前面一节我们分析了Binder通信相关的两个重要类:ProcessState 和 IPCThreadState。ProcessState负责打开Binder 驱动,每个进程只有一个。而 IPCThreadState负责提供与Binder通信相关的接口,每个线程有一个。下面我们通过具体 示例MediaPlayerService来分析我们应用程序中怎么通过Binder通信的。
frameworks/base/media/mediaserver/main_mediaserver.cpp
int main(int argc, char*argv[])
{
sp<ProcessState> proc(ProcessState)::self(); // 获得ProcessState在构造函数中打开binder驱动
sp<IServiceManager> sm = defaultServiceManager();
MediaPlayService::instantiate();
ProcessState::self()->startThreadPool();
IPCThreadState::self()->joinThreadPool();
}
1)获得ServiceManager的代理BpServiceManager
sp<IServiceManager> sm = defaultServiceManager();
sp<IServiceManager> defaultServiceManager()
{
if(gDefaultServiceManager != NULL) return gDefaultServiceManager;
{
AutoMutex -l(gDefaultServiceManagerLock);
if(gDefaultServiceManager == NULL)
gDefaultServiceManager = interface_cast<IServiceManager>(
ProcessState::self()->getContextObject(NULL));
}
return gDefaultServiceManager;
}
这里又是一个单例模式,每个进程只需要一个BpServiceManager代理,通过interface_cast获得。 首先看看ProcessState::self()->getContextObject(NULL)
sp<IBinder> ProcessState::getContextObject(const sp<IBinder>& caller)
{
return getStrongProxyForHandle(0);
}
sp<IBinder> ProcessState::getStrongProxyForHandle(int32_t handle)
{
sp<IBinder> result;
AutoMutex _l(mLock);
handle_entry *e = lookupHandleLocked(handle);
if( e != NULL) {
IBinder* b = e->binder;
if(b == NULL || !e->refs->attemptIncWeak(this)) {
b = new BpBinder(handle);
e->binder = b;
if(b) e->refs = b->getWeakRefs();
result = b;
}else{
result.force_set(b);
e->refs->decWeak(this);
}
}
return result;
}
struct handle_entry{
IBinder* binder;
RefBase::weakref_type* refs;
}
ProcessState::handle_entry* ProcessState::lookupHandleLocked()从数组mHandleToObject里面根据handle索引,查找 一个handle_entry结构体。然后根据传入的句柄handle这里为0,表示ServiceManager,new一个BpBinder 所以现在相当于: gDefaultServiceManager = interface_cast<IServiceManager>(new BpBinder(0));
现在我们看看interface_cast是什么?
frameworks/base/include/binder/IInterface.h
template<typename INTERFACE>
inline sp<INTERFACE> interface_cast(const sp<IBinder>& obj)
{
return INTERFACE::asInterface(obj);
}
等价于:
inline sp<IServiceManager> interface_cast(const sp<IBinder>& obj)
{
return IServiceManager::asInterface(obj);
}
继续我们跟到IServiceManager里面去:
frameworks/base/include/binder/IServiceManager.h
class IServiceManager:public IInterface
{
public:
DECLARE_META_INTERFACE(ServiceManager);// MLGB的又是宏!!!
virtual status_t addService(const String16& name, const sp<IBinder>& service) = 0;
virtual sp<IBinder> getService(const String16& name) const = 0;
}
#define DECLARE_META_INTERFACE(INTERFACE) \
static const android::String16 descriptor; \
static android::sp<I##INTERFACE> asInterface( \
const android::sp<android::IBinder>& obj); \
virtual const android::String16& getInterfaceDescriptor() const;\
I##INTERFACE(); \
virtual !I##INTERFACE();
替换之后就是:
static const android::String16 descriptor;
static android::sp<IServiceManager> asInterface(
const android::sp<android::IBinder>& obj);
virtual const android::String16& getInterfaceDescriptor() const;
IServiceManager();
virtual !IServiceManager();
都是一些函数声明,既然有声明的地方,肯定有实现的地方了。
#define IMPLEMENT_META_INTERFACE(INTERFACE, NAME) \
const android::String16 I##INTERFACE::descriptor(NAME); \
const android::String16& \
I##INTERFACE::getInterfaceDescriptor() const { \
return I##INTERFACE::descriptor; \
} \
android::sp<I##INTERFACE> I##INTERFACE::asInterface( \
const android::sp<android::IBinder>& obj) \
{ \
android::sp<I##INTERFACE> intr; \
if (obj != NULL) { \
intr = static_cast<I##INTERFACE*>( \
obj->queryLocalInterface( \
I##INTERFACE::descriptor).get()); \
if (intr == NULL) { \
intr = new Bp##INTERFACE(obj); \
} \
} \
return intr; \
} \
I##INTERFACE::I##INTERFACE() { } \
I##INTERFACE::~I##INTERFACE() { }
继续替换:
{
const android::String16 IServiceManager::descriptor(NAME);
const android::String16&
IServiceManager::getInterfaceDescriptor() const {
return IServiceManager::descriptor;
}
android::sp<IServiceManager> IServiceManager::asInterface(
const android::sp<android::IBinder>& obj) // 参数为new BpBinder(0)
{
android::sp<IServiceManager> intr;
if (obj != NULL) {
intr = static_cast<IServiceManager*>(
obj->queryLocalInterface(
IServiceManager::descriptor).get());
if (intr == NULL) {
intr = new BpServiceManager(obj); // 原来在这里new 了一个BpServiceManager对象
}
}
return intr;
}
IServiceManager::IServiceManager() { }
IServiceManager::~IServiceManager() { }
}
总结:根据句柄handle 0 创建一个new BpBinder(0),根据这个BpBinder创建了一个BpServiceManager代理。
下面来看看BpServiceManager代理:
class BpServiceManager : public BpInterface<IServiceManager>
{
public:
BpServiceManager(const sp<IBinder>& impl) : BpInterface<IServiceManager>(iml)
{}
}
这里BpInterface是一个模板类,表示这里BpServiceManager同时继承与BpInterface和IServiceManager类
template<typename INTERFACE>
class BpInterface : public INTERFACE, public BpRefBase
{
public: BpInterface(const sp<IBinder>& remote);
...
}
调用了基类BpInterface构造函数:
BpInterface<IServiceManager>::BpInterace(const sp<IBinder>& remote) : BpRefBase(remote)
{}
//这里的remote就是刚刚的new BpBinder(0)
BpRefBase::BpRefBase(const sp<IBinder>& o) : mRemote(o.get()),mRefs(NULL), mState(0)
{
}
2)添加服务 MediaPlayerService::instantiate();
frameworks/base/media/libmediaplayerservice/ibMediaPlayerService.cpp
void MediaPlayerService::instantiate()
{
defaultServiceManager()->addService(String16("media.player"), new MediaPlayerService);
}
defaultServiceManager()返回的是刚创建的BpServiceManager,调用add函数。 BpMediaPlayService作为服务代理端,那么BnMediaPlayerService一定是实现端,MediaPlayerService继承于 BnMediaPlayerService,实现了真正的业务函数。
来看看BpServiceManager的addService()函数:
virtual status_t addService(const String16& name, const sp<IBinder>& service)
{
Parcel data, reply;
data.writeInterfaceToken(IServiceManager.getInterfaceDescriptor()); // android.os.IServiceManager
data.writeString16(name); // media.player
data.writeStrongBinder(service); // 也就是MediaPlayerService
status_t err = remote()->transact(ADD_SERVICE_TRANSACTION, data, &reply);
return err == NO_ERROR ? reply.readInt32() : err;
}
这里remote()就是前面创建的BpBinder(0)对象。
status_t BpBinder::transact(uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags)
{
IPCThreadState::self()->transact(mHandle, code, data, reply, flags);
}
status_t IPCThreadState::transact(int32_t handle, uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags)
{
// 发送ADD_SERVICE_TRANSACTION请求
writeTransactionData(BC_TRANSACTION, flags, handle, code, data, NULL);
if(reply) // 等待响应
waitForResponse(NULL, reply);
}
status_t IPCThreadState::writeTransactionData(int32_t cmd, uint32_t binderFlags, int32_t handle,
uint32_t code, const Parcel& data, status_t *statusBuffer)
{
// cmd BC_TRANSACTION 应用程序向BINDER发送的命令
binder_transaction_data tr;
tr.target.handle = handle; // 0
tr.code = code; // ADD_SERVICE_TRANSACTION
tr.flags = binderFlags;
// 把命令和数据一起发送到 Parcel mOut中
mOut.writeInt32(cmd);
mOut.write(&tr, sizeof(tr));
}
status_t IPCThreadState::waitForResponse(Parcel* reply, status_t *acquireResult)
{
int32_t cmd;
while(1)
talkWithDriver();
cmd = mIn.readInt32();
switch(cmd) {
case BR_TRANSACTION_COMPLETE:
...
break;
}
{
return err;
}
status_t IPCThreadState::talkWithDriver(bool doReceive)
{
binder_write_read bwr;
bwr.write_size = outAvail;
bwr.write_buf = (long unsigned int)mOut.data(); // 写入mOut的数据
bwr.read_size = mIn.dataCapacity;
bwr.read_buffer = (long unsigned int)mIn.data();
ioctl(mProcess->mDriverFD, BINDER_WRITE_READm &bwr); // 把mOut写到Binder,并读取mIn数据
}
3)IPCThreadState::joinThreadPool(), ProcessState::self()->startThreadPool() 进入线程循环talkWithDriver 等待客户端Client请求,从Binder读取命令请求进行处理。
到现在为止MediaPlayerService的服务端已经向服务总管ServiceManager注册了,下面我们看看客户端是如何获得服务的代理并和服务端通信的。 我们以MediaPlayer的业务函数decode解析播放一个URL为例
sp<IMemory> MediaPlayer::decode(const char*url, uint32_t *pSampleRate, ...)
{
sp<IMemory> p;
const sp<IMediaPlayerService>& service = getMediaPlayerService(); // 获得BpMediaPlayerSerivce代理
if(service != 0)
p = service->decode(url, ....);
return p;
}
这里我们主要分析getMediaPlayerService,客户端是如何向ServiceManager总管查询服务并获得代理的。
sp<IMediaPlayerService>& IMediaDeathNotifier::getMediaPlayerService()
{
sp<IServiceManager> sm = defaultServiceManager(); // 生成一个BpServiceManager代理对象
sp<IBinder> binder;
do {
binder = sm->getService(String16("media.player"));
if(binder != 0)
break;
usleep(500000)
} while(true);
sMediaPlayerService = interface_cast<IMediaPlayerService>(binder);
return sMediaPlayerService;
}
1)首先获得BpServiceManager的代理,然后调用getService()函数向服务总管ServiceManager查询服务。 frameworks/base/libs/binder/IServiceManager.cpp
class BpServiceManager : public BpInterface<IServiceManager>
{
public:
virtual sp<IBinder> getService(const String16& name) const
{
for(n = 0; n < 5; n++) {
sp<IBinder> svc = checkService(name); // 调用checkService函数
if(svc != NULL) return svc;
sleep(1);
}
return NULL;
}
virtual sp<IBinder> checkService(const String16& name) const
{
Parcel data, reply;
data.writeInterfaceToken(IServiceManager::getInterfaceDescriptor());
// 首先调用data.writeInt32(IPCThreadState::self()->getStrictModePolicy())
// 然后再写入android.os.IServiceManager
data.writeString16(name); // 写入 media.player
remote()->transact(CHECK_SERVICE_TRANSACTION, data, &reply);
return reply.readStrongBinder();
}
}
这里首先将请求打包成Parcel各式,然后调用remote()->transact()函数,前面我们分析过BpServiceManager::remote()返回 的就是前面new BpBinder(0)对应句柄为ServiceManager。继续去BpBinder中寻找实现代码: frameworks/base/libs/binder/BpBinder.cpp
status_t BpBinder::transact(uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags)
{
IPCThreadState::self()->transact(mHandle, code, data, reply, flags);
}
最后调用的IPCThreadState的transact()函数,IPCThreadState是专门提供通过Binder进程间通信的接口的。
status_t IPCTheadState::transact(int32_t handle, uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags)
{
// 填充binder_transaction_data 结构体,写入到mOut中去
writeTransactionData(BC_TRANSACTION, flags, handle, code, data, NULL);
// 调用talkWithDriver() 将mOut写到Binder驱动,并从Binder驱动读取mIn数据
waitForResponse(reply);
}
首先通过writeTransactionData函数来填充mOut结构体,mOut里面内容为: mOut.writeInt32(BC_TRANSACTION); mOut.write(&tr, sizeof(tr)); 这里binder_transaction_data tr内容为: tr.target.handle = 0; // 表面是发往ServiceManager的 tr.code = CHECK_SERVICE_TRANSACTION; tr.flags = 0; tr.data内容为: data.writeInt32(IPCThreadState::self()->getStrictModePolicy() | STRICT_MODE_PENALTY_GATHER); data.writeString16("android.os.IServiceManager"); data.writeString16("media.player"); 根据前面Android开发之ServiceManager一章中我们分析,svcmgr_handler处理从句柄为0的Binder的请求: strict_policy = bio_get_string32(); s = bio_get_string16(); // 就是上面的android.os.IServiceManager s = bio_get_string16(); // 就是上面的 media.player 根据media.player遍历全局链表svclist找到相应的服务,调用bio_put_ref(reply, ptr) 返回目标Binder实体。
这个waitForResponse()函数是关键:
status_t IPCThreadState::waitForResponse(Parcel* reply)
{
while(1) {
talkWithDriver(); // 输入mOut 输出mIn
cmd = mIn.readInt32();
switch(cmd) {
case BR_REPLY:
{
binder_transaction_data tr;
mIn.read(&tr, sizeof(tr));
if(reply) {
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(sizt_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(sizt_t), freeBuffer, this)
}
}
}
}
}
最后返回的是:return reply.readStrongBinder();进入到Parcel的readStrongBinder()函数
sp<IBinder> Parcel::readStrongBinder() const
{
sp<IBinder> val;
unflatten_binder(ProcessState::self(), *this, &val);
return val;
}
status_t unflatten_binder(const sp<ProcessState>& proc, const Parcel& in, sp<IBinder>* out)
{
const flat_binder_object* flat = in.readObject(false);
if(flat) {
switch(flat->type) {
case BINDER_TYPE_BINDER:
*out = static_cast<IBinder*>(flat->cookie);
return finish_unflatten_binder(NULL, *flat, in);
case BINDER_TYPE_HANDLE:
*out = proc->getStrongProxyForHandle(flat->handle);
return finish_unflatten_binder(static_cast<BpBinder*>(out->get()), *flat, in);
}
}
}
这里flat->type是BINDER_TYPE_HANDLE,所以调用ProcessState::getStrongProxyForHandle()函数
sp<IBinder> ProcessState::getStrongProxyForHandle(int32_t handle)
{
sp<IBinder> result;
handle_entry* e = lookupHandleLocked(handle);
if(e != NULL) {
IBinder* b = e->binder;
if(b == NULL || !e->refs->attemptIncWeak(this)) {
b = new BpBinder(handle);
e->binder = b;
if( b ) e->refs = e->getWeakRefs();
result = b;
} else {
result.force_set(b);
e->refs->decWeak(this);
}
}
return result;
}
这里的handle就是ServiceManager内维护的MediaPlayerService对应的Binder句柄,这个ProcessState根据这个句柄 new 了一个BpBinder,并将其保存起来,这样下次需要从ServiceManager请求获取到相同句柄的时候就可以直接返回了。 最后根据这个返回的BpBinder获得MediaPlayerService的代理: sMediaPlayerService = interface_cast<IMediaPlayerService>(binder); 根据前面ServiceManager一样,最后调用的是IMediaPlayerService的asInterface()宏函数
android::sp<IMediaPlayerService> IMediaPlayerService::asInterface(const android::sp<android::IBinder>& obj)
{
android::sp<IMediaPlayerService> intr;
if(obj != NULL ) {
intr = static_cast<IMediaPlayerService>(
obj->queryLocalInterface(IMediaPlayerService::descriptor).get);
if (intr == NULL) {
intr = new BpMediaPlayerService(obj);
}
}
return intr;
}
这样我就获得了一个代理BpMediaPlayerService对象,它的remote()为BpBinder(handle),这个handle就是向服务总共ServiceManager 查询到的MediaPlayerService对应的Binder句柄。
下一章我们分析,客户端如何通过这个BpServiceManager代理对象调用服务端MediaPlayerService的业务函数的?
原文二:Android的MediaPlayer分析
链接:http://www.byywee.com/page/M0/S846/846697.html
前段时候接到任务,商量一下MediaPlayer在Android体系中是如何实现的。到如今为止终于有些端倪了,在查询拜访代码时辰固然有些吃力,然则还算是有所得的。今天把本身对MediaPlayer的粗浅懂得拿出来,一来作为笔记备份;一来和大师分享。以下代码都是以Android4.2代码为根蒂根基的,迎接大师斧正,共同窗习。
为了有一个完全的浏览代码的脉络和思路,我是以一个很是简单的Java应用法度开端。法度很是简单,代码如下:
1 MediaPlayer mediaPlayer = new MediaPlayer();
2 mediaPlayer.setDataSource("/sdcard/test.mp3");
3 mediaPlayer.prepare();
4 mediaPlayer.start();
5 mediaPlayer.stop();
创建MediaPlayer对象
从第一行代码开端看:
MediaPlayer mediaPlayer = new MediaPlayer();这行代码是在Java应用法度中,功能是新建一个对象。既然如此,我们就看看MediaPlayer这个类在Java层的机关函数,如下:
[/frameworks/base/media/java/android/media/MediaPlayer.java]
1 public MediaPlayer() {
2
3 Looper looper;
4 if ((looper = Looper.myLooper()) != null) {
5 mEventHandler = new EventHandler(this, looper);
6 } else if ((looper = Looper.getMainLooper()) != null) {
7 mEventHandler = new EventHandler(this, looper);
8 } else {
9 mEventHandler = null;
10 }
11
12 /* Native setup requires a weak reference to our object.
13 * It""s easier to create it here than in C++.
14 */
15 native_setup(new WeakReference<MediaPlayer>(this));
16 }
在这个机关函数中,较为首要的就是native_setup(). 这个函数在之前的声明中包含native,申明这是一个native函数。也就是说native_setup()函数其实native层实现的。接下来我们要做的就是按照JNI找到这个函数是如何实现的。native函数实现地点的文件的文件名都是如下定名的,把包名中的"."调换为"_"+类名。所以我们要找文件就是android_media_MediaPlayer.cpp.在这个文件中我们有个首要的数组,如下:
【/frameworks/base/media/jni/android_media_MediaPlayer.cpp】
1 static JNINativeMethod gMethods[] = {
2 {
3 "_setDataSource",
4 "(Ljava/lang/String;[Ljava/lang/String;[Ljava/lang/String;)V",
5 (void *)android_media_MediaPlayer_setDataSourceAndHeaders
6 },
7
8 {"setDataSource", "(Ljava/io/FileDescriptor;JJ)V", (void *)android_media_MediaPlayer_setDataSourceFD},
9 {"_setVideoSurface", "(Landroid/view/Surface;)V", (void *)android_media_MediaPlayer_setVideoSurface},
10 {"prepare", "()V", (void *)android_media_MediaPlayer_prepare},
11 {"prepareAsync", "()V", (void *)android_media_MediaPlayer_prepareAsync},
12 {"_start", "()V", (void *)android_media_MediaPlayer_start},
13 {"_stop", "()V", (void *)android_media_MediaPlayer_stop},
14 {"getVideoWidth", "()I", (void *)android_media_MediaPlayer_getVideoWidth},
15 {"getVideoHeight", "()I", (void *)android_media_MediaPlayer_getVideoHeight},
16 {"seekTo", "(I)V", (void *)android_media_MediaPlayer_seekTo},
17 {"_pause", "()V", (void *)android_media_MediaPlayer_pause},
18 {"isPlaying", "()Z", (void *)android_media_MediaPlayer_isPlaying},
19 {"getCurrentPosition", "()I", (void *)android_media_MediaPlayer_getCurrentPosition},
20 {"getDuration", "()I", (void *)android_media_MediaPlayer_getDuration},
21 {"_release", "()V", (void *)android_media_MediaPlayer_release},
22 {"_reset", "()V", (void *)android_media_MediaPlayer_reset},
23 {"setAudioStreamType", "(I)V", (void *)android_media_MediaPlayer_setAudioStreamType},
24 {"setLooping", "(Z)V", (void *)android_media_MediaPlayer_setLooping},
25 {"isLooping", "()Z", (void *)android_media_MediaPlayer_isLooping},
26 {"setVolume", "(FF)V", (void *)android_media_MediaPlayer_setVolume},
27 {"getFrameAt", "(I)Landroid/graphics/Bitmap;", (void *)android_media_MediaPlayer_getFrameAt},
28 {"native_invoke", "(Landroid/os/Parcel;Landroid/os/Parcel;)I",(void *)android_media_MediaPlayer_invoke},
29 {"native_setMetadataFilter", "(Landroid/os/Parcel;)I", (void *)android_media_MediaPlayer_setMetadataFilter},
30 {"native_getMetadata", "(ZZLandroid/os/Parcel;)Z", (void *)android_media_MediaPlayer_getMetadata},
31 {"native_init", "()V", (void *)android_media_MediaPlayer_native_init},
32 {"native_setup", "(Ljava/lang/Object;)V", (void *)android_media_MediaPlayer_native_setup},
33 {"native_finalize", "()V", (void *)android_media_MediaPlayer_native_finalize},
34 {"getAudioSessionId", "()I", (void *)android_media_MediaPlayer_get_audio_session_id},
35 {"setAudioSessionId", "(I)V", (void *)android_media_MediaPlayer_set_audio_session_id},
36 {"setAuxEffectSendLevel", "(F)V", (void *)android_media_MediaPlayer_setAuxEffectSendLevel},
37 {"attachAuxEffect", "(I)V", (void *)android_media_MediaPlayer_attachAuxEffect},
38 {"native_pullBatteryData", "(Landroid/os/Parcel;)I", (void *)android_media_MediaPlayer_pullBatteryData},
39 {"setParameter", "(ILandroid/os/Parcel;)Z", (void *)android_media_MediaPlayer_setParameter},
40 {"getParameter", "(ILandroid/os/Parcel;)V", (void *)android_media_MediaPlayer_getParameter},
41 {"native_setRetransmitEndpoint", "(Ljava/lang/String;I)I", (void *)android_media_MediaPlayer_setRetransmitEndpoint},
42 {"setNextMediaPlayer", "(Landroid/media/MediaPlayer;)V", (void *)android_media_MediaPlayer_setNextMediaPlayer},
43 };
在这个数组中,第一列默示的是Java层中函数名,第二列是Java层函数对应的参数以及返回值类型,第三列就是对应在Native层的函数名。从这个数组中,我们可以找到native_setup(),对应的函数名是android_media_MediaPlayer_native_setup,接下来我们就看看这个函数的实现:
【/frameworks/base/media/jni/android_media_MediaPlayer.cpp】
1 static void
2 android_media_MediaPlayer_native_setup(JNIEnv *env, jobject thiz, jobject weak_this)
3 {
4 ALOGV("native_setup");
5 sp<MediaPlayer> mp = new MediaPlayer();
6 if (mp == NULL) {
7 jniThrowException(env, "java/lang/RuntimeException", "Out of memory");
8 return;
9 }
10
11 // create new listener and give it to MediaPlayer
12 sp<JNIMediaPlayerListener> listener = new JNIMediaPlayerListener(env, thiz, weak_this);
13 mp->setListener(listener);
14
15 // Stow our new C++ MediaPlayer in an opaque field in the Java object.
16 setMediaPlayer(env, thiz, mp);
17 }
在本文中,对于解析MediaPlayer流程较为首要的函数,我都邑用红色字体标识出来。这不是说没有标红的就不首要,在法度里没有无用的代码,只是对于我们此次的流程解析不太首要罢了。先看第5行
sp<MediaPlayer> mp = new MediaPlayer();
这行代码的功能就是创建一个MediaPlayer类的对象,不过这个对象是Native层的对象,也就是说这个Native层的MediaPlayer类是应用C++代码实现的。我们先往下看下面一句较为首要的代码,然后在后头解析Native层MediaPlayer对象的机关函数。
16 setMediaPlayer(env, thiz, mp);
这行代码的功能就是把我们新创建的Native层的MediaPlayer对象保存到Java层。也就是说将来我们经由过程getMediaplayer()的时辰获取到的就是这个对象。接下来,我就开端解析Native层的MediaPlayer的机关函数,代码如下:
1 MediaPlayer::MediaPlayer()
2 {
3 ALOGV("constructor");
4 mListener = NULL;
5 mCookie = NULL;
6 mStreamType = AUDIO_STREAM_MUSIC;
7 mCurrentPosition = -1;
8 mSeekPosition = -1;
9 mCurrentState = MEDIA_PLAYER_IDLE;
10 mPrepareSync = false;
11 mPrepareStatus = NO_ERROR;
12 mLoop = false;
13 mLeftVolume = mRightVolume = 1.0;
14 mVideoWidth = mVideoHeight = 0;
15 mLockThreadId = 0;
16 mAudioSessionId = AudioSystem::newAudioSessionId();
17 AudioSystem::acquireAudioSessionId(mAudioSessionId);
18 mSendLevel = 0;
19 mRetransmitEndpointValid = false;
20 }
在机关Native层的MediaPlayer对象的时辰,也会机关父类的对象。只不过在这些对象机关过程中,对于我们解析MediaPlayer流程并没有希罕首要的器材。然则我们要懂得的一点就是在MediaPlayer的父类IMediaDeathNotifier中有个很首要的办法getMediaPlayerService(),对于这个办法我们在之后的解析过程中还会用到,到时辰在具体介绍。
到这里为止,这一末节根蒂根基停止。总结一下,在MediaPlayer的初始化过程中,就是创建了MediaPlayer的对象。
首要的函数setDataSource()
在写这篇blog之前,已经对MediaPlayer的流程有了可能的懂得。如今回头再看,发明这个函数真的很是首要。在这个函数中,MediaPlayer和MediaPlayerService建树了接洽,MediaPlayerService和Stagefright或者OpenCore也建树了接洽。 我们持续RFCD,在Java层,我们应用的办法如下:
mediaPlayer.setDataSource("/sdcard/test.mp3");
在Java层这个函数代码如下:
【/frameworks/base/media/java/android/media/MediaPlayer.java】
1 public void setDataSource(String path)
2 throws IOException, IllegalArgumentException, SecurityException, IllegalStateException {
3 setDataSource(path, null, null);
4 }
5 调用setDataSource()的重载办法,之后的重载调用过程我们就略过,终极会调用到如下办法:
6 public void setDataSource(FileDescriptor fd)
7 throws IOException, IllegalArgumentException, IllegalStateException {
8 // intentionally less than LONG_MAX
9 setDataSource(fd, 0, 0 x7ffffffffffffffL);
10 }
11 public native void setDataSource(FileDescriptor fd, long offset, long length)
12 throws IOException, IllegalArgumentException, IllegalStateException;
所以这个函数终极还是调用到类Native层。在之后的过程中,我们都邑跳过对于JNI层的寻找过程,直接进入Native层的MediaPlayer对象寻找响应的办法。在Native层中setDataSource()对应的代码如下:
【/frameworks/av/media/libmedia/mediaplayer.cpp】
1 status_t MediaPlayer::setDataSource(int fd, int64_t offset, int64_t length)
2 {
3 ALOGV("setDataSource(%d, %lld, %lld)", fd, offset, length);
4 status_t err = UNKNOWN_ERROR;
5 const sp<IMediaPlayerService>& service(getMediaPlayerService());
6 if (service != 0) {
7 sp<IMediaPlayer> player(service->create(getpid(), this, mAudioSessionId));
8 if ((NO_ERROR != doSetRetransmitEndpoint(player)) ||
9 (NO_ERROR != player->setDataSource(fd, offset, length))) {
10 player.clear();
11 }
12 err = attachNewPlayer(player);
13 }
14 return err;
15 }
上方这段代码重点处所有三个,都已经标示出来。先看第一个,getMediaPlayerService(),这个函数从其名字上看是获取MediaPlayerService。是以
5 const sp<IMediaPlayerService>& service(getMediaPlayerService());
的功能就是获取MediaPlayerService并且把返回值赋值给service。关于getMediaPlayerService函数,我们在前面说过一次,这个办法是MediaPlayer的父类IMediaDeathNotifier的一个办法,关系到MediaPlayer和MediaPlayerService之间的通信。下面我们就看看这个办法的代码:
[/frameworks/av/media/libmedia/IMediaDeathNotifier.cpp]
1 // establish binder interface to MediaPlayerService
2 /*static*/const sp<IMediaPlayerService>&
3 IMediaDeathNotifier::getMediaPlayerService()
4 {
5 ALOGV("getMediaPlayerService");
6 Mutex::Autolock _l(sServiceLock);
7 if (sMediaPlayerService == 0) {
8 sp<IServiceManager> sm = defaultServiceManager();
9 sp<IBinder> binder;
10 do {
11 binder = sm->getService(String16("media.player"));
12 if (binder != 0) {
13 break;
14 }
15 ALOGW("Media player service not published, waiting...");
16 usleep(500000); // 0.5 s
17 } while (true);
18
19 if (sDeathNotifier == NULL) {
20 sDeathNotifier = new DeathNotifier();
21 }
22 binder->linkToDeath(sDeathNotifier);
23 sMediaPlayerService = interface_cast<IMediaPlayerService>(binder);
24 }
25 ALOGE_IF(sMediaPlayerService == 0, "no media player service!?");
26 return sMediaPlayerService;
27 }
在getMediaPlayerService()函数中,重点内容有三处,不过都是为了最后获取MediaPlayerService办事的。第8行是获取ServiceManager;第11行是从ServiceManager中获取到我们所须要的一个对象,这个对象是BpBinder类型,为我们最后取得MediaPlayerService作筹办。第23行是获取MediaPlayerService,不过终极获得这个对象是BpMediaPlayerService类型。我们获取到得这个BpMediaPlayerService,这个对象的所有操纵终极都邑由真正的MediaPlayerService对象来完成,是以完全可以看做是MediaPlayerService。这个过程和Binder通信相干,不是我们此次商量的重点,这里不再赘述。
到这里,我们解析完成了getMediaPlayerService办法,获得了返回值,并把它赋值给了service。接下来我们就要看看MediaPlayer应用这个service都做了些什么工作。从代码中看,统共有两件工作,分别是:
1)按照过程ID,对象本身,AudioSessionId这三个参数,在server端创建了一个Player对象并返回给MediaPlayer端的player。这个Player也是具有过程间通信才能的。
7 sp<IMediaPlayer> player(service->create(getpid(), this, mAudioSessionId));
2)为server端的Player设置播放资料
player->setDataSource(fd, offset, length)
我们开端看看第一件工作时如何完成的,以及在Server段创建player的时辰都做了那些工作。这个Create操纵终极会在Server端履行,所以我们就去MediaPlayerService段在这个过程中都做了些什么,代码如下:
[/frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp]
1 sp<IMediaPlayer> MediaPlayerService::create(pid_t pid, const sp<IMediaPlayerClient>& client,
2 int audioSessionId)
3 {
4 int32_t connId = android_atomic_inc(&mNextConnId);
5
6 sp<Client> c = new Client(
7 this, pid, connId, client, audioSessionId,
8 IPCThreadState::self()->getCallingUid());
9
10 ALOGV("Create new client(%d) pid %d, uid %d, ", connId, pid,
11 IPCThreadState::self()->getCallingUid());
12
13 wp<Client> w = c;
14 {
15 Mutex::Autolock lock(mLock);
16 mClients.add(w);
17 }
18 return c;
19 }
从代码中,我们可以看出,在Server端,我们新建的对象是Client,是MediaPlayerService的内部类。然则这个Client履行了MediaPlayerService的绝大项目组操纵。同样我们将来传输过来的数据也都是会交给内部类Client进行处理惩罚的。到这里,在Server端的对象就创建完成了。
接下来就是要做第二件工作了,就是在MediaPlayer中调用player->setDataSource();,这件事还是要在Server完成的,代码如下:
[/frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp]
1 status_t MediaPlayerService::Client::setDataSource(int fd, int64_t offset, int64_t length)
2 {
3 ALOGV("setDataSource fd=%d, offset=%lld, length=%lld", fd, offset, length);
4 struct stat sb;
5 int ret = fstat(fd, &sb);
6 if (ret != 0) {
7 ALOGE("fstat(%d) failed: %d, %s", fd, ret, strerror(errno));
8 return UNKNOWN_ERROR;
9 }
10
11 ALOGV("st_dev = %llu", sb.st_dev);
12 ALOGV("st_mode = %u", sb.st_mode);
13 ALOGV("st_uid = %lu", sb.st_uid);
14 ALOGV("st_gid = %lu", sb.st_gid);
15 ALOGV("st_size = %llu", sb.st_size);
16
17 if (offset >= sb.st_size) {
18 ALOGE("offset error");
19 ::close(fd);
20 return UNKNOWN_ERROR;
21 }
22 if (offset + length > sb.st_size) {
23 length = sb.st_size - offset;
24 ALOGV("calculated length = %lld", length);
25 }
26
27 player_type playerType = MediaPlayerFactory::getPlayerType(this,
28 fd,
29 offset,
30 length);
31 sp<MediaPlayerBase> p = setDataSource_pre(playerType);
32 if (p == NULL) {
33 return NO_INIT;
34 }
35
36 // now set data source
37 setDataSource_post(p, p->setDataSource(fd, offset, length));
38 return mStatus;
39 }
上方这段代码感化就是获取一个PlayerType,然后一这个PlayerType作为参数传递到setDataSource _pre(),在这个办法中按照playerType做一些工作。关于PlayerType是如何获取的,以及我们传入的文件test.mp3,对应什么类型的playerType,我就不再介绍了。之所以不介绍是因为牵扯到资料的一些特有的参数和属性断定的,我不太懂得。不过我们可以从代码中找到在Android中一共又几种如许的Type,如下:
/frameworks/av/include/media/MediaPlayerInterface.h
1 enum player_type {
2 PV_PLAYER = 1,
3 SONIVOX_PLAYER = 2,
4 STAGEFRIGHT_PLAYER = 3,
5 NU_PLAYER = 4,
6 // Test players are available only in the ""test"" and ""eng"" builds.
7 // The shared library with the test player is passed passed as an
8 // argument to the ""test:"" url in the setDataSource call.
9 TEST_PLAYER = 5,
10 };
一种有五种type,然则PV_PLAYER如同是从Android2.3之后就作废了,因为PVPlayer是和Opencore相干的,自从OpenCore从Android2.3中移出之后,这个type对应的Player就没有了。所以在Android4.2中只剩下
四种type了。这些type和他们对应的player都邑以键值对的情势放入一个数据布局中,以备后用。在后面我们在应用这些type的时辰,就会发明这些键值对是存放在sFactoryMap中的,而sFactoryMap 是tFactoryMap型的,
也就是就是KeyVector。貌似KeyVector是android定义的数据布局吧,类似Hashmap。而把这些type和player键值对添参加sFactorymap中过程如下:
[/frameworks/av/media/libmediaplayerservice/MediaPlayerFactory.cpp]
1 void MediaPlayerFactory::registerBuiltinFactories() {
2 Mutex::Autolock lock_(&sLock);
3
4 if (sInitComplete)
5 return;
6
7 registerFactory_l(new StagefrightPlayerFactory(), STAGEFRIGHT_PLAYER);
8 registerFactory_l(new NuPlayerFactory(), NU_PLAYER);
9 registerFactory_l(new SonivoxPlayerFactory(), SONIVOX_PLAYER);
10 registerFactory_l(new TestPlayerFactory(), TEST_PLAYER);
11
12 sInitComplete = true;
13 }
看看上方这两段代码还是有益处的,最起码让我们熟悉打听工作的来龙去脉。若是我们想要在Android中扩大,应用新的解码对象,可以直接持续MediaPlayerFactory类,然后在这里注册所须要的type就行了。
接着上方的说,假设我们mp3对应的时defaultType,也就是STAGEFRIGHT_PLAYER,那么这个数据就会作为setDataSource_pre()的参数应用。我们就看看这个函数是如何哄骗Type参数创建新的对象的,代码如下:
【/frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp】
1 sp<MediaPlayerBase> MediaPlayerService::Client::setDataSource_pre(
2 player_type playerType)
3 {
4 ALOGV("player type = %d", playerType);
5
6 // create the right type of player
7 sp<MediaPlayerBase> p = createPlayer(playerType);
8 if (p == NULL) {
9 return p;
10 }
11
12 if (!p->hardwareOutput()) {
13 mAudioOutput = new AudioOutput(mAudioSessionId);
14 static_cast<MediaPlayerInterface*>(p.get())->setAudioSink(mAudioOutput);
15 }
16
17 return p;
18 }
先申明一下,上方的红色代码都是很首要的,不过13和14行代码在setDataSource()临时还是用不上,要比及start()的时辰,和AudioFlinger建树接洽的时辰才干用到。如今先标识表记标帜下,有个印象。我们持续我们的工作,解析createPlayer()办法,代码如下:
/frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp
1 sp<MediaPlayerBase> MediaPlayerService::Client::createPlayer(player_type playerType)
2 {
3 // determine if we have the right player type
4 sp<MediaPlayerBase> p = mPlayer;
5 if ((p != NULL) && (p->playerType() != playerType)) {
6 ALOGV(" player");
7 p.clear();
8 }
9 if (p == NULL) {
10 p = MediaPlayerFactory::createPlayer(playerType, this, notify);
11 }
12
13 if (p != NULL) {
14 p->setUID(mUID);
15 }
16
17 return p;
18 }
看红色代码标识的第10行代码,这里应用的设计模式中的--Factory模式,大师可以参考进修下。Android体系时一个如此重大错杂的体系,为了杰出的扩大性应用了很多设计模式。无论是进修策画机哪方面内容,Android源码都是一份不错的教材。还是持续看我们的任务,接着看代码,在MediaPlayerFactory::createPlayer()的代码如下:
[/frameworks/av/media/libmediaplayerservice/MediaPlayerFactory.cpp]
1 sp<MediaPlayerBase> MediaPlayerFactory::createPlayer(
2 player_type playerType,
3 void* cookie,
4 notify_callback_f notifyFunc) {
5 sp<MediaPlayerBase> p;
6 IFactory* factory;
7 status_t init_result;
8 Mutex::Autolock lock_(&sLock);
9
10 if (sFactoryMap.indexOfKey(playerType) < 0) {
11 ALOGE("Failed to create player object of type %d, no registered"
12 " factory", playerType);
13 return p;
14 }
15
16 factory = sFactoryMap.valueFor(playerType);
17 CHECK(NULL != factory);
18 p = factory->createPlayer();
19
20 if (p == NULL) {
21 ALOGE("Failed to create player object of type %d, create failed",
22 playerType);
23 return p;
24 }
25
26 init_result = p->initCheck();
27 if (init_result == NO_ERROR) {
28 p->setNotifyCallback(cookie, notifyFunc);
29 } else {
30 ALOGE("Failed to create player object of type %d, initCheck failed"
31 " (res = %d)", playerType, init_result);
32 p.clear();
33 }
34
35 return p;
36 }
我们经由过程playerType获取到得是IFactory对象,然后应用这个对象进行createPlayer。对于这一点的懂得和Factory模式有关,若是不熟悉打听的话,抽出一点点时候看看Factory模式必然是会有益处的。按照
playerType,我们这里的factory应当是StageFrightPlayerFactory的对象,然后调用它的的createPlayer函数。createPlayer()的实现如下:
[/frameworks/av/media/libmediaplayerservice/MediaPlayerFactory.cpp]
1 virtual sp<MediaPlayerBase> createPlayer() {
2 ALOGV(" create StagefrightPlayer");
3 return new StagefrightPlayer();
4 }
从这里StageFrightPlayerFactory的createPlayer中我们获得的返回值是一个新的对象--StagefrightPlayer的实例。到这里,我们就算是进入stagefright了,我们代码查询拜访到这儿也根蒂根基将近停止了。
我们还是接着看看这个新的对象的机关办法,代码如下:
[/frameworks/av/media/libmediaplayerservice/StagefrightPlayer.cpp]
1 StagefrightPlayer::StagefrightPlayer()
2 : mPlayer(new AwesomePlayer) {
3 ALOGV("StagefrightPlayer");
4
5 mPlayer->setListener(this);
6 }
在这个机关函数中,有个很轻易忽视的处所,就是机关函数的属性赋值的体式格式,在C++中经常应用,叫做机关函数的初始化列表。乍一看,这个机关函数什么也没有,其实重点内容就在机关函数的初始化列表中,在这里我们又
创建了一个对象AwesomPlayer. 我感觉我们进行到这儿,也可以告一段落了,因为我们已经接触到了Stagefright,再向下查询拜访代码的话就应当是零丁研究stagefright的工作了。今后,在查询拜访过stagefrigh后,我会
再补上一篇博客说说stagefright的。
好了,骚年们,让我们像函数调用停止时一样,一步一步return吧,直到可以或许持续向下履行的处所。我们一向查询拜访到这儿,是在解析MediaPlayerService中要做的两件事中的第二件工作:setDataSource()。我们方才停止
解析的处所是:sp<MediaPlayerBase> p = setDataSource_pre(playerType);让我们先总结一下我们方才都做了哪些工作,其实挺简单的就是创建了一个对象StagefrightPlayer,并把这个对象保存到p中。StagefrightPlayer
对象在创建的过程中,我们也设置了一些参数,比如p->setAudioSink(),还创建了AwesomePlayer对象,保存在StagefrightPlayer对象中的mPlayer属性中。既然setDataSource_pre()已经查询拜访停止,那么我们就要开端下一步
操纵了,就是下面已经用红色代码标示出来的项目组:
[/frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp]
1 status_t MediaPlayerService::Client::setDataSource(int fd, int64_t offset, int64_t length)
2 {
3 ...
4 player_type playerType = MediaPlayerFactory::getPlayerType(this,
5 fd,
6 offset,
7 length);
8 sp<MediaPlayerBase> p = setDataSource_pre(playerType);
9 ...
10 // now set data source
11 setDataSource_post(p, p->setDataSource(fd, offset, length));
12 return mStatus;
13 }
其实函数setDataSource_post()对于我们解析MediaPlayer流程的感化不太大,可能是要设置MediaPlayer的一些功能
的属,我们此次重点存眷的是这个办法中的第二个参数。由上方的解析,我们知道p指向的对象是一个StagefrightPlayer
对象,所以p->setDataSource()的实现,我们就应当去StagefrightPlayer类中寻找。找到响应的代码如下:
[/frameworks/av/media/libmediaplayerservice/StagefrightPlayer.cpp]
1 status_t StagefrightPlayer::setDataSource(int fd, int64_t offset, int64_t length) {
2 ALOGV("setDataSource(%d, %lld, %lld)", fd, offset, length);
3 return mPlayer->setDataSource(dup(fd), offset, length);
4 }
本来是应用StagefrightPlayer的属性mPlayer去实现的,前面我们也说到了stagefrightPlayer.mPlayer指向的是
AwesomePlayer.到这儿我们到此为止吧。
到这里,在MediaPlayerService端的的setDataSource()的工作也停止了,总结一下就是:从Java应用法度中的
MediaPlayer的setDataSource()会传递到Native层中的MediaPlayer的setDataSource()去履行,而
MediaPlayer会把这个办法交给MediaPlayerservice去履行。MediaPlayerService的是应用stagefrightPlayer
实现的,最后,最后,我们的setdataSource还是交给了AwesomePlayer去履行了。这个流程把MediaPlayer和
MediaPlayerService之间的接洽建树起来,同时又把MediaPlayerService是如何应用stagefright实现的关系建树了
起来。到这里,我们的解析MediaPlayer的流程的目标也算是根蒂根基达到了。若是再持续解析MediaPlayer的话,我们可
以分为两个标的目标了,一个是解析stagefright是如何解码的;一个是解析MediaPlayer与AudioTrack,AudioFlinger
之间的关系了。由此看来,将来的任务和路子还是任重道远的。