总所周知在MRC时代，经常会用到AutoreleasePool帮助开发者管理内存；但在ARC时，这部分的内存管理操作却被系统所代替了，不再手动的去release或者autorelease等操作了，不过AutoreleasePool仍然在背后发挥着极大的作用，今天我们就从以下几个方面来探索下AutoreleasePool
- 1.什么是
autorelease&autoreleasePool - 2.
autoreleasePool原理解析 - 3.线程,
Runloop以及autoreleasePool
1.什么是autorelease& autoreleasePool
对象执行 autorelease 方法或者直接在 autoreleasePool 中创建对象，会将对象添加到 autoreleasePool 中，当自动释放池销毁的时候，会对所有对象做 release 操作
@interface MGMan : NSObject
+ (instancetype)object;
@end
@implementation MGMan
- (void)dealloc {
NSLog(@"MGMan dealloc");
}
//这个方法返回autorelease对象
+ (instancetype)object {
return [[MGMan alloc] init];
}
@end
- (void)testARCRule {
__weak id tmp1 = nil;
__weak id tmp2 = nil;
{
MGMan *man1 = [MGMan object];
MGMan *man2 = [[MGMan alloc] init];
tmp1 = man1;
tmp2 = man2;
}
NSLog(@"tmp1 = %@", tmp1);
NSLog(@"tmp2 = %@", tmp2);
}
//打印结果
//MGMan dealloc
//tmp1 = <MGMan: 0x600001188230>
//tmp2 = (null)
//MGMan dealloc
在作用域外，声明了两个弱引用对象tmp1和tmp2，在作用域内，分别对tmp1和tmp2赋值，并在作用域外分别打印tmp1和tmp2，为什么tmp1和tmp2的打印结果不一样呢，man1和man2的释放时机也不一样
-
[MGMan object];声明的对象并不会立即释放，man1这个对象被注册进了autoreleasePool中，man1的释放周期和autoreleasePool的释放周期相关，和作用域关系不大 -
[[MGMan alloc] init];声明的对象，不会加入到autoreleasePool中，并不是一个autorelease对象
⚠️注意
ARC的规则：alloc/new/copy/mutableCopy开头的方法返回的对象不是 autorelease 对象
@interface MGMan : NSObject
+ (instancetype)object;
+ (NSString *)newString;
+ (NSString *)allocString;
+ (NSString *)copyString;
+ (NSString *)initString;
+ (NSString *)helloString;
+ (NSString *)createString;
@end
#import "MGMan.h"
@implementation MGMan
- (void)dealloc {
NSLog(@"MGMan dealloc");
}
//这个方法返回autorelease对象
+ (instancetype)object {
return [[MGMan alloc] init];
}
+ (NSString *)newString {
return [[NSString alloc] initWithCString:"Hello Objective-c" encoding:NSUTF8StringEncoding];
}
+ (NSString *)allocString {
return [[NSString alloc] initWithCString:"Hello Objective-c" encoding:NSUTF8StringEncoding];
}
+ (NSString *)copyString {
//帮我们插入autorelease
return [[NSString alloc] initWithCString:"Hello Objective-c" encoding:NSUTF8StringEncoding];
}
+ (NSString *)initString {
return [[NSString alloc] initWithCString:"Hello Objective-c" encoding:NSUTF8StringEncoding];
}
+ (NSString *)helloString {
return [[NSString alloc] initWithCString:"Hello Objective-c" encoding:NSUTF8StringEncoding];
}
+ (NSString *)createString {
return [[NSString alloc] initWithCString:"Hello Objective-c" encoding:NSUTF8StringEncoding];
}
- (void)testARCRule {
@autoreleasepool {
//Tagged Pointer
//LLVM 并不会发送autorelease消息
__weak NSString *tmp1 = [MGMan newString];
__weak NSString *tmp2 = [MGMan allocString];
__weak NSString *tmp3 = [MGMan copyString];
NSLog(@"%@", tmp1);
NSLog(@"%@", tmp2);
NSLog(@"%@", tmp3);
__weak NSString *tmp4 = [MGMan initString];
__weak NSString *tmp5 = [MGMan helloString];
__weak NSString *tmp6 = [MGMan createString];
NSLog(@"%@", tmp4);
NSLog(@"%@", tmp5);
NSLog(@"%@", tmp6);
}
}
//打印结果
//(null)
//(null)
//(null)
//Hello Objective-c
//Hello Objective-c
//Hello Objective-c
上面这个例子验证了ARC的规则：alloc/new/copy/mutableCopy 开头的方法返回的对象不是 autorelease 对象
autoreleasePool的使用
- 1.在ARC和MRC环境下
@autoreleasePool {} - 2.在MRC环境下
NSAutoreleasePool
// 1、生成一个NSAutoreleasePool对象
NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
// 2、调用autorelease方法
id object = [[NSObject alloc] init];
[object autorelease];
// 3、对象销毁
[pool drain];
@autoreleasepool {
//LLVM
id object = [[NSObject alloc] init];
}
- 1.生成一个NSAutoreleasePool对象
- 2.调用autorelease方法
- 3.对象销毁
在ARC环境下其实是我们的系统自动帮我们调用了autorelease,我们可以结合源码GNUStep来看下
#ifndef AUTORELEASE
/**
* Basic autorelease operation ... calls [NSObject-autorelease]<br />
* Does nothing when ARC is in use.
*/
#define AUTORELEASE(object) [(id)(object) autorelease]
#endif
+ (id) arrayWithCapacity: (NSUInteger)numItems
{
return AUTORELEASE([[self allocWithZone: NSDefaultMallocZone()]
initWithCapacity: numItems]);
}
当我们初始化的时候，系统会自动帮我们调用AUTORELEASE宏定义去包裹
2.autoreleasePool原理解析
新创建一个项目
#import <Foundation/Foundation.h>
int main(int argc, const char * argv[]) {
@autoreleasepool {
// insert code here...
NSLog(@"Hello, World!");
}
return 0;
}
我们clang下，看下@autoreleasepool{}具体做了什么
通过clang我们得到以下代码
struct __AtAutoreleasePool {
__AtAutoreleasePool() {atautoreleasepoolobj = objc_autoreleasePoolPush();}
~__AtAutoreleasePool() {objc_autoreleasePoolPop(atautoreleasepoolobj);}
void * atautoreleasepoolobj;
};
int main(int argc, const char * argv[]) {
/* @autoreleasepool */ { __AtAutoreleasePool __autoreleasepool;
NSLog((NSString *)&__NSConstantStringImpl__var_folders_fc_dm6qr3kj7mbfs5kj1td4_f1c0000gn_T_main_3eea55_mi_0);
}
return 0;
}
static struct IMAGE_INFO { unsigned version; unsigned flag; } _OBJC_IMAGE_INFO = { 0, 2 };
- 1.声明了一个
__autoreleasepool私有对象 - 2.
__AtAutoreleasePool是一个结构体 - 3.
__AtAutoreleasePool结构体中包含了一个构造函数和一个析构函数 - 4.构造函数中调用了
objc_autoreleasePoolPush ()方法，析构函数中调用了objc_autoreleasePoolPop()方法
接下来我们来探究下objc_autoreleasePoolPush ()和objc_autoreleasePoolPop()方法 源码下载地址
我这里使用的是objc4-866.9
objc_autoreleasePoolPush ()方法
void *
objc_autoreleasePoolPush(void)
{
return AutoreleasePoolPage::push();
}
返回AutoreleasePoolPage,并且调用了push()函数
首先来看下AutoreleasePoolPage
class AutoreleasePoolPage : private AutoreleasePoolPageData
struct AutoreleasePoolPageData
{
#if SUPPORT_AUTORELEASEPOOL_DEDUP_PTRS
struct AutoreleasePoolEntry {
uintptr_t ptr: 48;
uintptr_t count: 16;
static const uintptr_t maxCount = 65535; // 2^16 - 1
};
static_assert((AutoreleasePoolEntry){ .ptr = OBJC_VM_MAX_ADDRESS }.ptr == OBJC_VM_MAX_ADDRESS, "OBJC_VM_MAX_ADDRESS doesn't fit into AutoreleasePoolEntry::ptr!");
#endif
magic_t const magic;
__unsafe_unretained id *next;
objc_thread_t const thread;
AutoreleasePoolPage * const parent;
AutoreleasePoolPage *child;
uint32_t const depth;
uint32_t hiwat;
AutoreleasePoolPageData(__unsafe_unretained id* _next, objc_thread_t _thread, AutoreleasePoolPage* _parent, uint32_t _depth, uint32_t _hiwat)
: magic(), next(_next), thread(_thread),
parent(_parent), child(nil),
depth(_depth), hiwat(_hiwat)
{
}
}
-
AutoreleasePoolPage是一个类，并且继承自AutoreleasePoolPageData
-
- 2.当我们看到
parent与child时，就能发现我们的自动释放池它是一个双向链表结构，parent指向上一个节点链表，child指向下一个节点链表
接着我们再来看下push()方法
static inline void *push()
{
ReturnAutoreleaseInfo info = getReturnAutoreleaseInfo();
moveTLSAutoreleaseToPool(info);
id *dest;
if (slowpath(DebugPoolAllocation)) {
// Each autorelease pool starts on a new pool page.
dest = autoreleaseNewPage(POOL_BOUNDARY);
} else {
dest = autoreleaseFast(POOL_BOUNDARY);
}
ASSERT(dest == (id *)EMPTY_POOL_PLACEHOLDER || *dest == POOL_BOUNDARY);
return dest;
}
- 1.
push()是一个静态的内链函数 - 2.
if (slowpath(DebugPoolAllocation))首先判断当前的自动释放池是否创建了，如果没有调用autoreleaseNewPage()创建，如果已经创建过则调用autoreleaseFast ()并传进去一个POOL_BOUNDARY哨兵对象（边界） - 3.每一个
autoreleaseNewPage只有一个哨兵对象
接着我们再来看下autoreleaseFast()方法
static inline id *autoreleaseFast(id obj)
{
AutoreleasePoolPage *page = hotPage();
if (page && !page->full()) {
return page->add(obj);
} else if (page) {
return autoreleaseFullPage(obj, page);
} else {
return autoreleaseNoPage(obj);
}
}
-
AutoreleasePoolPage *page = hotPage();通过hotPage()方法获取AutoreleasePoolPage对象
-
-
if (page && !page->full())判断page是否存在，以及是否满了
-
- 3.如果
page存在并且没满则page->add(obj);链表插入对象 - 4.如果
page满了则调用autoreleaseFullPage(obj, page); - 5.如果
page不存在则调用autoreleaseNoPage(obj)
接着我们再来看下hotPage()方法
static tls_direct(AutoreleasePoolPage *, tls_key::autorelease_pool, HotPageDealloc)
hotPage_;
static inline AutoreleasePoolPage *hotPage()
{
AutoreleasePoolPage *result = hotPage_;
if (result == EMPTY_POOL_PLACEHOLDER) return nil;
if (result) result->fastcheck();
return result;
}
从上面的代码中我们不难发现，
- 1.
AutoreleasePoolPage是从当前线程的私有的局部空间取出来static tls_direct(AutoreleasePoolPage *, tls_key::autorelease_pool, HotPageDealloc) hotPage_;, - 通过
tls_key值取出AutoreleasePoolPage
- 通过
- 3.每一个线程都有一个与之对应的
autoreleasePool - 4.
tls只有当前线程才能访问到
然后我们来看下page->add(obj)是怎么把obj插入到链表中去的
id *add(id obj)
{
ASSERT(!full());
unprotect();
id *ret;
#if SUPPORT_AUTORELEASEPOOL_DEDUP_PTRS
if (!DisableAutoreleaseCoalescing || !DisableAutoreleaseCoalescingLRU) {
if (!DisableAutoreleaseCoalescingLRU) {
if (!empty() && (obj != POOL_BOUNDARY)) {
AutoreleasePoolEntry *topEntry = (AutoreleasePoolEntry *)next - 1;
for (uintptr_t offset = 0; offset < 4; offset++) {
AutoreleasePoolEntry *offsetEntry = topEntry - offset;
if (offsetEntry <= (AutoreleasePoolEntry*)begin() || *(id *)offsetEntry == POOL_BOUNDARY) {
break;
}
if (offsetEntry->ptr == (uintptr_t)obj && offsetEntry->count < AutoreleasePoolEntry::maxCount) {
if (offset > 0) {
AutoreleasePoolEntry found = *offsetEntry;
memmove(offsetEntry, offsetEntry + 1, offset * sizeof(*offsetEntry));
*topEntry = found;
}
topEntry->count++;
ret = (id *)topEntry; // need to reset ret
goto done;
}
}
}
} else {
if (!empty() && (obj != POOL_BOUNDARY)) {
AutoreleasePoolEntry *prevEntry = (AutoreleasePoolEntry *)next - 1;
if (prevEntry->ptr == (uintptr_t)obj && prevEntry->count < AutoreleasePoolEntry::maxCount) {
prevEntry->count++;
ret = (id *)prevEntry; // need to reset ret
goto done;
}
}
}
}
#endif
ret = next; // faster than `return next-1` because of aliasing
*next++ = obj;
#if SUPPORT_AUTORELEASEPOOL_DEDUP_PTRS
// Make sure obj fits in the bits available for it
ASSERT(((AutoreleasePoolEntry *)ret)->ptr == (uintptr_t)obj);
#endif
done:
protect();
return ret;
}
如图所示链表有一个起始位置begin(),begin()指向的位置为整个存储的空间减去page数据结构的空间，同时也有一个结束位置end(), end则是this指针加上整个链表的长度，当前有个指针叫next,next指针指向的是当前空地址的首地址,如果当前链表是空的，那么next指针指向的是begin()位置，当前如果向链表中插入一个对象obj，那么此时next指针则会*next++ = obj;指向下一个地址，同时插入obj之前会存放一个POOL_BOUNDARY(哨兵对象)，并且返回ret及next上一个指向的位置，如果链表存满后，则会重新声明新的链表继续存放对象，每张链表的容积为2的14次方(16384)字节(以前老版本为4096）
objc_autoreleasePoolPop ()方法
NEVER_INLINE
void
objc_autoreleasePoolPop(void *ctxt)
{
AutoreleasePoolPage::pop(ctxt);
}
在前文的__AtAutoreleasePool结构体中我们得到，构造函数会调用objc_autoreleasePoolPush并返回atautoreleasepoolobj,这个atautoreleasepoolobj其实就是POOL_BOUNDARY的地址，当调用objc_autoreleasePoolPop又会把POOL_BOUNDARY地址传进去
static inline void
pop(void *token)
{
// We may have an object in the ReturnAutorelease TLS when the pool is
// otherwise empty. Release that first before checking for an empty pool
// so we don't return prematurely. Loop in case the release placed a new
// object in the TLS.
while (releaseReturnAutoreleaseTLS())
;
AutoreleasePoolPage *page;
id *stop;
if (token == (void*)EMPTY_POOL_PLACEHOLDER) {
// Popping the top-level placeholder pool.
page = hotPage();
if (!page) {
// Pool was never used. Clear the placeholder.
return setHotPage(nil);
}
// Pool was used. Pop its contents normally.
// Pool pages remain allocated for re-use as usual.
page = coldPage();
token = page->begin();
} else {
page = pageForPointer(token);
}
stop = (id *)token;
if (*stop != POOL_BOUNDARY) {
if (stop == page->begin() && !page->parent) {
// Start of coldest page may correctly not be POOL_BOUNDARY:
// 1. top-level pool is popped, leaving the cold page in place
// 2. an object is autoreleased with no pool
} else {
// Error. For bincompat purposes this is not
// fatal in executables built with old SDKs.
return badPop(token);
}
}
if (slowpath(PrintPoolHiwat || DebugPoolAllocation || DebugMissingPools)) {
return popPageDebug(token, page, stop);
}
return popPage<false>(token, page, stop);
}
- 1.
pop()是一个静态的内链函数 -
if (token == (void*)EMPTY_POOL_PLACEHOLDER)对token及对POOL_BOUNDARY进行判断
-
- 调用
page = pageForPointer(token);通过token地址获取page
- 调用
- 4.通过
token,page,stop调用popPage<false>(token, page, stop);
template<bool allowDebug>
static void
popPage(void *token, AutoreleasePoolPage *page, id *stop)
{
if (allowDebug && PrintPoolHiwat) printHiwat();
page->releaseUntil(stop);
// memory: delete empty children
if (allowDebug && DebugPoolAllocation && page->empty()) {
// special case: delete everything during page-per-pool debugging
AutoreleasePoolPage *parent = page->parent;
page->kill();
setHotPage(parent);
} else if (allowDebug && DebugMissingPools && page->empty() && !page->parent) {
// special case: delete everything for pop(top)
// when debugging missing autorelease pools
page->kill();
setHotPage(nil);
} else if (page->child) {
// hysteresis: keep one empty child if page is more than half full
if (page->lessThanHalfFull()) {
page->child->kill();
}
else if (page->child->child) {
page->child->child->kill();
}
}
}
通过调用page->releaseUntil(stop);对当前链表里的对象进行release操作
void releaseUntil(id *stop)
{
// Not recursive: we don't want to blow out the stack
// if a thread accumulates a stupendous amount of garbage
do {
while (this->next != stop) {
// Restart from hotPage() every time, in case -release
// autoreleased more objects
AutoreleasePoolPage *page = hotPage();
// fixme I think this `while` can be `if`, but I can't prove it
while (page->empty()) {
page = page->parent;
setHotPage(page);
}
page->unprotect();
#if SUPPORT_AUTORELEASEPOOL_DEDUP_PTRS
AutoreleasePoolEntry* entry = (AutoreleasePoolEntry*) --page->next;
// create an obj with the zeroed out top byte and release that
id obj = (id)entry->ptr;
int count = (int)entry->count; // grab these before memset
#else
id obj = *--page->next;
#endif
memset((void*)page->next, SCRIBBLE, sizeof(*page->next));
page->protect();
if (obj != POOL_BOUNDARY) {
#if SUPPORT_AUTORELEASEPOOL_DEDUP_PTRS
// release count+1 times since it is count of the additional
// autoreleases beyond the first one
for (int i = 0; i < count + 1; i++) {
objc_release(obj);
}
#else
objc_release(obj);
#endif
}
}
// Stale return autorelease info is conceptually autoreleased. If
// there is any, release the object in the info. If stale info is
// present, we have to loop in case it autoreleased more objects
// when it was released.
} while (releaseReturnAutoreleaseTLS());
setHotPage(this);
#if DEBUG
// we expect any children to be completely empty
for (AutoreleasePoolPage *page = child; page; page = page->child) {
ASSERT(page->empty());
}
#endif
}
- 1.
(this->next != stop)对next指针进行判断，当前的next指针指向的是最后加入释放池中的对象 -
while (this->next != stop) {}用来判断是否到达边界
-
-
id obj = *--page->next;找到当前要释放的对象
-
-
memset((void*)page->next, SCRIBBLE, sizeof(*page->next));置空当前内存空间
-
-
if (obj != POOL_BOUNDARY)如果不等于POOL_BOUNDARY那就调用runtime中的objc_release(obj);对当前对象进行释放
-
- 6.释放完对象后
next指针会移动到上一个对象
我们来看这样一个例子
- (void)viewDidLoad {
[super viewDidLoad];
@autoreleasepool {
MGMan *man1 = [[MGMan alloc] init];
@autoreleasepool {
MGMan *man2 = [[MGMan alloc] init];
@autoreleasepool {
MGMan *man3 = [[MGMan alloc] init];
};
};
};
}
- 1.
autoreleasepool支持嵌套使用 - 2.当执行
@autoreleasepool{}时，首先会调用构造函数__AtAutoreleasePool() {atautoreleasepoolobj = objc_autoreleasePoolPush();} - 3.生成
autoreleasePoolPage对象 - 4.插入
POOL_BOUNDARY - 5.执行
page->add(man) - 6.返回
POOL_BOUNDARY地址 - 7.离开作用域时，则优先释放man3这个对象
- 8.
objc_autoreleasePoolPop(POOL_BOUNDARY) - 9.移动
next指针，使之指向man2；
⚠️注意
@autoreleasepool{}嵌套使用不一定会创建新的释放池，创建新的释放池唯一标准，当前是否有释放池并且当前释放池是否已经存满
3.线程,Runloop以及autoreleasePool
- 1.每一个线程都会维护自己的自动释放池,每一个autoreleasePool对应一个线程
- 2.子线程在使用autorelease对象，懒加载出来一个autoreleasePoolPage
- 3.线程退出时
exit释放当前的资源(tls) - 4.Runloop退出的时候，进行pop操作
通过一下示例观察线程与AutoreleasePool关系
__weak id tmp = nil;
/**
线程和 AutoreleasePool
*/
- (void)test_one {
[NSThread detachNewThreadSelector:@selector(testAction) toTarget:self withObject:nil];
}
- (void)testAction {
__autoreleasing id test = [NSObject new];
NSLog(@"obj=%@", test);
tmp = test;
[[NSThread currentThread] setName:@"test runloop thread"];
NSLog(@"thread ending");
}
首先我们添加一个断点[NSThread exit]查看线程何时退出
我们观察下tmp 值的变化
tmp的值发生变化，同时也打印了thread ending说明任务已经执行完成，[NSThread exit]断点也触发了
线程退出后tmp也被释放了
这也正好契合我们开始的观点
