在前面一篇文章Android系統進程間通信（IPC）機制Binder中的Server和Client獲得Service Manager接口之路中，介紹了在Android系統中Binder進程間通信機制中的Server角色是如何獲得Service Manager遠程接口的，即defaultServiceManager函數的實現。Server獲得了Service Manager遠程接口之后，就要把自己的Service添加到Service Manager中去，然后把自己啟動起來，等待Client的請求。本文將通過分析源代碼了解Server的啟動過程是怎么樣的。
本文通過一個具體的例子來說明Binder機制中Server的啟動過程。我們知道，在Android系統中，提供了多媒體播放的功能，這個功能是以服務的形式來提供的。這里，我們就通過分析MediaPlayerService的實現來了解Media Server的啟動過程。
首先，看一下MediaPlayerService的類圖，以便我們理解下面要描述的內容。
我們將要介紹的主角MediaPlayerService繼承于BnMediaPlayerService類，熟悉Binder機制的同學應該知道BnMediaPlayerService是一個Binder Native類，用來處理Client請求的。BnMediaPlayerService繼承于BnInterface<IMediaPlayerService>類，BnInterface是一個模板類，它定義在frameworks/base/include/binder/IInterface.h文件中：
template<typename INTERFACE> class BnInterface : public INTERFACE, public BBinder { public: virtual sp<IInterface> queryLocalInterface(const String16& _descriptor); virtual const String16& getInterfaceDescriptor() const; protected: virtual IBinder* onAsBinder(); }; 這里可以看出，BnMediaPlayerService實際是繼承了IMediaPlayerService和BBinder類。IMediaPlayerService和BBinder類又分別繼承了IInterface和IBinder類，IInterface和IBinder類又同時繼承了RefBase類。
實際上，BnMediaPlayerService并不是直接接收到Client處發送過來的請求，而是使用了IPCThreadState接收Client處發送過來的請求，而IPCThreadState又借助了ProcessState類來與Binder驅動程序交互。有關IPCThreadState和ProcessState的關系，可以參考上一篇文章Android系統進程間通信（IPC）機制Binder中的Server和Client獲得Service Manager接口之路，接下來也會有相應的描述。IPCThreadState接收到了Client處的請求后，就會調用BBinder類的transact函數，并傳入相關參數，BBinder類的transact函數最終調用BnMediaPlayerService類的onTransact函數，于是，就開始真正地處理Client的請求了。
了解了MediaPlayerService類結構之后，就要開始進入到本文的主題了。
首先，看看MediaPlayerService是如何啟動的。啟動MediaPlayerService的代碼位于frameworks/base/media/mediaserver/main_mediaserver.cpp文件中：
int main(int argc, char** argv) { sp<ProcessState> proc(ProcessState::self()); sp<IServiceManager> sm = defaultServiceManager(); LOGI("ServiceManager: %p", sm.get()); AudioFlinger::instantiate(); MediaPlayerService::instantiate(); CameraService::instantiate(); AudioPolicyService::instantiate(); ProcessState::self()->startThreadPool(); IPCThreadState::self()->joinThreadPool(); } 這里我們不關注AudioFlinger和CameraService相關的代碼。
先看下面這句代碼：
sp<ProcessState> proc(ProcessState::self());
這句代碼的作用是通過ProcessState::self()調用創建一個ProcessState實例。ProcessState::self()是ProcessState類的一個靜態成員變量，定義在frameworks/base/libs/binder/ProcessState.cpp文件中：
sp<ProcessState> ProcessState::self() { if (gProcess != NULL) return gProcess; AutoMutex _l(gProcessMutex); if (gProcess == NULL) gProcess = new ProcessState; return gProcess; } 這里可以看出，這個函數作用是返回一個全局唯一的ProcessState實例gProcess。全局唯一實例變量gProcess定義在frameworks/base/libs/binder/Static.cpp文件中：
Mutex gProcessMutex;
sp<ProcessState> gProcess;
再來看ProcessState的構造函數：
ProcessState::ProcessState() : mDriverFD(open_driver()) , mVMStart(MAP_FAILED) , mManagesContexts(false) , mBinderContextCheckFunc(NULL) , mBinderContextUserData(NULL) , mThreadPoolStarted(false) , mThreadPoolSeq(1) { if (mDriverFD >= 0) { // XXX Ideally, there should be a specific define for whether we // have mmap (or whether we could possibly have the kernel module // availabla). #if !defined(HAVE_WIN32_IPC) // mmap the binder, providing a chunk of virtual address space to receive transactions. mVMStart = mmap(0, BINDER_VM_SIZE, PROT_READ, MAP_PRIVATE | MAP_NORESERVE, mDriverFD, 0); if (mVMStart == MAP_FAILED) { // *sigh* LOGE("Using /dev/binder failed: unable to mmap transaction memory./n"); close(mDriverFD); mDriverFD = -1; } #else mDriverFD = -1; #endif } if (mDriverFD < 0) { // Need to run without the driver, starting our own thread pool. } } 這個函數有兩個關鍵地方，一是通過open_driver函數打開Binder設備文件/dev/binder，并將打開設備文件描述符保存在成員變量mDriverFD中；二是通過mmap來把設備文件/dev/binder映射到內存中。
先看open_driver函數的實現，這個函數同樣位于frameworks/base/libs/binder/ProcessState.cpp文件中：
static int open_driver() { if (gSingleProcess) { return -1; } int fd = open("/dev/binder", O_RDWR); if (fd >= 0) { fcntl(fd, F_SETFD, FD_CLOEXEC); int vers; #if defined(HAVE_ANDROID_OS) status_t result = ioctl(fd, BINDER_VERSION, &vers); #else status_t result = -1; errno = EPERM; #endif if (result == -1) { LOGE("Binder ioctl to obtain version failed: %s", strerror(errno)); close(fd); fd = -1; } if (result != 0 || vers != BINDER_CURRENT_PROTOCOL_VERSION) { LOGE("Binder driver protocol does not match user space protocol!"); close(fd); fd = -1; } #if defined(HAVE_ANDROID_OS) size_t maxThreads = 15; result = ioctl(fd, BINDER_SET_MAX_THREADS, &maxThreads); if (result == -1) { LOGE("Binder ioctl to set max threads failed: %s", strerror(errno)); } #endif } else { LOGW("Opening '/dev/binder' failed: %s/n", strerror(errno)); } return fd; } 這個函數的作用主要是通過open文件操作函數來打開/dev/binder設備文件，然后再調用ioctl文件控制函數來分別執行BINDER_VERSION和BINDER_SET_MAX_THREADS兩個命令來和Binder驅動程序進行交互，前者用于獲得當前Binder驅動程序的版本號，后者用于通知Binder驅動程序，MediaPlayerService最多可同時啟動15個線程來處理Client端的請求。
open在Binder驅動程序中的具體實現，請參考前面一篇文章淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路，這里不再重復描述。打開/dev/binder設備文件后，Binder驅動程序就為MediaPlayerService進程創建了一個struct binder_proc結構體實例來維護MediaPlayerService進程上下文相關信息。
我們來看一下ioctl文件操作函數執行BINDER_VERSION命令的過程：
status_t result = ioctl(fd, BINDER_VERSION, &vers);
這個函數調用最終進入到Binder驅動程序的binder_ioctl函數中，我們只關注BINDER_VERSION相關的部分邏輯：
static long binder_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { int ret; struct binder_proc *proc = filp->private_data; struct binder_thread *thread; unsigned int size = _IOC_SIZE(cmd); void __user *ubuf = (void __user *)arg; /*printk(KERN_INFO "binder_ioctl: %d:%d %x %lx/n", proc->pid, current->pid, cmd, arg);*/ ret = wait_event_interruptible(binder_user_error_wait, binder_stop_on_user_error < 2); if (ret) return ret; mutex_lock(&binder_lock); thread = binder_get_thread(proc); if (thread == NULL) { ret = -ENOMEM; goto err; } switch (cmd) { ...... case BINDER_VERSION: if (size != sizeof(struct binder_version)) { ret = -EINVAL; goto err; } if (put_user(BINDER_CURRENT_PROTOCOL_VERSION, &((struct binder_version *)ubuf)->protocol_version)) { ret = -EINVAL; goto err; } break; ...... } ret = 0; err: ...... return ret; } 很簡單，只是將BINDER_CURRENT_PROTOCOL_VERSION寫入到傳入的參數arg指向的用戶緩沖區中去就返回了。BINDER_CURRENT_PROTOCOL_VERSION是一個宏，定義在kernel/common/drivers/staging/android/binder.h文件中：
/* This is the current protocol version. */
#define BINDER_CURRENT_PROTOCOL_VERSION 7
這里為什么要把ubuf轉換成struct binder_version之后，再通過其protocol_version成員變量再來寫入呢，轉了一圈，最終內容還是寫入到ubuf中。我們看一下struct binder_version的定義就會明白，同樣是在kernel/common/drivers/staging/android/binder.h文件中：
/* Use with BINDER_VERSION, driver fills in fields. */ struct binder_version { /* driver protocol version -- increment with incompatible change */ signed long protocol_version; }; 從注釋中可以看出來，這里是考慮到兼容性，因為以后很有可能不是用signed long來表示版本號。
這里有一個重要的地方要注意的是，由于這里是打開設備文件/dev/binder之后，第一次進入到binder_ioctl函數，因此，這里調用binder_get_thread的時候，就會為當前線程創建一個struct binder_thread結構體變量來維護線程上下文信息，具體可以參考淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路一文。
接著我們再來看一下ioctl文件操作函數執行BINDER_SET_MAX_THREADS命令的過程：
result = ioctl(fd, BINDER_SET_MAX_THREADS, &maxThreads);
這個函數調用最終進入到Binder驅動程序的binder_ioctl函數中，我們只關注BINDER_SET_MAX_THREADS相關的部分邏輯：
static long binder_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { int ret; struct binder_proc *proc = filp->private_data; struct binder_thread *thread; unsigned int size = _IOC_SIZE(cmd); void __user *ubuf = (void __user *)arg; /*printk(KERN_INFO "binder_ioctl: %d:%d %x %lx/n", proc->pid, current->pid, cmd, arg);*/ ret = wait_event_interruptible(binder_user_error_wait, binder_stop_on_user_error < 2); if (ret) return ret; mutex_lock(&binder_lock); thread = binder_get_thread(proc); if (thread == NULL) { ret = -ENOMEM; goto err; } switch (cmd) { ...... case BINDER_SET_MAX_THREADS: if (copy_from_user(&proc->max_threads, ubuf, sizeof(proc->max_threads))) { ret = -EINVAL; goto err; } break; ...... } ret = 0; err: ...... return ret; } 這里實現也是非常簡單，只是簡單地把用戶傳進來的參數保存在proc->max_threads中就完畢了。注意，這里再調用binder_get_thread函數的時候，就可以在proc->threads中找到當前線程對應的struct binder_thread結構了，因為前面已經創建好并保存在proc->threads紅黑樹中。
回到ProcessState的構造函數中，這里還通過mmap函數來把設備文件/dev/binder映射到內存中，這個函數在淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路一文也已經有詳細介紹，這里不再重復描述。宏BINDER_VM_SIZE就定義在ProcessState.cpp文件中：
#define BINDER_VM_SIZE ((1*1024*1024) - (4096 *2))
mmap函數調用完成之后，Binder驅動程序就為當前進程預留了BINDER_VM_SIZE大小的內存空間了。
這樣，ProcessState全局唯一變量gProcess就創建完畢了，回到frameworks/base/media/mediaserver/main_mediaserver.cpp文件中的main函數，下一步是調用defaultServiceManager函數來獲得Service Manager的遠程接口，這個已經在上一篇文章淺談Android系統進程間通信（IPC）機制Binder中的Server和Client獲得Service Manager接口之路有詳細描述，讀者可以回過頭去參考一下。
再接下來，就進入到MediaPlayerService::instantiate函數把MediaPlayerService添加到Service Manger中去了。這個函數定義在frameworks/base/media/libmediaplayerservice/MediaPlayerService.cpp文件中：
void MediaPlayerService::instantiate() { defaultServiceManager()->addService( String16("media.player"), new MediaPlayerService()); } 我們重點看一下IServiceManger::addService的過程，這有助于我們加深對Binder機制的理解。
在上一篇文章淺談Android系統進程間通信（IPC）機制Binder中的Server和Client獲得Service Manager接口之路中說到，defaultServiceManager返回的實際是一個BpServiceManger類實例，因此，我們看一下BpServiceManger::addService的實現，這個函數實現在frameworks/base/libs/binder/IServiceManager.cpp文件中：
class BpServiceManager : public BpInterface<IServiceManager> { public: BpServiceManager(const sp<IBinder>& impl) : BpInterface<IServiceManager>(impl) { } ...... virtual status_t addService(const String16& name, const sp<IBinder>& service) { Parcel data, reply; data.writeInterfaceToken(IServiceManager::getInterfaceDescriptor()); data.writeString16(name); data.writeStrongBinder(service); status_t err = remote()->transact(ADD_SERVICE_TRANSACTION, data, &reply); return err == NO_ERROR ? reply.readExceptionCode() } ...... }; 這里的Parcel類是用來于序列化進程間通信數據用的。
先來看這一句的調用：
data.writeInterfaceToken(IServiceManager::getInterfaceDescriptor());
IServiceManager::getInterfaceDescriptor()返回來的是一個字符串，即"android.os.IServiceManager"，具體可以參考IServiceManger的實現。我們看一下Parcel::writeInterfaceToken的實現，位于frameworks/base/libs/binder/Parcel.cpp文件中：
// Write RPC headers. (previously just the interface token) status_t Parcel::writeInterfaceToken(const String16& interface) { writeInt32(IPCThreadState::self()->getStrictModePolicy() | STRICT_MODE_PENALTY_GATHER); // currently the interface identification token is just its name as a string return writeString16(interface); } 它的作用是寫入一個整數和一個字符串到Parcel中去。
再來看下面的調用：
data.writeString16(name);
這里又是寫入一個字符串到Parcel中去，這里的name即是上面傳進來的“media.player”字符串。
往下看：
data.writeStrongBinder(service);
這里定入一個Binder對象到Parcel去。我們重點看一下這個函數的實現，因為它涉及到進程間傳輸Binder實體的問題，比較復雜，需要重點關注，同時，也是理解Binder機制的一個重點所在。注意，這里的service參數是一個MediaPlayerService對象。
status_t Parcel::writeStrongBinder(const sp<IBinder>& val) { return flatten_binder(ProcessState::self(), val, this); }
看到flatten_binder函數，是不是似曾相識的感覺？我們在前面一篇文章淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路中，曾經提到在Binder驅動程序中，使用struct flat_binder_object來表示傳輸中的一個binder對象，它的定義如下所示：
/* * This is the flattened representation of a Binder object for transfer * between processes. The 'offsets' supplied as part of a binder transaction * contains offsets into the data where these structures occur. The Binder * driver takes care of re-writing the structure type and data as it moves * between processes. */ struct flat_binder_object { /* 8 bytes for large_flat_header. */ unsigned long type; unsigned long flags; /* 8 bytes of data. */ union { void *binder; /* local object */ signed long handle; /* remote object */ }; /* extra data associated with local object */ void *cookie; }; 各個成員變量的含義請參考資料Android Binder設計與實現。
我們進入到flatten_binder函數看看：
status_t flatten_binder(const sp<ProcessState>& proc, const sp<IBinder>& binder, Parcel* out) { flat_binder_object obj; obj.flags = 0x7f | FLAT_BINDER_FLAG_ACCEPTS_FDS; if (binder != NULL) { IBinder *local = binder->localBinder(); if (!local) { BpBinder *proxy = binder->remoteBinder(); if (proxy == NULL) { LOGE("null proxy"); } const int32_t handle = proxy ? proxy->handle() : 0; obj.type = BINDER_TYPE_HANDLE; obj.handle = handle; obj.cookie = NULL; } else { obj.type = BINDER_TYPE_BINDER; obj.binder = local->getWeakRefs(); obj.cookie = local; } } else { obj.type = BINDER_TYPE_BINDER; obj.binder = NULL; obj.cookie = NULL; } return finish_flatten_binder(binder, obj, out); } 首先是初始化flat_binder_object的flags域：
obj.flags = 0x7f | FLAT_BINDER_FLAG_ACCEPTS_FDS;
0x7f表示處理本Binder實體請求數據包的線程的最低優先級，FLAT_BINDER_FLAG_ACCEPTS_FDS表示這個Binder實體可以接受文件描述符，Binder實體在收到文件描述符時，就會在本進程中打開這個文件。
傳進來的binder即為MediaPlayerService::instantiate函數中new出來的MediaPlayerService實例，因此，不為空。又由于MediaPlayerService繼承自BBinder類，它是一個本地Binder實體，因此binder->localBinder返回一個BBinder指針，而且肯定不為空，于是執行下面語句：
obj.type = BINDER_TYPE_BINDER; obj.binder = local->getWeakRefs(); obj.cookie = local;
設置了flat_binder_obj的其他成員變量，注意，指向這個Binder實體地址的指針local保存在flat_binder_obj的成員變量cookie中。
函數調用finish_flatten_binder來將這個flat_binder_obj寫入到Parcel中去：
inline static status_t finish_flatten_binder( const sp<IBinder>& binder, const flat_binder_object& flat, Parcel* out) { return out->writeObject(flat, false); } Parcel::writeObject的實現如下：
status_t Parcel::writeObject(const flat_binder_object& val, bool nullMetaData) { const bool enoughData = (mDataPos+sizeof(val)) <= mDataCapacity; const bool enoughObjects = mObjectsSize < mObjectsCapacity; if (enoughData && enoughObjects) { restart_write: *reinterpret_cast<flat_binder_object*>(mData+mDataPos) = val; // Need to write meta-data? if (nullMetaData || val.binder != NULL) { mObjects[mObjectsSize] = mDataPos; acquire_object(ProcessState::self(), val, this); mObjectsSize++; } // remember if it's a file descriptor if (val.type == BINDER_TYPE_FD) { mHasFds = mFdsKnown = true; } return finishWrite(sizeof(flat_binder_object)); } if (!enoughData) { const status_t err = growData(sizeof(val)); if (err != NO_ERROR) return err; } if (!enoughObjects) { size_t newSize = ((mObjectsSize+2)*3)/2; size_t* objects = (size_t*)realloc(mObjects, newSize*sizeof(size_t)); if (objects == NULL) return NO_MEMORY; mObjects = objects; mObjectsCapacity = newSize; } goto restart_write; } 這里除了把flat_binder_obj寫到Parcel里面之內，還要記錄這個flat_binder_obj在Parcel里面的偏移位置：
mObjects[mObjectsSize] = mDataPos;
這里因為，如果進程間傳輸的數據間帶有Binder對象的時候，Binder驅動程序需要作進一步的處理，以維護各個Binder實體的一致性，下面我們將會看到Binder驅動程序是怎么處理這些Binder對象的。
再回到BpServiceManager::addService函數中，調用下面語句：
status_t err = remote()->transact(ADD_SERVICE_TRANSACTION, data, &reply);
回到淺談Android系統進程間通信（IPC）機制Binder中的Server和Client獲得Service Manager接口之路一文中的類圖中去看一下，這里的remote成員函數來自于BpRefBase類，它返回一個BpBinder指針。因此，我們繼續進入到BpBinder::transact函數中去看看：
status_t BpBinder::transact( uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags) { // Once a binder has died, it will never come back to life. if (mAlive) { status_t status = IPCThreadState::self()->transact( mHandle, code, data, reply, flags); if (status == DEAD_OBJECT) mAlive = 0; return status; } return DEAD_OBJECT; } 這里又調用了IPCThreadState::transact進執行實際的操作。注意，這里的mHandle為0，code為ADD_SERVICE_TRANSACTION。ADD_SERVICE_TRANSACTION是上面以參數形式傳進來的，那mHandle為什么是0呢？因為這里表示的是Service Manager遠程接口，它的句柄值一定是0，具體請參考淺談Android系統進程間通信（IPC）機制Binder中的Server和Client獲得Service Manager接口之路一文。
再進入到IPCThreadState::transact函數，看看做了些什么事情：
status_t IPCThreadState::transact(int32_t handle, uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags) { status_t err = data.errorCheck(); flags |= TF_ACCEPT_FDS; IF_LOG_TRANSACTIONS() { TextOutput::Bundle _b(alog); alog << "BC_TRANSACTION thr " << (void*)pthread_self() << " / hand " << handle << " / code " << TypeCode(code) << ": " << indent << data << dedent << endl; } if (err == NO_ERROR) { LOG_ONEWAY(">>>> SEND from pid %d uid %d %s", getpid(), getuid(), (flags & TF_ONE_WAY) == 0 ? "READ REPLY" : "ONE WAY"); err = writeTransactionData(BC_TRANSACTION, flags, handle, code, data, NULL); } if (err != NO_ERROR) { if (reply) reply->setError(err); return (mLastError = err); } if ((flags & TF_ONE_WAY) == 0) { #if 0 if (code == 4) { // relayout LOGI(">>>>>> CALLING transaction 4"); } else { LOGI(">>>>>> CALLING transaction %d", code); } #endif if (reply) { err = waitForResponse(reply); } else { Parcel fakeReply; err = waitForResponse(&fakeReply); } #if 0 if (code == 4) { // relayout LOGI("<<<<<< RETURNING transaction 4"); } else { LOGI("<<<<<< RETURNING transaction %d", code); } #endif IF_LOG_TRANSACTIONS() { TextOutput::Bundle _b(alog); alog << "BR_REPLY thr " << (void*)pthread_self() << " / hand " << handle << ": "; if (reply) alog << indent << *reply << dedent << endl; else alog << "(none requested)" << endl; } } else { err = waitForResponse(NULL, NULL); } return err; } IPCThreadState::transact函數的參數flags是一個默認值為0的參數，上面沒有傳相應的實參進來，因此，這里就為0。
函數首先調用writeTransactionData函數準備好一個struct binder_transaction_data結構體變量，這個是等一下要傳輸給Binder驅動程序的。struct binder_transaction_data的定義我們在淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路一文中有詳細描述，讀者不妨回過去讀一下。這里為了方便描述，將struct binder_transaction_data的定義再次列出來：
struct binder_transaction_data { /* The first two are only used for bcTRANSACTION and brTRANSACTION, * identifying the target and contents of the transaction. */ union { size_t handle; /* target descriptor of command transaction */ void *ptr; /* target descriptor of return transaction */ } target; void *cookie; /* target object cookie */ unsigned int code; /* transaction command */ /* General information about the transaction. */ unsigned int flags; pid_t sender_pid; uid_t sender_euid; size_t data_size; /* number of bytes of data */ size_t offsets_size; /* number of bytes of offsets */ /* If this transaction is inline, the data immediately * follows here; otherwise, it ends with a pointer to * the data buffer. */ union { struct { /* transaction data */ const void *buffer; /* offsets from buffer to flat_binder_object structs */ const void *offsets; } ptr; uint8_t buf[8]; } data; }; writeTransactionData函數的實現如下：
status_t IPCThreadState::writeTransactionData(int32_t cmd, uint32_t binderFlags, int32_t handle, uint32_t code, const Parcel& data, status_t* statusBuffer) { binder_transaction_data tr; tr.target.handle = handle; tr.code = code; tr.flags = binderFlags; const status_t err = data.errorCheck(); if (err == NO_ERROR) { tr.data_size = data.ipcDataSize(); tr.data.ptr.buffer = data.ipcData(); tr.offsets_size = data.ipcObjectsCount()*sizeof(size_t); tr.data.ptr.offsets = data.ipcObjects(); } else if (statusBuffer) { tr.flags |= TF_STATUS_CODE; *statusBuffer = err; tr.data_size = sizeof(status_t); tr.data.ptr.buffer = statusBuffer; tr.offsets_size = 0; tr.data.ptr.offsets = NULL; } else { return (mLastError = err); } mOut.writeInt32(cmd); mOut.write(&tr, sizeof(tr)); return NO_ERROR; } 注意，這里的cmd為BC_TRANSACTION。 這個函數很簡單，在這個場景下，就是執行下面語句來初始化本地變量tr：
tr.data_size = data.ipcDataSize(); tr.data.ptr.buffer = data.ipcData(); tr.offsets_size = data.ipcObjectsCount()*sizeof(size_t); tr.data.ptr.offsets = data.ipcObjects();
回憶一下上面的內容，寫入到tr.data.ptr.buffer的內容相當于下面的內容：
writeInt32(IPCThreadState::self()->getStrictModePolicy() | STRICT_MODE_PENALTY_GATHER); writeString16("android.os.IServiceManager"); writeString16("media.player"); writeStrongBinder(new MediaPlayerService());
其中包含了一個Binder實體MediaPlayerService，因此需要設置tr.offsets_size就為1，tr.data.ptr.offsets就指向了這個MediaPlayerService的地址在tr.data.ptr.buffer中的偏移量。最后，將tr的內容保存在IPCThreadState的成員變量mOut中。
回到IPCThreadState::transact函數中，接下去看，(flags & TF_ONE_WAY) == 0為true，并且reply不為空，所以最終進入到waitForResponse(reply)這條路徑來。我們看一下waitForResponse函數的實現：
status_t IPCThreadState::waitForResponse(Parcel *reply, status_t *acquireResult) { int32_t cmd; int32_t err; while (1) { if ((err=talkWithDriver()) < NO_ERROR) break; err = mIn.errorCheck(); if (err < NO_ERROR) break; if (mIn.dataAvail() == 0) continue; cmd = mIn.readInt32(); IF_LOG_COMMANDS() { alog << "Processing waitForResponse Command: " << getReturnString(cmd) << endl; } switch (cmd) { case BR_TRANSACTION_COMPLETE: if (!reply && !acquireResult) goto finish; break; case BR_DEAD_REPLY: err = DEAD_OBJECT; goto finish; case BR_FAILED_REPLY: err = FAILED_TRANSACTION; goto finish; case BR_ACQUIRE_RESULT: { LOG_ASSERT(acquireResult != NULL, "Unexpected brACQUIRE_RESULT"); const int32_t result = mIn.readInt32(); if (!acquireResult) continue; *acquireResult = result ? NO_ERROR : INVALID_OPERATION; } goto finish; case BR_REPLY: { binder_transaction_data tr; err = mIn.read(&tr, sizeof(tr)); LOG_ASSERT(err == NO_ERROR, "Not enough command data for brREPLY"); if (err != NO_ERROR) goto finish; if (reply) { if ((tr.flags & TF_STATUS_CODE) == 0) { reply->ipcSetDataReference( reinterpret_cast<const uint8_t*>(tr.data.ptr.buffer), tr.data_size, reinterpret_cast<const size_t*>(tr.data.ptr.offsets), tr.offsets_size/sizeof(size_t), freeBuffer, this); } else { err = *static_cast<const status_t*>(tr.data.ptr.buffer); freeBuffer(NULL, reinterpret_cast<const uint8_t*>(tr.data.ptr.buffer), tr.data_size, reinterpret_cast<const size_t*>(tr.data.ptr.offsets), tr.offsets_size/sizeof(size_t), this); } } else { freeBuffer(NULL, reinterpret_cast<const uint8_t*>(tr.data.ptr.buffer), tr.data_size, reinterpret_cast<const size_t*>(tr.data.ptr.offsets), tr.offsets_size/sizeof(size_t), this); continue; } } goto finish; default: err = executeCommand(cmd); if (err != NO_ERROR) goto finish; break; } } finish: if (err != NO_ERROR) { if (acquireResult) *acquireResult = err; if (reply) reply->setError(err); mLastError = err; } return err; } 這個函數雖然很長，但是主要調用了talkWithDriver函數來與Binder驅動程序進行交互：
status_t IPCThreadState::talkWithDriver(bool doReceive) { LOG_ASSERT(mProcess->mDriverFD >= 0, "Binder driver is not opened"); binder_write_read bwr; // Is the read buffer empty? const bool needRead = mIn.dataPosition() >= mIn.dataSize(); // We don't want to write anything if we are still reading // from data left in the input buffer and the caller // has requested to read the next data. const size_t outAvail = (!doReceive || needRead) ? mOut.dataSize() : 0; bwr.write_size = outAvail; bwr.write_buffer = (long unsigned int)mOut.data(); // This is what we'll read. if (doReceive && needRead) { bwr.read_size = mIn.dataCapacity(); bwr.read_buffer = (long unsigned int)mIn.data(); } else { bwr.read_size = 0; } IF_LOG_COMMANDS() { TextOutput::Bundle _b(alog); if (outAvail != 0) { alog << "Sending commands to driver: " << indent; const void* cmds = (const void*)bwr.write_buffer; const void* end = ((const uint8_t*)cmds)+bwr.write_size; alog << HexDump(cmds, bwr.write_size) << endl; while (cmds < end) cmds = printCommand(alog, cmds); alog << dedent; } alog << "Size of receive buffer: " << bwr.read_size << ", needRead: " << needRead << ", doReceive: " << doReceive << endl; } // Return immediately if there is nothing to do. if ((bwr.write_size == 0) && (bwr.read_size == 0)) return NO_ERROR; bwr.write_consumed = 0; bwr.read_consumed = 0; status_t err; do { IF_LOG_COMMANDS() { alog << "About to read/write, write size = " << mOut.dataSize() << endl; } #if defined(HAVE_ANDROID_OS) if (ioctl(mProcess->mDriverFD, BINDER_WRITE_READ, &bwr) >= 0) err = NO_ERROR; else err = -errno; #else err = INVALID_OPERATION; #endif IF_LOG_COMMANDS() { alog << "Finished read/write, write size = " << mOut.dataSize() << endl; } } while (err == -EINTR); IF_LOG_COMMANDS() { alog << "Our err: " << (void*)err << ", write consumed: " << bwr.write_consumed << " (of " << mOut.dataSize() << "), read consumed: " << bwr.read_consumed << endl; } if (err >= NO_ERROR) { if (bwr.write_consumed > 0) { if (bwr.write_consumed < (ssize_t)mOut.dataSize()) mOut.remove(0, bwr.write_consumed); else mOut.setDataSize(0); } if (bwr.read_consumed > 0) { mIn.setDataSize(bwr.read_consumed); mIn.setDataPosition(0); } IF_LOG_COMMANDS() { TextOutput::Bundle _b(alog); alog << "Remaining data size: " << mOut.dataSize() << endl; alog << "Received commands from driver: " << indent; const void* cmds = mIn.data(); const void* end = mIn.data() + mIn.dataSize(); alog << HexDump(cmds, mIn.dataSize()) << endl; while (cmds < end) cmds = printReturnCommand(alog, cmds); alog << dedent; } return NO_ERROR; } return err; } 這里doReceive和needRead均為1，有興趣的讀者可以自已分析一下。因此，這里告訴Binder驅動程序，先執行write操作，再執行read操作，下面我們將會看到。
最后，通過ioctl(mProcess->mDriverFD, BINDER_WRITE_READ, &bwr)進行到Binder驅動程序的binder_ioctl函數，我們只關注cmd為BINDER_WRITE_READ的邏輯：
static long binder_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { int ret; struct binder_proc *proc = filp->private_data; struct binder_thread *thread; unsigned int size = _IOC_SIZE(cmd); void __user *ubuf = (void __user *)arg; /*printk(KERN_INFO "binder_ioctl: %d:%d %x %lx/n", proc->pid, current->pid, cmd, arg);*/ ret = wait_event_interruptible(binder_user_error_wait, binder_stop_on_user_error < 2); if (ret) return ret; mutex_lock(&binder_lock); thread = binder_get_thread(proc); if (thread == NULL) { ret = -ENOMEM; goto err; } switch (cmd) { case BINDER_WRITE_READ: { struct binder_write_read bwr; if (size != sizeof(struct binder_write_read)) { ret = -EINVAL; goto err; } if (copy_from_user(&bwr, ubuf, sizeof(bwr))) { ret = -EFAULT; goto err; } if (binder_debug_mask & BINDER_DEBUG_READ_WRITE) printk(KERN_INFO "binder: %d:%d write %ld at %08lx, read %ld at %08lx/n", proc->pid, thread->pid, bwr.write_size, bwr.write_buffer, bwr.read_size, bwr.read_buffer); if (bwr.write_size > 0) { ret = binder_thread_write(proc, thread, (void __user *)bwr.write_buffer, bwr.write_size, &bwr.write_consumed); if (ret < 0) { bwr.read_consumed = 0; if (copy_to_user(ubuf, &bwr, sizeof(bwr))) ret = -EFAULT; goto err; } } if (bwr.read_size > 0) { ret = binder_thread_read(proc, thread, (void __user *)bwr.read_buffer, bwr.read_size, &bwr.read_consumed, filp->f_flags & O_NONBLOCK); if (!list_empty(&proc->todo)) wake_up_interruptible(&proc->wait); if (ret < 0) { if (copy_to_user(ubuf, &bwr, sizeof(bwr))) ret = -EFAULT; goto err; } } if (binder_debug_mask & BINDER_DEBUG_READ_WRITE) printk(KERN_INFO "binder: %d:%d wrote %ld of %ld, read return %ld of %ld/n", proc->pid, thread->pid, bwr.write_consumed, bwr.write_size, bwr.read_consumed, bwr.read_size); if (copy_to_user(ubuf, &bwr, sizeof(bwr))) { ret = -EFAULT; goto err; } break; } ...... } ret = 0; err: ...... return ret; } 函數首先是將用戶傳進來的參數拷貝到本地變量struct binder_write_read bwr中去。這里bwr.write_size > 0為true，因此，進入到binder_thread_write函數中，我們只關注BC_TRANSACTION部分的邏輯：
binder_thread_write(struct binder_proc *proc, struct binder_thread *thread, void __user *buffer, int size, signed long *consumed) { uint32_t cmd; void __user *ptr = buffer + *consumed; void __user *end = buffer + size; while (ptr < end && thread->return_error == BR_OK) { if (get_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); if (_IOC_NR(cmd) < ARRAY_SIZE(binder_stats.bc)) { binder_stats.bc[_IOC_NR(cmd)]++; proc->stats.bc[_IOC_NR(cmd)]++; thread->stats.bc[_IOC_NR(cmd)]++; } switch (cmd) { ..... case BC_TRANSACTION: case BC_REPLY: { struct binder_transaction_data tr; if (copy_from_user(&tr, ptr, sizeof(tr))) return -EFAULT; ptr += sizeof(tr); binder_transaction(proc, thread, &tr, cmd == BC_REPLY); break; } ...... } *consumed = ptr - buffer; } return 0; } 首先將用戶傳進來的transact參數拷貝在本地變量struct binder_transaction_data tr中去，接著調用binder_transaction函數進一步處理，這里我們忽略掉無關代碼：
static void binder_transaction(struct binder_proc *proc, struct binder_thread *thread, struct binder_transaction_data *tr, int reply) { struct binder_transaction *t; struct binder_work *tcomplete; size_t *offp, *off_end; struct binder_proc *target_proc; struct binder_thread *target_thread = NULL; struct binder_node *target_node = NULL; struct list_head *target_list; wait_queue_head_t *target_wait; struct binder_transaction *in_reply_to = NULL; struct binder_transaction_log_entry *e; uint32_t return_error; ...... if (reply) { ...... } else { if (tr->target.handle) { ...... } else { target_node = binder_context_mgr_node; if (target_node == NULL) { return_error = BR_DEAD_REPLY; goto err_no_context_mgr_node; } } ...... target_proc = target_node->proc; if (target_proc == NULL) { return_error = BR_DEAD_REPLY; goto err_dead_binder; } ...... } if (target_thread) { ...... } else { target_list = &target_proc->todo; target_wait = &target_proc->wait; } ...... /* TODO: reuse incoming transaction for reply */ t = kzalloc(sizeof(*t), GFP_KERNEL); if (t == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_t_failed; } ...... tcomplete = kzalloc(sizeof(*tcomplete), GFP_KERNEL); if (tcomplete == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_tcomplete_failed; } ...... if (!reply && !(tr->flags & TF_ONE_WAY)) t->from = thread; else t->from = NULL; t->sender_euid = proc->tsk->cred->euid; t->to_proc = target_proc; t->to_thread = target_thread; t->code = tr->code; t->flags = tr->flags; t->priority = task_nice(current); t->buffer = binder_alloc_buf(target_proc, tr->data_size, tr->offsets_size, !reply && (t->flags & TF_ONE_WAY)); if (t->buffer == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_alloc_buf_failed; } t->buffer->allow_user_free = 0; t->buffer->debug_id = t->debug_id; t->buffer->transaction = t; t->buffer->target_node = target_node; if (target_node) binder_inc_node(target_node, 1, 0, NULL); offp = (size_t *)(t->buffer->data + ALIGN(tr->data_size, sizeof(void *))); if (copy_from_user(t->buffer->data, tr->data.ptr.buffer, tr->data_size)) { ...... return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } if (copy_from_user(offp, tr->data.ptr.offsets, tr->offsets_size)) { ...... return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } ...... off_end = (void *)offp + tr->offsets_size; for (; offp < off_end; offp++) { struct flat_binder_object *fp; ...... fp = (struct flat_binder_object *)(t->buffer->data + *offp); switch (fp->type) { case BINDER_TYPE_BINDER: case BINDER_TYPE_WEAK_BINDER: { struct binder_ref *ref; struct binder_node *node = binder_get_node(proc, fp->binder); if (node == NULL) { node = binder_new_node(proc, fp->binder, fp->cookie); if (node == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_new_node_failed; } node->min_priority = fp->flags & FLAT_BINDER_FLAG_PRIORITY_MASK; node->accept_fds = !!(fp->flags & FLAT_BINDER_FLAG_ACCEPTS_FDS); } if (fp->cookie != node->cookie) { ...... goto err_binder_get_ref_for_node_failed; } ref = binder_get_ref_for_node(target_proc, node); if (ref == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_get_ref_for_node_failed; } if (fp->type == BINDER_TYPE_BINDER) fp->type = BINDER_TYPE_HANDLE; else fp->type = BINDER_TYPE_WEAK_HANDLE; fp->handle = ref->desc; binder_inc_ref(ref, fp->type == BINDER_TYPE_HANDLE, &thread->todo); ...... } break; ...... } } if (reply) { ...... } else if (!(t->flags & TF_ONE_WAY)) { BUG_ON(t->buffer->async_transaction != 0); t->need_reply = 1; t->from_parent = thread->transaction_stack; thread->transaction_stack = t; } else { ...... } t->work.type = BINDER_WORK_TRANSACTION; list_add_tail(&t->work.entry, target_list); tcomplete->type = BINDER_WORK_TRANSACTION_COMPLETE; list_add_tail(&tcomplete->entry, &thread->todo); if (target_wait) wake_up_interruptible(target_wait); return; ...... } 注意，這里傳進來的參數reply為0，tr->target.handle也為0。因此，target_proc、target_thread、target_node、target_list和target_wait的值分別為：
target_node = binder_context_mgr_node; target_proc = target_node->proc; target_list = &target_proc->todo; target_wait = &target_proc->wait;
接著，分配了一個待處理事務t和一個待完成工作項tcomplete，并執行初始化工作：
/* TODO: reuse incoming transaction for reply */ t = kzalloc(sizeof(*t), GFP_KERNEL); if (t == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_t_failed; } ...... tcomplete = kzalloc(sizeof(*tcomplete), GFP_KERNEL); if (tcomplete == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_tcomplete_failed; } ...... if (!reply && !(tr->flags & TF_ONE_WAY)) t->from = thread; else t->from = NULL; t->sender_euid = proc->tsk->cred->euid; t->to_proc = target_proc; t->to_thread = target_thread; t->code = tr->code; t->flags = tr->flags; t->priority = task_nice(current); t->buffer = binder_alloc_buf(target_proc, tr->data_size, tr->offsets_size, !reply && (t->flags & TF_ONE_WAY)); if (t->buffer == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_alloc_buf_failed; } t->buffer->allow_user_free = 0; t->buffer->debug_id = t->debug_id; t->buffer->transaction = t; t->buffer->target_node = target_node; if (target_node) binder_inc_node(target_node, 1, 0, NULL); offp = (size_t *)(t->buffer->data + ALIGN(tr->data_size, sizeof(void *))); if (copy_from_user(t->buffer->data, tr->data.ptr.buffer, tr->data_size)) { ...... return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } if (copy_from_user(offp, tr->data.ptr.offsets, tr->offsets_size)) { ...... return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } 注意，這里的事務t是要交給target_proc處理的，在這個場景之下，就是Service Manager了。因此，下面的語句：
t->buffer = binder_alloc_buf(target_proc, tr->data_size, tr->offsets_size, !reply && (t->flags & TF_ONE_WAY));
就是在Service Manager的進程空間中分配一塊內存來保存用戶傳進入的參數了：
if (copy_from_user(t->buffer->data, tr->data.ptr.buffer, tr->data_size)) { ...... return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } if (copy_from_user(offp, tr->data.ptr.offsets, tr->offsets_size)) { ...... return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } 由于現在target_node要被使用了，增加它的引用計數：
if (target_node) binder_inc_node(target_node, 1, 0, NULL);
接下去的for循環，就是用來處理傳輸數據中的Binder對象了。在我們的場景中，有一個類型為BINDER_TYPE_BINDER的Binder實體MediaPlayerService：
switch (fp->type) { case BINDER_TYPE_BINDER: case BINDER_TYPE_WEAK_BINDER: { struct binder_ref *ref; struct binder_node *node = binder_get_node(proc, fp->binder); if (node == NULL) { node = binder_new_node(proc, fp->binder, fp->cookie); if (node == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_new_node_failed; } node->min_priority = fp->flags & FLAT_BINDER_FLAG_PRIORITY_MASK; node->accept_fds = !!(fp->flags & FLAT_BINDER_FLAG_ACCEPTS_FDS); } if (fp->cookie != node->cookie) { ...... goto err_binder_get_ref_for_node_failed; } ref = binder_get_ref_for_node(target_proc, node); if (ref == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_get_ref_for_node_failed; } if (fp->type == BINDER_TYPE_BINDER) fp->type = BINDER_TYPE_HANDLE; else fp->type = BINDER_TYPE_WEAK_HANDLE; fp->handle = ref->desc; binder_inc_ref(ref, fp->type == BINDER_TYPE_HANDLE, &thread->todo); ...... } break; 由于是第一次在Binder驅動程序中傳輸這個MediaPlayerService，調用binder_get_node函數查詢這個Binder實體時，會返回空，于是binder_new_node在proc中新建一個，下次就可以直接使用了。
現在，由于要把這個Binder實體MediaPlayerService交給target_proc，也就是Service Manager來管理，也就是說Service Manager要引用這個MediaPlayerService了，于是通過binder_get_ref_for_node為MediaPlayerService創建一個引用，并且通過binder_inc_ref來增加這個引用計數，防止這個引用還在使用過程當中就被銷毀。注意，到了這里的時候，t->buffer中的flat_binder_obj的type已經改為BINDER_TYPE_HANDLE，handle已經改為ref->desc，跟原來不一樣了，因為這個flat_binder_obj是最終是要傳給Service Manager的，而Service Manager只能夠通過句柄值來引用這個Binder實體。
最后，把待處理事務加入到target_list列表中去：
list_add_tail(&t->work.entry, target_list);
并且把待完成工作項加入到本線程的todo等待執行列表中去：
list_add_tail(&tcomplete->entry, &thread->todo);
現在目標進程有事情可做了，于是喚醒它：
if (target_wait)
wake_up_interruptible(target_wait);
這里就是要喚醒Service Manager進程了。回憶一下前面這篇文章，此時， Service Manager正在binder_t淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路hread_read函數中調用wait_event_interruptible進入休眠狀態。
這里我們先忽略一下Service Manager被喚醒之后的場景，繼續MedaPlayerService的啟動過程，然后再回來。
回到binder_ioctl函數，bwr.read_size > 0為true，于是進入binder_thread_read函數：
static int binder_thread_read(struct binder_proc *proc, struct binder_thread *thread, void __user *buffer, int size, signed long *consumed, int non_block) { void __user *ptr = buffer + *consumed; void __user *end = buffer + size; int ret = 0; int wait_for_proc_work; if (*consumed == 0) { if (put_user(BR_NOOP, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); } retry: wait_for_proc_work = thread->transaction_stack == NULL && list_empty(&thread->todo); ....... if (wait_for_proc_work) { ....... } else { if (non_block) { if (!binder_has_thread_work(thread)) ret = -EAGAIN; } else ret = wait_event_interruptible(thread->wait, binder_has_thread_work(thread)); } ...... while (1) { uint32_t cmd; struct binder_transaction_data tr; struct binder_work *w; struct binder_transaction *t = NULL; if (!list_empty(&thread->todo)) w = list_first_entry(&thread->todo, struct binder_work, entry); else if (!list_empty(&proc->todo) && wait_for_proc_work) w = list_first_entry(&proc->todo, struct binder_work, entry); else { if (ptr - buffer == 4 && !(thread->looper & BINDER_LOOPER_STATE_NEED_RETURN)) /* no data added */ goto retry; break; } if (end - ptr < sizeof(tr) + 4) break; switch (w->type) { ...... case BINDER_WORK_TRANSACTION_COMPLETE: { cmd = BR_TRANSACTION_COMPLETE; if (put_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); binder_stat_br(proc, thread, cmd); if (binder_debug_mask & BINDER_DEBUG_TRANSACTION_COMPLETE) printk(KERN_INFO "binder: %d:%d BR_TRANSACTION_COMPLETE/n", proc->pid, thread->pid); list_del(&w->entry); kfree(w); binder_stats.obj_deleted[BINDER_STAT_TRANSACTION_COMPLETE]++; } break; ...... } if (!t) continue; ...... } done: ...... return 0; } 這里，thread->transaction_stack和thread->todo均不為空，于是wait_for_proc_work為false，由于binder_has_thread_work的時候，返回true，這里因為thread->todo不為空，因此，線程雖然調用了wait_event_interruptible，但是不會睡眠，于是繼續往下執行。
由于thread->todo不為空，執行下列語句：
if (!list_empty(&thread->todo)) w = list_first_entry(&thread->todo, struct binder_work, entry);
w->type為BINDER_WORK_TRANSACTION_COMPLETE，這是在上面的binder_transaction函數設置的，于是執行：
switch (w->type) { ...... case BINDER_WORK_TRANSACTION_COMPLETE: { cmd = BR_TRANSACTION_COMPLETE; if (put_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); ...... list_del(&w->entry); kfree(w); } break; ...... } 這里就將w從thread->todo刪除了。由于這里t為空，重新執行while循環，這時由于已經沒有事情可做了，最后就返回到binder_ioctl函數中。注間，這里一共往用戶傳進來的緩沖區buffer寫入了兩個整數，分別是BR_NOOP和BR_TRANSACTION_COMPLETE。
binder_ioctl函數返回到用戶空間之前，把數據消耗情況拷貝回用戶空間中：
if (copy_to_user(ubuf, &bwr, sizeof(bwr))) { ret = -EFAULT; goto err; } 最后返回到IPCThreadState::talkWithDriver函數中，執行下面語句：
if (err >= NO_ERROR) { if (bwr.write_consumed > 0) { if (bwr.write_consumed < (ssize_t)mOut.dataSize()) mOut.remove(0, bwr.write_consumed); else mOut.setDataSize(0); } if (bwr.read_consumed > 0) { <pre code_snippet_id="134056" snippet_file_name="blog_20131230_54_6706870" name="code" class="cpp"> mIn.setDataSize(bwr.read_consumed); mIn.setDataPosition(0);</pre> } ...... return NO_ERROR; } 首先是把mOut的數據清空：
mOut.setDataSize(0);
然后設置已經讀取的內容的大小：
mIn.setDataSize(bwr.read_consumed);
mIn.setDataPosition(0);
然后返回到IPCThreadState::waitForResponse函數中。在IPCThreadState::waitForResponse函數，先是從mIn讀出一個整數，這個便是BR_NOOP了，這是一個空操作，什么也不做。然后繼續進入IPCThreadState::talkWithDriver函數中。
這時候，下面語句執行后：
const bool needRead = mIn.dataPosition() >= mIn.dataSize();
needRead為false，因為在mIn中，尚有一個整數BR_TRANSACTION_COMPLETE未讀出。
這時候，下面語句執行后：
const size_t outAvail = (!doReceive || needRead) ? mOut.dataSize() : 0;
outAvail等于0。因此，最后bwr.write_size和bwr.read_size均為0，IPCThreadState::talkWithDriver函數什么也不做，直接返回到IPCThreadState::waitForResponse函數中。在IPCThreadState::waitForResponse函數，又繼續從mIn讀出一個整數，這個便是BR_TRANSACTION_COMPLETE：
switch (cmd) { case BR_TRANSACTION_COMPLETE: if (!reply && !acquireResult) goto finish; break; ...... } reply不為NULL，因此，IPCThreadState::waitForResponse的循環沒有結束，繼續執行，又進入到IPCThreadState::talkWithDrive中。
這次，needRead就為true了，而outAvail仍為0，所以bwr.read_size不為0，bwr.write_size為0。于是通過：
ioctl(mProcess->mDriverFD, BINDER_WRITE_READ, &bwr)
進入到Binder驅動程序中的binder_ioctl函數中。由于bwr.write_size為0，bwr.read_size不為0，這次直接就進入到binder_thread_read函數中。這時候，thread->transaction_stack不等于0，但是thread->todo為空，于是線程就通過：
[cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
wait_event_interruptible(thread->wait, binder_has_thread_work(thread));
進入睡眠狀態，等待Service Manager來喚醒了。
現在，我們可以回到Service Manager被喚醒的過程了。我們接著前面淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路這篇文章的最后，繼續描述。此時， Service Manager正在binder_thread_read函數中調用wait_event_interruptible_exclusive進入休眠狀態。上面被MediaPlayerService啟動后進程喚醒后，繼續執行binder_thread_read函數：
static int binder_thread_read(struct binder_proc *proc, struct binder_thread *thread, void __user *buffer, int size, signed long *consumed, int non_block) { void __user *ptr = buffer + *consumed; void __user *end = buffer + size; int ret = 0; int wait_for_proc_work; if (*consumed == 0) { if (put_user(BR_NOOP, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); } retry: wait_for_proc_work = thread->transaction_stack == NULL && list_empty(&thread->todo); ...... if (wait_for_proc_work) { ...... if (non_block) { if (!binder_has_proc_work(proc, thread)) ret = -EAGAIN; } else ret = wait_event_interruptible_exclusive(proc->wait, binder_has_proc_work(proc, thread)); } else { ...... } ...... while (1) { uint32_t cmd; struct binder_transaction_data tr; struct binder_work *w; struct binder_transaction *t = NULL; if (!list_empty(&thread->todo)) w = list_first_entry(&thread->todo, struct binder_work, entry); else if (!list_empty(&proc->todo) && wait_for_proc_work) w = list_first_entry(&proc->todo, struct binder_work, entry); else { if (ptr - buffer == 4 && !(thread->looper & BINDER_LOOPER_STATE_NEED_RETURN)) /* no data added */ goto retry; break; } if (end - ptr < sizeof(tr) + 4) break; switch (w->type) { case BINDER_WORK_TRANSACTION: { t = container_of(w, struct binder_transaction, work); } break; ...... } if (!t) continue; BUG_ON(t->buffer == NULL); if (t->buffer->target_node) { struct binder_node *target_node = t->buffer->target_node; tr.target.ptr = target_node->ptr; tr.cookie = target_node->cookie; ...... cmd = BR_TRANSACTION; } else { ...... } tr.code = t->code; tr.flags = t->flags; tr.sender_euid = t->sender_euid; if (t->from) { struct task_struct *sender = t->from->proc->tsk; tr.sender_pid = task_tgid_nr_ns(sender, current->nsproxy->pid_ns); } else { tr.sender_pid = 0; } tr.data_size = t->buffer->data_size; tr.offsets_size = t->buffer->offsets_size; tr.data.ptr.buffer = (void *)t->buffer->data + proc->user_buffer_offset; tr.data.ptr.offsets = tr.data.ptr.buffer + ALIGN(t->buffer->data_size, sizeof(void *)); if (put_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); if (copy_to_user(ptr, &tr, sizeof(tr))) return -EFAULT; ptr += sizeof(tr); ...... list_del(&t->work.entry); t->buffer->allow_user_free = 1; if (cmd == BR_TRANSACTION && !(t->flags & TF_ONE_WAY)) { t->to_parent = thread->transaction_stack; t->to_thread = thread; thread->transaction_stack = t; } else { t->buffer->transaction = NULL; kfree(t); binder_stats.obj_deleted[BINDER_STAT_TRANSACTION]++; } break; } done: ...... return 0; } Service Manager被喚醒之后，就進入while循環開始處理事務了。這里wait_for_proc_work等于1，并且proc->todo不為空，所以從proc->todo列表中得到第一個工作項：
w = list_first_entry(&proc->todo, struct binder_work, entry);
從上面的描述中，我們知道，這個工作項的類型為BINDER_WORK_TRANSACTION，于是通過下面語句得到事務項：
t = container_of(w, struct binder_transaction, work);
接著就是把事務項t中的數據拷貝到本地局部變量struct binder_transaction_data tr中去了：
if (t->buffer->target_node) { struct binder_node *target_node = t->buffer->target_node; tr.target.ptr = target_node->ptr; tr.cookie = target_node->cookie; ...... cmd = BR_TRANSACTION; } else { ...... } tr.code = t->code; tr.flags = t->flags; tr.sender_euid = t->sender_euid; if (t->from) { struct task_struct *sender = t->from->proc->tsk; tr.sender_pid = task_tgid_nr_ns(sender, current->nsproxy->pid_ns); } else { tr.sender_pid = 0; } tr.data_size = t->buffer->data_size; tr.offsets_size = t->buffer->offsets_size; tr.data.ptr.buffer = (void *)t->buffer->data + proc->user_buffer_offset; tr.data.ptr.offsets = tr.data.ptr.buffer + ALIGN(t->buffer->data_size, sizeof(void *)); 這里有一個非常重要的地方，是Binder進程間通信機制的精髓所在：
tr.data.ptr.buffer = (void *)t->buffer->data + proc->user_buffer_offset; tr.data.ptr.offsets = tr.data.ptr.buffer + ALIGN(t->buffer->data_size, sizeof(void *));
t->buffer->data所指向的地址是內核空間的，現在要把數據返回給Service Manager進程的用戶空間，而Service Manager進程的用戶空間是不能訪問內核空間的數據的，所以這里要作一下處理。怎么處理呢？我們在學面向對象語言的時候，對象的拷貝有深拷貝和淺拷貝之分，深拷貝是把另外分配一塊新內存，然后把原始對象的內容搬過去，淺拷貝是并沒有為新對象分配一塊新空間，而只是分配一個引用，而個引用指向原始對象。Binder機制用的是類似淺拷貝的方法，通過在用戶空間分配一個虛擬地址，然后讓這個用戶空間虛擬地址與 t->buffer->data這個內核空間虛擬地址指向同一個物理地址，這樣就可以實現淺拷貝了。怎么樣用戶空間和內核空間的虛擬地址同時指向同一個物理地址呢？請參考前面一篇文章淺談Service Manager成為Android進程間通信（IPC）機制Binder守護進程之路，那里有詳細描述。這里只要將t->buffer->data加上一個偏移值proc->user_buffer_offset就可以得到t->buffer->data對應的用戶空間虛擬地址了。調整了tr.data.ptr.buffer的值之后，不要忘記也要一起調整tr.data.ptr.offsets的值。
接著就是把tr的內容拷貝到用戶傳進來的緩沖區去了，指針ptr指向這個用戶緩沖區的地址：
if (put_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); if (copy_to_user(ptr, &tr, sizeof(tr))) return -EFAULT; ptr += sizeof(tr);
這里可以看出，這里只是對作tr.data.ptr.bufferr和tr.data.ptr.offsets的內容作了淺拷貝。
最后，由于已經處理了這個事務，要把它從todo列表中刪除：
list_del(&t->work.entry); t->buffer->allow_user_free = 1; if (cmd == BR_TRANSACTION && !(t->flags & TF_ONE_WAY)) { t->to_parent = thread->transaction_stack; t->to_thread = thread; thread->transaction_stack = t; } else { t->buffer->transaction = NULL; kfree(t); binder_stats.obj_deleted[BINDER_STAT_TRANSACTION]++; } 注意，這里的cmd == BR_TRANSACTION && !(t->flags & TF_ONE_WAY)為true，表明這個事務雖然在驅動程序中已經處理完了，但是它仍然要等待Service Manager完成之后，給驅動程序一個確認，也就是需要等待回復，于是把當前事務t放在thread->transaction_stack隊列的頭部：
t->to_parent = thread->transaction_stack; t->to_thread = thread; thread->transaction_stack = t;
如果cmd == BR_TRANSACTION && !(t->flags & TF_ONE_WAY)為false，那就不需要等待回復了，直接把事務t刪掉。
這個while最后通過一個break跳了出來，最后返回到binder_ioctl函數中：
static long binder_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { int ret; struct binder_proc *proc = filp->private_data; struct binder_thread *thread; unsigned int size = _IOC_SIZE(cmd); void __user *ubuf = (void __user *)arg; ...... switch (cmd) { case BINDER_WRITE_READ: { struct binder_write_read bwr; if (size != sizeof(struct binder_write_read)) { ret = -EINVAL; goto err; } if (copy_from_user(&bwr, ubuf, sizeof(bwr))) { ret = -EFAULT; goto err; } ...... if (bwr.read_size > 0) { ret = binder_thread_read(proc, thread, (void __user *)bwr.read_buffer, bwr.read_size, &bwr.read_consumed, filp->f_flags & O_NONBLOCK); if (!list_empty(&proc->todo)) wake_up_interruptible(&proc->wait); if (ret < 0) { if (copy_to_user(ubuf, &bwr, sizeof(bwr))) ret = -EFAULT; goto err; } } ...... if (copy_to_user(ubuf, &bwr, sizeof(bwr))) { ret = -EFAULT; goto err; } break; } ...... default: ret = -EINVAL; goto err; } ret = 0; err: ...... return ret; } 從binder_thread_read返回來后，再看看proc->todo是否還有事務等待處理，如果是，就把睡眠在proc->wait隊列的線程喚醒來處理。最后，把本地變量struct binder_write_read bwr的內容拷貝回到用戶傳進來的緩沖區中，就返回了。
這里就是返回到frameworks/base/cmds/servicemanager/binder.c文件中的binder_loop函數了：
void binder_loop(struct binder_state *bs, binder_handler func) { int res; struct binder_write_read bwr; unsigned readbuf[32]; bwr.write_size = 0; bwr.write_consumed = 0; bwr.write_buffer = 0; readbuf[0] = BC_ENTER_LOOPER; binder_write(bs, readbuf, sizeof(unsigned)); for (;;) { bwr.read_size = sizeof(readbuf); bwr.read_consumed = 0; bwr.read_buffer = (unsigned) readbuf; res = ioctl(bs->fd, BINDER_WRITE_READ, &bwr); if (res < 0) { LOGE("binder_loop: ioctl failed (%s)/n", strerror(errno)); break; } res = binder_parse(bs, 0, readbuf, bwr.read_consumed, func); if (res == 0) { LOGE("binder_loop: unexpected reply?!/n"); break; } if (res < 0) { LOGE("binder_loop: io error %d %s/n", res, strerror(errno)); break; } } } 返回來的數據都放在readbuf中，接著調用binder_parse進行解析：
int binder_parse(struct binder_state *bs, struct binder_io *bio, uint32_t *ptr, uint32_t size, binder_handler func) { int r = 1; uint32_t *end = ptr + (size / 4); while (ptr < end) { uint32_t cmd = *ptr++; ...... case BR_TRANSACTION: { struct binder_txn *txn = (void *) ptr; if ((end - ptr) * sizeof(uint32_t) < sizeof(struct binder_txn)) { LOGE("parse: txn too small!/n"); return -1; } binder_dump_txn(txn); if (func) { unsigned rdata[256/4]; struct binder_io msg; struct binder_io reply; int res; bio_init(&reply, rdata, sizeof(rdata), 4); bio_init_from_txn(&msg, txn); res = func(bs, txn, &msg, &reply); binder_send_reply(bs, &reply, txn->data, res); } ptr += sizeof(*txn) / sizeof(uint32_t); break; } ...... default: LOGE("parse: OOPS %d/n", cmd); return -1; } } return r; } 首先把從Binder驅動程序讀出來的數據轉換為一個struct binder_txn結構體，保存在txn本地變量中，struct binder_txn定義在frameworks/base/cmds/servicemanager/binder.h文件中：
struct binder_txn { void *target; void *cookie; uint32_t code; uint32_t flags; uint32_t sender_pid; uint32_t sender_euid; uint32_t data_size; uint32_t offs_size; void *data; void *offs; }; 函數中還用到了另外一個數據結構struct binder_io，也是定義在frameworks/base/cmds/servicemanager/binder.h文件中：
struct binder_io { char *data; /* pointer to read/write from */ uint32_t *offs; /* array of offsets */ uint32_t data_avail; /* bytes available in data buffer */ uint32_t offs_avail; /* entries available in offsets array */ char *data0; /* start of data buffer */ uint32_t *offs0; /* start of offsets buffer */ uint32_t flags; uint32_t unused; }; 接著往下看，函數調bio_init來初始化reply變量：
void bio_init(struct binder_io *bio, void *data, uint32_t maxdata, uint32_t maxoffs) { uint32_t n = maxoffs * sizeof(uint32_t); if (n > maxdata) { bio->flags = BIO_F_OVERFLOW; bio->data_avail = 0; bio->offs_avail = 0; return; } bio->data = bio->data0 = data + n; bio->offs = bio->offs0 = data; bio->data_avail = maxdata - n; bio->offs_avail = maxoffs; bio->flags = 0; } 接著又調用bio_init_from_txn來初始化msg變量：
void bio_init_from_txn(struct binder_io *bio, struct binder_txn *txn) { bio->data = bio->data0 = txn->data; bio->offs = bio->offs0 = txn->offs; bio->data_avail = txn->data_size; bio->offs_avail = txn->offs_size / 4; bio->flags = BIO_F_SHARED; } 最后，真正進行處理的函數是從參數中傳進來的函數指針func，這里就是定義在frameworks/base/cmds/servicemanager/service_manager.c文件中的svcmgr_handler函數：
int svcmgr_handler(struct binder_state *bs, struct binder_txn *txn, struct binder_io *msg, struct binder_io *reply) { struct svcinfo *si; uint16_t *s; unsigned len; void *ptr; uint32_t strict_policy; if (txn->target != svcmgr_handle) return -1; // Equivalent to Parcel::enforceInterface(), reading the RPC // header with the strict mode policy mask and the interface name. // Note that we ignore the strict_policy and don't propagate it // further (since we do no outbound RPCs anyway). strict_policy = bio_get_uint32(msg); s = bio_get_string16(msg, &len); if ((len != (sizeof(svcmgr_id) / 2)) || memcmp(svcmgr_id, s, sizeof(svcmgr_id))) { fprintf(stderr,"invalid id %s/n", str8(s)); return -1; } switch(txn->code) { ...... case SVC_MGR_ADD_SERVICE: s = bio_get_string16(msg, &len); ptr = bio_get_ref(msg); if (do_add_service(bs, s, len, ptr, txn->sender_euid)) return -1; break; ...... } bio_put_uint32(reply, 0); return 0; } 回憶一下，在BpServiceManager::addService時，傳給Binder驅動程序的參數為：
writeInt32(IPCThreadState::self()->getStrictModePolicy() | STRICT_MODE_PENALTY_GATHER); writeString16("android.os.IServiceManager"); writeString16("media.player"); writeStrongBinder(new MediaPlayerService()); 這里的語句：
strict_policy = bio_get_uint32(msg); s = bio_get_string16(msg, &len); s = bio_get_string16(msg, &len); ptr = bio_get_ref(msg);
就是依次把它們讀取出來了，這里，我們只要看一下bio_get_ref的實現。先看一個數據結構struct binder_obj的定義：
struct binder_object { uint32_t type; uint32_t flags; void *pointer; void *cookie; }; 這個結構體其實就是對應struct flat_binder_obj的。
接著看bio_get_ref實現：
void *bio_get_ref(struct binder_io *bio) { struct binder_object *obj; obj = _bio_get_obj(bio); if (!obj) return 0; if (obj->type == BINDER_TYPE_HANDLE) return obj->pointer; return 0; } _bio_get_obj這個函數就不跟進去看了，它的作用就是從binder_io中取得第一個還沒取獲取過的binder_object。在這個場景下，就是我們最開始傳過來代表MediaPlayerService的flat_binder_obj了，這個原始的flat_binder_obj的type為BINDER_TYPE_BINDER，binder為指向MediaPlayerService的弱引用的地址。在前面我們說過，在Binder驅動驅動程序里面，會把這個flat_binder_obj的type改為BINDER_TYPE_HANDLE，handle改為一個句柄值。這里的handle值就等于obj->pointer的值。
回到svcmgr_handler函數，調用do_add_service進一步處理：
int do_add_service(struct binder_state *bs, uint16_t *s, unsigned len, void *ptr, unsigned uid) { struct svcinfo *si; // LOGI("add_service('%s',%p) uid=%d/n", str8(s), ptr, uid); if (!ptr || (len == 0) || (len > 127)) return -1; if (!svc_can_register(uid, s)) { LOGE("add_service('%s',%p) uid=%d - PERMISSION DENIED/n", str8(s), ptr, uid); return -1; } si = find_svc(s, len); if (si) { if (si->ptr) { LOGE("add_service('%s',%p) uid=%d - ALREADY REGISTERED/n", str8(s), ptr, uid); return -1; } si->ptr = ptr; } else { si = malloc(sizeof(*si) + (len + 1) * sizeof(uint16_t)); if (!si) { LOGE("add_service('%s',%p) uid=%d - OUT OF MEMORY/n", str8(s), ptr, uid); return -1; } si->ptr = ptr; si->len = len; memcpy(si->name, s, (len + 1) * sizeof(uint16_t)); si->name[len] = '/0'; si->death.func = svcinfo_death; si->death.ptr = si; si->next = svclist; svclist = si; } binder_acquire(bs, ptr); binder_link_to_death(bs, ptr, &si->death); return 0; } 這個函數的實現很簡單，就是把MediaPlayerService這個Binder實體的引用寫到一個struct svcinfo結構體中，主要是它的名稱和句柄值，然后插入到鏈接svclist的頭部去。這樣，Client來向Service Manager查詢服務接口時，只要給定服務名稱，Service Manger就可以返回相應的句柄值了。
這個函數執行完成后，返回到svcmgr_handler函數，函數的最后，將一個錯誤碼0寫到reply變量中去，表示一切正常：
bio_put_uint32(reply, 0);
svcmgr_handler函數執行完成后，返回到binder_parse函數，執行下面語句：
binder_send_reply(bs, &reply, txn->data, res);
我們看一下binder_send_reply的實現，從函數名就可以猜到它要做什么了，告訴Binder驅動程序，它完成了Binder驅動程序交給它的任務了。
void binder_send_reply(struct binder_state *bs, struct binder_io *reply, void *buffer_to_free, int status) { struct { uint32_t cmd_free; void *buffer; uint32_t cmd_reply; struct binder_txn txn; } __attribute__((packed)) data; data.cmd_free = BC_FREE_BUFFER; data.buffer = buffer_to_free; data.cmd_reply = BC_REPLY; data.txn.target = 0; data.txn.cookie = 0; data.txn.code = 0; if (status) { data.txn.flags = TF_STATUS_CODE; data.txn.data_size = sizeof(int); data.txn.offs_size = 0; data.txn.data = &status; data.txn.offs = 0; } else { data.txn.flags = 0; data.txn.data_size = reply->data - reply->data0; data.txn.offs_size = ((char*) reply->offs) - ((char*) reply->offs0); data.txn.data = reply->data0; data.txn.offs = reply->offs0; } binder_write(bs, &data, sizeof(data)); } 從這里可以看出，binder_send_reply告訴Binder驅動程序執行BC_FREE_BUFFER和BC_REPLY命令，前者釋放之前在binder_transaction分配的空間，地址為buffer_to_free，buffer_to_free這個地址是Binder驅動程序把自己在內核空間用的地址轉換成用戶空間地址再傳給Service Manager的，所以Binder驅動程序拿到這個地址后，知道怎么樣釋放這個空間；后者告訴MediaPlayerService，它的addService操作已經完成了，錯誤碼是0，保存在data.txn.data中。
再來看binder_write函數：
int binder_write(struct binder_state *bs, void *data, unsigned len) { struct binder_write_read bwr; int res; bwr.write_size = len; bwr.write_consumed = 0; bwr.write_buffer = (unsigned) data; bwr.read_size = 0; bwr.read_consumed = 0; bwr.read_buffer = 0; res = ioctl(bs->fd, BINDER_WRITE_READ, &bwr); if (res < 0) { fprintf(stderr,"binder_write: ioctl failed (%s)/n", strerror(errno)); } return res; } 這里可以看出，只有寫操作，沒有讀操作，即read_size為0。
這里又是一個ioctl的BINDER_WRITE_READ操作。直入到驅動程序的binder_ioctl函數后，執行BINDER_WRITE_READ命令，這里就不累述了。
最后，從binder_ioctl執行到binder_thread_write函數，我們首先看第一個命令BC_FREE_BUFFER：
int binder_thread_write(struct binder_proc *proc, struct binder_thread *thread, void __user *buffer, int size, signed long *consumed) { uint32_t cmd; void __user *ptr = buffer + *consumed; void __user *end = buffer + size; while (ptr < end && thread->return_error == BR_OK) { if (get_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); if (_IOC_NR(cmd) < ARRAY_SIZE(binder_stats.bc)) { binder_stats.bc[_IOC_NR(cmd)]++; proc->stats.bc[_IOC_NR(cmd)]++; thread->stats.bc[_IOC_NR(cmd)]++; } switch (cmd) { ...... case BC_FREE_BUFFER: { void __user *data_ptr; struct binder_buffer *buffer; if (get_user(data_ptr, (void * __user *)ptr)) return -EFAULT; ptr += sizeof(void *); buffer = binder_buffer_lookup(proc, data_ptr); if (buffer == NULL) { binder_user_error("binder: %d:%d " "BC_FREE_BUFFER u%p no match/n", proc->pid, thread->pid, data_ptr); break; } if (!buffer->allow_user_free) { binder_user_error("binder: %d:%d " "BC_FREE_BUFFER u%p matched " "unreturned buffer/n", proc->pid, thread->pid, data_ptr); break; } if (binder_debug_mask & BINDER_DEBUG_FREE_BUFFER) printk(KERN_INFO "binder: %d:%d BC_FREE_BUFFER u%p found buffer %d for %s transaction/n", proc->pid, thread->pid, data_ptr, buffer->debug_id, buffer->transaction ? "active" : "finished"); if (buffer->transaction) { buffer->transaction->buffer = NULL; buffer->transaction = NULL; } if (buffer->async_transaction && buffer->target_node) { BUG_ON(!buffer->target_node->has_async_transaction); if (list_empty(&buffer->target_node->async_todo)) buffer->target_node->has_async_transaction = 0; else list_move_tail(buffer->target_node->async_todo.next, &thread->todo); } binder_transaction_buffer_release(proc, buffer, NULL); binder_free_buf(proc, buffer); break; } ...... *consumed = ptr - buffer; } return 0; } 首先通過看這個語句：
get_user(data_ptr, (void * __user *)ptr)
這個是獲得要刪除的Buffer的用戶空間地址，接著通過下面這個語句來找到這個地址對應的struct binder_buffer信息：
buffer = binder_buffer_lookup(proc, data_ptr);
因為這個空間是前面在binder_transaction里面分配的，所以這里一定能找到。
最后，就可以釋放這塊內存了：
binder_transaction_buffer_release(proc, buffer, NULL);
binder_free_buf(proc, buffer);
再來看另外一個命令BC_REPLY：
int binder_thread_write(struct binder_proc *proc, struct binder_thread *thread, void __user *buffer, int size, signed long *consumed) { uint32_t cmd; void __user *ptr = buffer + *consumed; void __user *end = buffer + size; while (ptr < end && thread->return_error == BR_OK) { if (get_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); if (_IOC_NR(cmd) < ARRAY_SIZE(binder_stats.bc)) { binder_stats.bc[_IOC_NR(cmd)]++; proc->stats.bc[_IOC_NR(cmd)]++; thread->stats.bc[_IOC_NR(cmd)]++; } switch (cmd) { ...... case BC_TRANSACTION: case BC_REPLY: { struct binder_transaction_data tr; if (copy_from_user(&tr, ptr, sizeof(tr))) return -EFAULT; ptr += sizeof(tr); binder_transaction(proc, thread, &tr, cmd == BC_REPLY); break; } ...... *consumed = ptr - buffer; } return 0; } 又再次進入到binder_transaction函數：
static void binder_transaction(struct binder_proc *proc, struct binder_thread *thread, struct binder_transaction_data *tr, int reply) { struct binder_transaction *t; struct binder_work *tcomplete; size_t *offp, *off_end; struct binder_proc *target_proc; struct binder_thread *target_thread = NULL; struct binder_node *target_node = NULL; struct list_head *target_list; wait_queue_head_t *target_wait; struct binder_transaction *in_reply_to = NULL; struct binder_transaction_log_entry *e; uint32_t return_error; ...... if (reply) { in_reply_to = thread->transaction_stack; if (in_reply_to == NULL) { ...... return_error = BR_FAILED_REPLY; goto err_empty_call_stack; } binder_set_nice(in_reply_to->saved_priority); if (in_reply_to->to_thread != thread) { ....... goto err_bad_call_stack; } thread->transaction_stack = in_reply_to->to_parent; target_thread = in_reply_to->from; if (target_thread == NULL) { return_error = BR_DEAD_REPLY; goto err_dead_binder; } if (target_thread->transaction_stack != in_reply_to) { ...... return_error = BR_FAILED_REPLY; in_reply_to = NULL; target_thread = NULL; goto err_dead_binder; } target_proc = target_thread->proc; } else { ...... } if (target_thread) { e->to_thread = target_thread->pid; target_list = &target_thread->todo; target_wait = &target_thread->wait; } else { ...... } /* TODO: reuse incoming transaction for reply */ t = kzalloc(sizeof(*t), GFP_KERNEL); if (t == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_t_failed; } tcomplete = kzalloc(sizeof(*tcomplete), GFP_KERNEL); if (tcomplete == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_tcomplete_failed; } if (!reply && !(tr->flags & TF_ONE_WAY)) t->from = thread; else t->from = NULL; t->sender_euid = proc->tsk->cred->euid; t->to_proc = target_proc; t->to_thread = target_thread; t->code = tr->code; t->flags = tr->flags; t->priority = task_nice(current); t->buffer = binder_alloc_buf(target_proc, tr->data_size, tr->offsets_size, !reply && (t->flags & TF_ONE_WAY)); if (t->buffer == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_alloc_buf_failed; } t->buffer->allow_user_free = 0; t->buffer->debug_id = t->debug_id; t->buffer->transaction = t; t->buffer->target_node = target_node; if (target_node) binder_inc_node(target_node, 1, 0, NULL); offp = (size_t *)(t->buffer->data + ALIGN(tr->data_size, sizeof(void *))); if (copy_from_user(t->buffer->data, tr->data.ptr.buffer, tr->data_size)) { binder_user_error("binder: %d:%d got transaction with invalid " "data ptr/n", proc->pid, thread->pid); return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } if (copy_from_user(offp, tr->data.ptr.offsets, tr->offsets_size)) { binder_user_error("binder: %d:%d got transaction with invalid " "offsets ptr/n", proc->pid, thread->pid); return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } ...... if (reply) { BUG_ON(t->buffer->async_transaction != 0); binder_pop_transaction(target_thread, in_reply_to); } else if (!(t->flags & TF_ONE_WAY)) { ...... } else { ...... } t->work.type = BINDER_WORK_TRANSACTION; list_add_tail(&t->work.entry, target_list); tcomplete->type = BINDER_WORK_TRANSACTION_COMPLETE; list_add_tail(&tcomplete->entry, &thread->todo); if (target_wait) wake_up_interruptible(target_wait); return; ...... } 注意，這里的reply為1，我們忽略掉其它無關代碼。
前面Service Manager正在binder_thread_read函數中被MediaPlayerService啟動后進程喚醒后，在最后會把當前處理完的事務放在thread->transaction_stack中：
if (cmd == BR_TRANSACTION && !(t->flags & TF_ONE_WAY)) { t->to_parent = thread->transaction_stack; t->to_thread = thread; thread->transaction_stack = t; } 所以，這里，首先是把它這個binder_transaction取回來，并且放在本地變量in_reply_to中：
in_reply_to = thread->transaction_stack;
接著就可以通過in_reply_to得到最終發出這個事務請求的線程和進程：
target_thread = in_reply_to->from;
target_proc = target_thread->proc;
然后得到target_list和target_wait：
target_list = &target_thread->todo;
target_wait = &target_thread->wait;
下面這一段代碼：
/* TODO: reuse incoming transaction for reply */ t = kzalloc(sizeof(*t), GFP_KERNEL); if (t == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_t_failed; } tcomplete = kzalloc(sizeof(*tcomplete), GFP_KERNEL); if (tcomplete == NULL) { return_error = BR_FAILED_REPLY; goto err_alloc_tcomplete_failed; } if (!reply && !(tr->flags & TF_ONE_WAY)) t->from = thread; else t->from = NULL; t->sender_euid = proc->tsk->cred->euid; t->to_proc = target_proc; t->to_thread = target_thread; t->code = tr->code; t->flags = tr->flags; t->priority = task_nice(current); t->buffer = binder_alloc_buf(target_proc, tr->data_size, tr->offsets_size, !reply && (t->flags & TF_ONE_WAY)); if (t->buffer == NULL) { return_error = BR_FAILED_REPLY; goto err_binder_alloc_buf_failed; } t->buffer->allow_user_free = 0; t->buffer->debug_id = t->debug_id; t->buffer->transaction = t; t->buffer->target_node = target_node; if (target_node) binder_inc_node(target_node, 1, 0, NULL); offp = (size_t *)(t->buffer->data + ALIGN(tr->data_size, sizeof(void *))); if (copy_from_user(t->buffer->data, tr->data.ptr.buffer, tr->data_size)) { binder_user_error("binder: %d:%d got transaction with invalid " "data ptr/n", proc->pid, thread->pid); return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } if (copy_from_user(offp, tr->data.ptr.offsets, tr->offsets_size)) { binder_user_error("binder: %d:%d got transaction with invalid " "offsets ptr/n", proc->pid, thread->pid); return_error = BR_FAILED_REPLY; goto err_copy_data_failed; } 我們在前面已經分析過了，這里不再重復。但是有一點要注意的是，這里target_node為NULL，因此，t->buffer->target_node也為NULL。
函數本來有一個for循環，用來處理數據中的Binder對象，這里由于沒有Binder對象，所以就略過了。到了下面這句代碼：
binder_pop_transaction(target_thread, in_reply_to);
我們看看做了什么事情：
static void binder_pop_transaction( struct binder_thread *target_thread, struct binder_transaction *t) { if (target_thread) { BUG_ON(target_thread->transaction_stack != t); BUG_ON(target_thread->transaction_stack->from != target_thread); target_thread->transaction_stack = target_thread->transaction_stack->from_parent; t->from = NULL; } t->need_reply = 0; if (t->buffer) t->buffer->transaction = NULL; kfree(t); binder_stats.obj_deleted[BINDER_STAT_TRANSACTION]++; } 由于到了這里，已經不需要in_reply_to這個transaction了，就把它刪掉。
回到binder_transaction函數：
t->work.type = BINDER_WORK_TRANSACTION; list_add_tail(&t->work.entry, target_list); tcomplete->type = BINDER_WORK_TRANSACTION_COMPLETE; list_add_tail(&tcomplete->entry, &thread->todo);
和前面一樣，分別把t和tcomplete分別放在target_list和thread->todo隊列中，這里的target_list指的就是最初調用IServiceManager::addService的MediaPlayerService的Server主線程的的thread->todo隊列了，而thread->todo指的是Service Manager中用來回復IServiceManager::addService請求的線程。
最后，喚醒等待在target_wait隊列上的線程了，就是最初調用IServiceManager::addService的MediaPlayerService的Server主線程了，它最后在binder_thread_read函數中睡眠在thread->wait上，就是這里的target_wait了：
if (target_wait)
wake_up_interruptible(target_wait);
這樣，Service Manger回復調用IServiceManager::addService請求就算完成了，重新回到frameworks/base/cmds/servicemanager/binder.c文件中的binder_loop函數等待下一個Client請求的到來。事實上，Service Manger回到binder_loop函數再次執行ioctl函數時候，又會再次進入到binder_thread_read函數。這時個會發現thread->todo不為空，這是因為剛才我們調用了：
list_add_tail(&tcomplete->entry, &thread->todo);
把一個工作項tcompelete放在了在thread->todo中，這個tcompelete的type為BINDER_WORK_TRANSACTION_COMPLETE，因此，Binder驅動程序會執行下面操作：
switch (w->type) { case BINDER_WORK_TRANSACTION_COMPLETE: { cmd = BR_TRANSACTION_COMPLETE; if (put_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); list_del(&w->entry); kfree(w); } break; ...... } binder_loop函數執行完這個ioctl調用后，才會在下一次調用ioctl進入到Binder驅動程序進入休眠狀態，等待下一次Client的請求。
上面講到調用IServiceManager::addService的MediaPlayerService的Server主線程被喚醒了，于是，重新執行binder_thread_read函數：
static int binder_thread_read(struct binder_proc *proc, struct binder_thread *thread, void __user *buffer, int size, signed long *consumed, int non_block) { void __user *ptr = buffer + *consumed; void __user *end = buffer + size; int ret = 0; int wait_for_proc_work; if (*consumed == 0) { if (put_user(BR_NOOP, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); } retry: wait_for_proc_work = thread->transaction_stack == NULL && list_empty(&thread->todo); ...... if (wait_for_proc_work) { ...... } else { if (non_block) { if (!binder_has_thread_work(thread)) ret = -EAGAIN; } else ret = wait_event_interruptible(thread->wait, binder_has_thread_work(thread)); } ...... while (1) { uint32_t cmd; struct binder_transaction_data tr; struct binder_work *w; struct binder_transaction *t = NULL; if (!list_empty(&thread->todo)) w = list_first_entry(&thread->todo, struct binder_work, entry); else if (!list_empty(&proc->todo) && wait_for_proc_work) w = list_first_entry(&proc->todo, struct binder_work, entry); else { if (ptr - buffer == 4 && !(thread->looper & BINDER_LOOPER_STATE_NEED_RETURN)) /* no data added */ goto retry; break; } ...... switch (w->type) { case BINDER_WORK_TRANSACTION: { t = container_of(w, struct binder_transaction, work); } break; ...... } if (!t) continue; BUG_ON(t->buffer == NULL); if (t->buffer->target_node) { ...... } else { tr.target.ptr = NULL; tr.cookie = NULL; cmd = BR_REPLY; } tr.code = t->code; tr.flags = t->flags; tr.sender_euid = t->sender_euid; if (t->from) { ...... } else { tr.sender_pid = 0; } tr.data_size = t->buffer->data_size; tr.offsets_size = t->buffer->offsets_size; tr.data.ptr.buffer = (void *)t->buffer->data + proc->user_buffer_offset; tr.data.ptr.offsets = tr.data.ptr.buffer + ALIGN(t->buffer->data_size, sizeof(void *)); if (put_user(cmd, (uint32_t __user *)ptr)) return -EFAULT; ptr += sizeof(uint32_t); if (copy_to_user(ptr, &tr, sizeof(tr))) return -EFAULT; ptr += sizeof(tr); ...... list_del(&t->work.entry); t->buffer->allow_user_free = 1; if (cmd == BR_TRANSACTION && !(t->flags & TF_ONE_WAY)) { ...... } else { t->buffer->transaction = NULL; kfree(t); binder_stats.obj_deleted[BINDER_STAT_TRANSACTION]++; } break; } done: ...... return 0; } 在while循環中，從thread->todo得到w，w->type為BINDER_WORK_TRANSACTION，于是，得到t。從上面可以知道，Service Manager反回了一個0回來，寫在t->buffer->data里面，現在把t->buffer->data加上proc->user_buffer_offset，得到用戶空間地址，保存在tr.data.ptr.buffer里面，這樣用戶空間就可以訪問這個返回碼了。由于cmd不等于BR_TRANSACTION，這時就可以把t刪除掉了，因為以后都不需要用了。
執行完這個函數后，就返回到binder_ioctl函數，執行下面語句，把數據返回給用戶空間：
if (copy_to_user(ubuf, &bwr, sizeof(bwr))) { ret = -EFAULT; goto err; } 接著返回到用戶空間IPCThreadState::talkWithDriver函數，最后返回到IPCThreadState::waitForResponse函數，最終執行到下面語句：
status_t IPCThreadState::waitForResponse(Parcel *reply, status_t *acquireResult) { int32_t cmd; int32_t err; while (1) { if ((err=talkWithDriver()) < NO_ERROR) break; ...... cmd = mIn.readInt32(); ...... switch (cmd) { ...... case BR_REPLY: { binder_transaction_data tr; err = mIn.read(&tr, sizeof(tr)); LOG_ASSERT(err == NO_ERROR, "Not enough command data for brREPLY"); if (err != NO_ERROR) goto finish; if (reply) { if ((tr.flags & TF_STATUS_CODE) == 0) { reply->ipcSetDataReference( reinterpret_cast<const uint8_t*>(tr.data.ptr.buffer), tr.data_size, reinterpret_cast<const size_t*>(tr.data.ptr.offsets), tr.offsets_size/sizeof(size_t), freeBuffer, this); } else { ...... } } else { ...... } } goto finish; ...... } } finish: ...... return err; } 注意，這里的tr.flags等于0，這個是在上面的binder_send_reply函數里設置的。最終把結果保存在reply了：
reply->ipcSetDataReference( reinterpret_cast<const uint8_t*>(tr.data.ptr.buffer), tr.data_size, reinterpret_cast<const size_t*>(tr.data.ptr.offsets), tr.offsets_size/sizeof(size_t), freeBuffer, this);
這個函數我們就不看了，有興趣的讀者可以研究一下。
從這里層層返回，最后回到MediaPlayerService::instantiate函數中。
至此，IServiceManager::addService終于執行完畢了。這個過程非常復雜，但是如果我們能夠深刻地理解這一過程，將能很好地理解Binder機制的設計思想和實現過程。這里，對IServiceManager::addService過程中MediaPlayerService、ServiceManager和BinderDriver之間的交互作一個小結：
回到frameworks/base/media/mediaserver/main_mediaserver.cpp文件中的main函數，接下去還要執行下面兩個函數：
ProcessState::self()->startThreadPool();
IPCThreadState::self()->joinThreadPool();
首先看ProcessState::startThreadPool函數的實現：
void ProcessState::startThreadPool() { AutoMutex _l(mLock); if (!mThreadPoolStarted) { mThreadPoolStarted = true; spawnPooledThread(true); } } 這里調用spwanPooledThread：
void ProcessState::spawnPooledThread(bool isMain) { if (mThreadPoolStarted) { int32_t s = android_atomic_add(1, &mThreadPoolSeq); char buf[32]; sprintf(buf, "Binder Thread #%d", s); LOGV("Spawning new pooled thread, name=%s/n", buf); sp<Thread> t = new PoolThread(isMain); t->run(buf); } } 這里主要是創建一個線程，PoolThread繼續Thread類，Thread類定義在frameworks/base/libs/utils/Threads.cpp文件中，其run函數最終調用子類的threadLoop函數，這里即為PoolThread::threadLoop函數：
virtual bool threadLoop() { IPCThreadState::self()->joinThreadPool(mIsMain); return false; } 這里和frameworks/base/media/mediaserver/main_mediaserver.cpp文件中的main函數一樣，最終都是調用了IPCThreadState::joinThreadPool函數，它們的區別是，一個參數是true，一個是默認值false。我們來看一下這個函數的實現：
void IPCThreadState::joinThreadPool(bool isMain) { LOG_THREADPOOL("**** THREAD %p (PID %d) IS JOINING THE THREAD POOL/n", (void*)pthread_self(), getpid()); mOut.writeInt32(isMain ? BC_ENTER_LOOPER : BC_REGISTER_LOOPER); ...... status_t result; do { int32_t cmd; ....... // now get the next command to be processed, waiting if necessary result = talkWithDriver(); if (result >= NO_ERROR) { size_t IN = mIn.dataAvail(); if (IN < sizeof(int32_t)) continue; cmd = mIn.readInt32(); ...... } result = executeCommand(cmd); } ...... } while (result != -ECONNREFUSED && result != -EBADF); ....... mOut.writeInt32(BC_EXIT_LOOPER); talkWithDriver(false); } 這個函數最終是在一個無窮循環中，通過調用talkWithDriver函數來和Binder驅動程序進行交互，實際上就是調用talkWithDriver來等待Client的請求，然后再調用executeCommand來處理請求，而在executeCommand函數中，最終會調用BBinder::transact來真正處理Client的請求：
status_t IPCThreadState::executeCommand(int32_t cmd) { BBinder* obj; RefBase::weakref_type* refs; status_t result = NO_ERROR; switch (cmd) { ...... case BR_TRANSACTION: { binder_transaction_data tr; result = mIn.read(&tr, sizeof(tr)); ...... Parcel reply; ...... if (tr.target.ptr) { sp<BBinder> b((BBinder*)tr.cookie); const status_t error = b->transact(tr.code, buffer, &reply, tr.flags); if (error < NO_ERROR) reply.setError(error); } else { const status_t error = the_context_object->transact(tr.code, buffer, &reply, tr.flags); if (error < NO_ERROR) reply.setError(error); } ...... } break; ....... } if (result != NO_ERROR) { mLastError = result; } return result; } 接下來再看一下BBinder::transact的實現：
status_t BBinder::transact( uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags) { data.setDataPosition(0); status_t err = NO_ERROR; switch (code) { case PING_TRANSACTION: reply->writeInt32(pingBinder()); break; default: err = onTransact(code, data, reply, flags); break; } if (reply != NULL) { reply->setDataPosition(0); } return err; } 最終會調用onTransact函數來處理。在這個場景中，BnMediaPlayerService繼承了BBinder類，并且重載了onTransact函數，因此，這里實際上是調用了BnMediaPlayerService::onTransact函數，這個函數定義在frameworks/base/libs/media/libmedia/IMediaPlayerService.cpp文件中：
status_t BnMediaPlayerService::onTransact( uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags) { switch(code) { case CREATE_URL: { ...... } break; case CREATE_FD: { ...... } break; case DECODE_URL: { ...... } break; case DECODE_FD: { ...... } break; case CREATE_MEDIA_RECORDER: { ...... } break; case CREATE_METADATA_RETRIEVER: { ...... } break; case GET_OMX: { ...... } break; default: return BBinder::onTransact(code, data, reply, flags); } } 至此，我們就以MediaPlayerService為例，完整地介紹了Android系統進程間通信Binder機制中的Server啟動過程。Server啟動起來之后，就會在一個無窮循環中等待Client的請求了。在下一篇文章中，我們將介紹Client如何通過Service Manager遠程接口來獲得Server遠程接口，進而調用Server遠程接口來使用Server提供的服務，敬請關注。
新聞熱點
疑難解答
