iOS底层探究:objc_msgSend流程分析的慢速查找

objc_msgSend慢速查找流程:系统先按照快速查找流程走的,如果快速的查找不到,然后进入到慢速查找流程里面。下面是慢速查找到到流程分析。

➢ 在配置好到objc4-781源码项目中导航面板区搜索objc_msgSend找到objc-msg-arm64.s文件,然后定位到 STATIC_ENTRY __objc_msgSend_uncached部分 ,通过查看得出__objc_msgSend_uncached汇编实现的核心MethodTableLookup,源码如下:

STATIC_ENTRY __objc_msgSend_uncached

UNWIND __objc_msgSend_uncached, FrameWithNoSaves

// THIS IS NOT A CALLABLE C FUNCTION

// Out-of-band p16 is the class to search

MethodTableLookup    // 即查询方法列表

TailCallFunctionPointer x17

END_ENTRY __objc_msgSend_uncached

➢ 通过查看MethodTableLookup的汇编实现,发现核心代码是_loopUpImpOrForward,下面是其源码内容:

.macro MethodTableLookup

// push frame

SignLR

stp fp, lr, [sp,#-16]!

mov fp, sp

// save parameter registers: x0..x8, q0..q7

sub sp, sp,#(10*8 + 8*16)

stp q0, q1, [sp,#(0*16)]

stp q2, q3, [sp,#(2*16)]

stp q4, q5, [sp,#(4*16)]

stp q6, q7, [sp,#(6*16)]

stp x0, x1, [sp,#(8*16+0*8)]

stp x2, x3, [sp,#(8*16+2*8)]

stp x4, x5, [sp,#(8*16+4*8)]

stp x6, x7, [sp,#(8*16+6*8)]

str x8,    [sp,#(8*16+8*8)]

// lookUpImpOrForward(obj, sel, cls, LOOKUP_INITIALIZE | LOOKUP_RESOLVER)

// receiver and selector already in x0 and x1

mov x2, x16

mov x3,#3

bl _lookUpImpOrForward

// IMP in x0

mov x17, x0

// restore registers and return

ldp q0, q1, [sp,#(0*16)]

ldp q2, q3, [sp,#(2*16)]

ldp q4, q5, [sp,#(4*16)]

ldp q6, q7, [sp,#(6*16)]

ldp x0, x1, [sp,#(8*16+0*8)]

ldp x2, x3, [sp,#(8*16+2*8)]

ldp x4, x5, [sp,#(8*16+4*8)]

ldp x6, x7, [sp,#(8*16+6*8)]

ldr x8,    [sp,#(8*16+8*8)]

mov sp, fp

ldp fp, lr, [sp],#16

AuthenticateLR

.endmacro

上面是通过汇编源码查看得出的,我们也可以通过汇编调试来验证上述部分,下面我们来一同看看汇编调试

➢ 我们以main为例:在类对象[person sayHello]这一行加一个断点,运行项目此时来到[person sayHello]断点这一行(如下图一),运用汇编调试方式(图二),在debug区汇编代码处就可以看到objc_msgSend(图三)这行代码,下面附上截图可以直观感受下:


图一
图二
图三

➢ 此时我们继续跟汇编继续走流程:汇编处objc_msgSend(如上图三)这行加一断点,然后断到该行,此时按住键盘control键+鼠标点住stepinto(如下图四),进入到objc_msgSend汇编(下图五):


图四
图五

➢ 此时来到来_objc_msgSend_uncached处,在此行加一断点,继续control+setpinto,此时来到了lookUpImpOrForward(如下图七)汇编区:


图七

注释:

① C/C++中调用汇编,去查找会变时,C/C++调用的方法前面需要多加一个下划线

② 汇编中调用C/C++方式时,去查找C/C++方法,需要将汇编调用到方法前面去掉一个下划线

慢速查找C/C++部分

➢ 根据汇编部分到提示,导航面板区全局搜索lookUpImpOrForward,进一步查找发现其藏到来objc-runtime-new.mm文件中,下面是其源码实现(c函数):

/***********************************************************************

* lookUpImpOrForward.

* The standard IMP lookup. 

* Without LOOKUP_INITIALIZE: tries to avoid +initialize (but sometimes fails)

* Without LOOKUP_CACHE: skips optimistic unlocked lookup (but uses cache elsewhere)

* Most callers should use LOOKUP_INITIALIZE and LOOKUP_CACHE

* inst is an instance of cls or a subclass thereof, or nil if none is known. 

*  If cls is an un-initialized metaclass then a non-nil inst is faster.

* May return _objc_msgForward_impcache. IMPs destined for external use 

*  must be converted to _objc_msgForward or _objc_msgForward_stret.

*  If you don't want forwarding at all, use LOOKUP_NIL.

**********************************************************************/

IMPlookUpImpOrForward(idinst,SELsel,Classcls,intbehavior)

{

    const IMP forward_imp = (IMP)_objc_msgForward_impcache;

    IMPimp =nil;

    ClasscurClass;

    runtimeLock.assertUnlocked();

    // Optimistic cache lookup

    if(fastpath(behavior &LOOKUP_CACHE)) {

        imp =cache_getImp(cls, sel);

        if(imp)gotodone_nolock;

    }

    // runtimeLock is held during isRealized and isInitialized checking

    // to prevent races against concurrent realization.

    // runtimeLock is held during method search to make

    // method-lookup + cache-fill atomic with respect to method addition.

    // Otherwise, a category could be added but ignored indefinitely because

    // the cache was re-filled with the old value after the cache flush on

    // behalf of the category.

    runtimeLock.lock();

    // We don't want people to be able to craft a binary blob that looks like

    // a class but really isn't one and do a CFI attack.

    //

    // To make these harder we want to make sure this is a class that was

    // either built into the binary or legitimately registered through

    // objc_duplicateClass, objc_initializeClassPair or objc_allocateClassPair.

    //

    // TODO: this check is quite costly during process startup.

    checkIsKnownClass(cls);

    if(slowpath(!cls->isRealized())) {

        cls =realizeClassMaybeSwiftAndLeaveLocked(cls, runtimeLock);

        // runtimeLock may have been dropped but is now locked again

    }

    if(slowpath((behavior &LOOKUP_INITIALIZE) && !cls->isInitialized())) {

        cls =initializeAndLeaveLocked(cls, inst,runtimeLock);

        // runtimeLock may have been dropped but is now locked again

        // If sel == initialize, class_initialize will send +initialize and 

        // then the messenger will send +initialize again after this 

        // procedure finishes. Of course, if this is not being called 

        // from the messenger then it won't happen. 2778172

    }

    runtimeLock.assertLocked();

    curClass = cls;

    // The code used to lookpu the class's cache again right after

    // we take the lock but for the vast majority of the cases

    // evidence shows this is a miss most of the time, hence a time loss.

    //

    // The only codepath calling into this without having performed some

    // kind of cache lookup is class_getInstanceMethod().

    for(unsignedattempts =unreasonableClassCount();;) {

        // curClass method list.

        Methodmeth =getMethodNoSuper_nolock(curClass, sel);

        if(meth) {

            imp = meth->imp;

            gotodone;

        }

        if(slowpath((curClass = curClass->superclass) ==nil)) {

            // No implementation found, and method resolver didn't help.

            // Use forwarding.

            imp = forward_imp;

            break;

        }

        // Halt if there is a cycle in the superclass chain.

        if(slowpath(--attempts ==0)) {

            _objc_fatal("Memory corruption in class list.");

        }

        // Superclass cache.

        imp =cache_getImp(curClass, sel); // 有问题???? cache_getImp - lookup - lookUpImpOrForward

        if(slowpath(imp == forward_imp)) {

            // Found a forward:: entry in a superclass.

            // Stop searching, but don't cache yet; call method

            // resolver for this class first.

            break;

        }

        if(fastpath(imp)) {

            // Found the method in a superclass. Cache it in this class.

            gotodone;

        }

    }

    // No implementation found. Try method resolver once.

    if(slowpath(behavior &LOOKUP_RESOLVER)) {

        behavior ^=LOOKUP_RESOLVER;

        returnresolveMethod_locked(inst, sel, cls, behavior);

    }

 done:

    log_and_fill_cache(cls, imp, sel, inst, curClass);

    runtimeLock.unlock();

 done_nolock:

    if(slowpath((behavior &LOOKUP_NIL) && imp == forward_imp)) {

        returnnil;

    }

    returnimp;

}

通过分析上面到源码实现得出整体到慢速流程图如下图八:


图八



➢下面是慢速流程图详细分析

⒈cache缓存中进行快速查找,如果找到则直接返回imp,否则进入第二步

⒉判断cls是否是已知类,如果不是就报错;类是否实现,如果没有则需要先实现,然后确定去父链,此时实例化的目的是为了确定父链、ro、rw等,方便方法后续数据的查找以及查找的循环;是否初始化,如果没有,则初始化

⒊for循环,按照类类继承链或者元类继承链的顺序查找:当前cls的方法列表中使用二分查找算法查找方法,如果找到,则进入cache写入流程,并返回imp,如果找不到就返回nil;当前cls被赋值为父类,如果父类为nil,则imp=消息转发,并中指递归,进入第四步;如果父类中存在循环,则报错终止循环;父类中查找方法(如果未找到,则直接返回nil,继续循环查找,如果找到,则返回imp,进行cache写入 ,方便后续的查找)

⒋判断是否执行过动态方法解析:如果没有,则执行动态方法解析;如果执行过一次动态方法解析,则走到消息转发流程

以上是方法的慢速查找流程,下面拓展下二分查找原理以及父类缓存查找详细步骤

➢ getMethodNoSuper_nolock方法:二分查找方法列表流程图以及核心源码如下:

二分方法查找流程图

/***********************************************************************

 * search_method_list_inline

 **********************************************************************/

ALWAYS_INLINE static method_t *

findMethodInSortedMethodList(SEL key, const method_list_t *list)

{

    ASSERT(list);

    constmethod_t*constfirst = &list->first;

    constmethod_t*base = first;

    constmethod_t*probe;

    uintptr_tkeyValue = (uintptr_t)key;

    uint32_tcount;


    for(count = list->count; count !=0; count >>=1) {

        probe = base + (count >>1);


        uintptr_tprobeValue = (uintptr_t)probe->name;


        if(keyValue == probeValue) {

            // `probe` is a match.

            // Rewind looking for the *first* occurrence of this value.

            // This is required for correct category overrides.

            while(probe > first && keyValue == (uintptr_t)probe[-1].name) {

                probe--;

            }

            return(method_t*)probe;

        }


        if(keyValue > probeValue) {

            base = probe +1;

            count--;

        }

    }


    return nil;

}


注释:

算法原理:从第一次开始查找,每次曲中间位置,与查找的key的value做对比,如果相等则需要排除分类方法,然后将查询到的位置发方法实现返回;如果不相等,则需要继续二分查找,如果循环至count=0还是没有找到,则直接返回nil,这里借用下网上的图片(图九):


图九

以查找LGPerson类的say666实例方法为例,其二分查找过程如图十:

图十


cache_getImp方法:父类缓存查找

cache_getImp方法是通过汇编_cache_getImp实现的,传入的$0是getImp,如下图十一:


图十一


⒜ 如果父类找到了方法实现,则CacheHit命中,直接返回imp

⒝ 如果父类缓存中没有找到方法实现,则CheckMiss或者JumpMiss,通过$0跳转至LGetImpMiss,直接返回nil


❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖

                                                                          总结

❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖❖

☞ 对于对象方法,即在类中查找,其慢速查找的父类链是:类➛父类➛根类➛nil

☞ 对于类方法,,即在元类中查找(类方法在元类中是以对象方法的形式存在的),其慢速查找的父类链是:元类➛根元类➛根类➛nil

☞ 如果快速查找和慢速查找都没有找到方法的实现,则需要走动态方法决议(苹果爸爸给了开发者后悔的补救措施)

➢  如果动态方法决议仍然没有找到,则进行消息转发(下一篇进行讲解,敬请期待)

你可能感兴趣的:(iOS底层探究:objc_msgSend流程分析的慢速查找)