nvidia---7.stagefright框架分析

0.1继上一篇 android应用开发---11.MediaPlayer实现音乐播放的demo 发现android播放音频文件只需要下面几步

mediaPlayer = new MediaPlayer();               //1
mediaPlayer.setDataSource(File_Path);          //2
mediaPlayer.setAudioStreamType(AudioManager.STREAM_MUSIC);
mediaPlayer.prepareAsync();       //异步的方式加载音乐文件
mediaPlayer.start();             //准备好之后才能start

0.2 stageFright的框架如下所示:


下面就逐个分析一下这个过程

---

一.new MediaPlayer的过程

./frameworks/base/media/java/android/media/MediaPlayer.java
public MediaPlayer() {
   //调用jni中的native_setup
   native_setup(new WeakReference<MediaPlayer>(this));
}

1.1 jni的native_setup

./frameworks/base/media/jni/android_media_MediaPlayer.cpp
android_media_MediaPlayer_native_setup(JNIEnv *env, jobject thiz, jobject weak_this)
{
    sp<MediaPlayer> mp = new MediaPlayer();
    sp<JNIMediaPlayerListener> listener = new JNIMediaPlayerListener(env, thiz, weak_this);
    mp->setListener(listener);
    setMediaPlayer(env, thiz, mp);    //把mp保存起来
}

---

二. setDataSource的过程

在./frameworks/base/media/java/android/media/MediaPlayer.java中
public void setDataSource(String path) {
    //打开path的文件，转为fd
    //调用jni的_setDataSource
}

2.1 jni中的setDataSource

./frameworks/base/media/jni/android_media_MediaPlayer.cpp
android_media_MediaPlayer_setDataSourceFD(JNIEnv *env, jobject thiz, jobject fileDescriptor, jlong offset, jlong length)
{
    sp<MediaPlayer> mp = getMediaPlayer(env, thiz);             //获取第1步保存的mp
    int fd = jniGetFDFromFileDescriptor(env, fileDescriptor);
    process_media_player_call( env, thiz, mp->setDataSource(fd, offset, length), "java/io/IOException", "setDataSourceFD failed." );
}

2.2

./frameworks/av/media/libmedia/mediaplayer.cpp
status_t MediaPlayer::setDataSource(int fd, int64_t offset, int64_t length)
{
    const sp<IMediaPlayerService>& service(getMediaPlayerService());    //获取MediaPlayerService
    sp<IMediaPlayer> player(service->create(this, mAudioSessionId));    //2.2.1通过MediaPlayerService创建player
    player->setDataSource(fd, offset, length)));                        //2.2.2
    attachNewPlayer(player);
    return err;
}

通过下面的2.3可知service->create实际上是创建了一个::Client>
引自《Android多媒体开发【4】-- 上层mediaPlayer 是如何调到Stagefright的【2】【图解】》
“MediaPlayerService会为每个client应用进程创建一个相应的MediaPlayerService::Client的实例，来提供服务”
2.2.1

./frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp
sp<IMediaPlayer> MediaPlayerService::create(const sp<IMediaPlayerClient>& client, int audioSessionId)
{
    pid_t pid = IPCThreadState::self()->getCallingPid();
    int32_t connId = android_atomic_inc(&mNextConnId);

    sp<Client> c = new Client(this, pid, connId, client, audioSessionId, IPCThreadState::self()->getCallingUid());

    wp<Client> w = c;
    {
        Mutex::Autolock lock(mLock);
        mClients.add(w);               //添加到成员数组中
    }
    return c;
}

2.2.2
知道了player是一个::Client>,那么player->setDataSource(fd, offset, length)));也就好理解了
setDataSource主要是调用createPlayer，其过程如下
MediaPlayerService::Client::setDataSource
  --> MediaPlayerService::Client::setDataSource_pre
    --> MediaPlayerService::Client::createPlayer
      --> MediaPlayerFactory::createPlayer
        --> MediaPlayerFactory::StagefrightPlayerFactory::createPlayer

./frameworks/av/media/libmediaplayerservice/MediaPlayerService.cpp
status_t MediaPlayerService::Client::setDataSource(int fd, int64_t offset, int64_t length)
{
    int ret = fstat(fd, &sb);
    //MediaPlayerFactory中通过读取file中的内容来判断是下面的哪种类型
    //在STAGEFRIGHT_PLAYER, NU_PLAYER, SONIVOX_PLAYER, TEST_PLAYER
    //默认是STAGEFRIGHT_PLAYER
    player_type playerType = MediaPlayerFactory::getPlayerType(this,fd, offset, length);
    sp<MediaPlayerBase> p = setDataSource_pre(playerType);       //创建player并设置output
    //这儿实际上是创建了一个StagefrightPlayer,所以下面的p->setDataSourc
    //就是调用StagefrightPlayer::setDataSource
    setDataSource_post(p, p->setDataSource(fd, offset, length));  //第2步--> 2.2.3
    return mStatus;
}

sp<MediaPlayerBase> MediaPlayerService::Client::setDataSource_pre(player_type playerType)
{
    sp<MediaPlayerBase> p = createPlayer(playerType);     //先创建player，这儿实际上是创建了一个StagefrightPlayer
    if (!p->hardwareOutput()) {
        mAudioOutput = new AudioOutput(mAudioSessionId, IPCThreadState::self()->getCallingUid());
        static_cast<MediaPlayerInterface*>(p.get())->setAudioSink(mAudioOutput);
    }
    return p;
}

sp<MediaPlayerBase> MediaPlayerService::Client::createPlayer(player_type playerType)
{
    sp<MediaPlayerBase> p = mPlayer;
    p = MediaPlayerFactory::createPlayer(playerType, this, notify);  //返回一个StagefrightPlayer
    return p;
}

在MediaPlayerFactory中

./frameworks/av/media/libmediaplayerservice/MediaPlayerFactory.cpp
sp<MediaPlayerBase> MediaPlayerFactory::createPlayer(player_type playerType, ...) {
    sp<MediaPlayerBase> p;
    factory = sFactoryMap.valueFor(playerType);
    p = factory->createPlayer();     //默认是创建StagefrightPlayer();
    return p;
}

virtual sp<MediaPlayerBase> createPlayer() {
    return new StagefrightPlayer();     //最终new了一个StagefrightPlayer
}

2.2.3
//这儿实际上是创建了一个StagefrightPlayer,所以下面的p->setDataSourc  
//就是调用StagefrightPlayer::setDataSource
    setDataSource_post(p, p->setDataSource(fd, offset, length));

./frameworks/av/media/libmediaplayerservice/StagefrightPlayer.cpp
StagefrightPlayer::setDataSource
AwesomePlayer::setDataSource(int fd, int64_t offset, int64_t length)
    --> setDataSource_l(const sp<DataSource> &dataSource)
        --> extractor = MediaExtractor::Create(dataSource); //这儿会调用sniff检测文件的类型，并根据文件类型创建相应的解码器
                                                            //如:WAV会new WAVExtractor, mp3会new MP3Extractor
        --> setDataSource_l(extractor);

三. prepare的过程

---

三. start的过程
1. 近似的认为：MediaPlayer.java是StagefrightPlayer的java层的调用，这样就很好理解了

mediaPlayer = new MediaPlayer(); --> new StagefrightPlayer
mediaPlayer.setDataSource(File_Path); --> StagefrightPlayer::setDataSource
mediaPlayer.prepareAsync(); --> StagefrightPlayer::prepareAsync
mediaPlayer.start(); --> StagefrightPlayer::start

但是StagefrightPlayer只是一个空壳，它是调用了AwesomePlayer，所以进一步:

mediaPlayer = new MediaPlayer --> new StagefrightPlayer        --> mPlayer(new AwesomePlayer)
mediaPlayer.setDataSource --> StagefrightPlayer::setDataSource --> mPlayer->setDataSource
mediaPlayer.prepareAsync  --> StagefrightPlayer::prepareAsync  --> mPlayer->prepareAsync
mediaPlayer.start();      --> StagefrightPlayer::start         --> mPlayer->play()

---

四. 数据的读取

frameworks/av/media/libstagefright/DataSource.cpp
status_t AwesomePlayer::setDataSource(int fd, int64_t offset, int64_t length) {
    sp<DataSource> dataSource;
    sp<DataSource> fileSource = new FileSource(fd, offset, length);     //new了一个FileSource,里面的内容是open
    dataSource = new NuCachedSource2(fileSource);

    MediaExtractor::Create(dataSource);
      --> source->sniff(&tmp, &confidence, &meta)                       //调用了DataSource的sniff过程
        --> mSniffer->sniff(this, mimeType, confidence, meta);          //this 就是刚刚new的fileSource
          --> (*it)(source, &newMimeType, &newConfidence, &newMeta)     //调用Extractor的sniffer的过程
            --> source->readAt(0, header, sizeof(header)                //数据读取的过程
}

下面就看一下数据读取的真正过程

./frameworks/av/media/libstagefright/FileSource.cpp
FileSource有两个构造函数，如果传的是文件名，则调用open打开，获取fd; 如果是fd,则保存在成员变量中
FileSource::FileSource(const char *filename)
{
    mFd = open(filename, O_LARGEFILE | O_RDONLY);  //open:必然要调用系统接口的
    mLength = lseek64(mFd, 0, SEEK_END);
}

FileSource::FileSource(int fd, int64_t offset, int64_t length)
    : mFd(fd),                                 //传的是fd,则直接保存起来
      mOffset(offset),
      mLength(length)
}
//真正的读取过程，还是调用系统的read，没有什么神秘的
ssize_t FileSource::readAt(off64_t offset, void *data, size_t size) {
    lseek64(mFd, offset + mOffset, SEEK_SET);
    read(mFd, data, size);                              //read:隐藏的很深
}

---

参考文章:
Android多媒体开发【4】-- 上层mediaPlayer 是如何调到Stagefright的【2】【图解】
http://blog.csdn.net/tx3344/article/details/8024489