_constructor(&mut vm, &[]));
let map_obj = unsafe { &*map_value.as_js_object_ptr() };
let native_ptr = map_obj.native_data() as *mut JsObject;
assert!(map_obj.hash_props_vec().is_none());
assert!(!map_obj.native_data().is_null());
assert!(!SessionGc::new()
.object_edges(map_obj)
.iter()
.any(|edge| std::ptr::eq(edge.as_js_object_ptr(), native_ptr)));
}
#[test]
fn session_gc_traces_map_object_key_and_value() {
let mut vm = vm_with_low_threshold();
let map_value = native_ok(map::map_constructor(&mut vm, &[]));
let key = JsValue::from_js_object(plain_object(&mut vm));
let value = JsValue::from_js_object(plain_object(&mut vm));
vm.regs[0] = map_value;
vm.regs[1] = key;
vm.regs[2] = value;
native_ok(map::map_set(&mut vm, &[0, 1, 2]));
let map_session = vm.promote_object(map_value.as_js_object_ptr());
vm.regs.fill(JsValue::undefined());
vm.regs[0] = JsValue::from_js_object(map_session);
let mut gc = std::mem::replace(&mut vm.session_gc, SessionGc::new());
gc.collect(&mut vm);
vm.session_gc = gc;
let live_map = unsafe { &*vm.regs[0].as_js_object_ptr() };
let edges = map::map_native_edges(live_map);
assert_eq!(edges.len(), 2);
assert!(edges.iter().all(|value| vm.is_session_ptr(value.as_js_object_ptr())));
assert_eq!(vm.session_object_ptrs.len(), 3);
}
#[test]
fn session_gc_traces_set_object_key() {
let mut vm = vm_with_low_threshold();
let set_value = native_ok(set::set_constructor(&mut vm, &[]));
let key = JsValue::from_js_object(plain_object(&mut vm));
vm.regs[0] = set_value;
vm.regs[1] = key;
native_ok(set::set_add(&mut vm, &[0, 1]));
let set_session = vm.promote_object(set_value.as_js_object_ptr());
vm.regs.fill(JsValue::undefined());
vm.regs[0] = JsValue::from_js_object(set_session);
let mut gc = std::mem::replace(&mut vm.session_gc, SessionGc::new());
gc.collect(&mut vm);
vm.session_gc = gc;
let live_set = unsafe { &*vm.regs[0].as_js_object_ptr() };
let edges = set::set_native_edges(live_set);
assert_eq!(edges.len(), 1);
assert!(vm.is_session_ptr(edges[0].as_js_object_ptr()));
assert_eq!(vm.session_object_ptrs.len(), 2);
}
}