MediaPlayer代码分析(3)-播放前的准备prepare

本文分析prepare和prepareAsyn。

JNI层的prepare*

static voidandroid_media_MediaPlayer_prepare(JNIEnv *env, jobject thiz){    sp<MediaPlayer> mp = getMediaPlayer(env, thiz);    if (mp == NULL ) {        jniThrowException(env, "java/lang/IllegalStateException", NULL);        return;    }    // Handle the case where the display surface was set before the mp was    // initialized. We try again to make it stick.    sp<IGraphicBufferProducer> st = getVideoSurfaceTexture(env, thiz);    mp->setVideoSurfaceTexture(st);    process_media_player_call( env, thiz, mp->prepare(), "java/io/IOException", "Prepare failed." );}static voidandroid_media_MediaPlayer_prepareAsync(JNIEnv *env, jobject thiz){    sp<MediaPlayer> mp = getMediaPlayer(env, thiz);    if (mp == NULL ) {        jniThrowException(env, "java/lang/IllegalStateException", NULL);        return;    }    // Handle the case where the display surface was set before the mp was    // initialized. We try again to make it stick.    sp<IGraphicBufferProducer> st = getVideoSurfaceTexture(env, thiz);    mp->setVideoSurfaceTexture(st);    process_media_player_call( env, thiz, mp->prepareAsync(), "java/io/IOException", "Prepare Async failed." );}

两个函数都调用了getVideoSurfaceTexture()

static sp<IGraphicBufferProducer>getVideoSurfaceTexture(JNIEnv* env, jobject thiz) {    IGraphicBufferProducer * const p = (IGraphicBufferProducer*)env->GetIntField(thiz, fields.surface_texture);    return sp<IGraphicBufferProducer>(p);}

获取了Java层的mNativeSurfaceTexture

private int mNativeSurfaceTexture;  // accessed by native methods

接着调用了mp->prepare和mp->prepareAsync。这里我们只分析mp->prepareAsync，它的操作结果经过回调通知给Java层的。

status_t MediaPlayer::prepareAsync(){    ALOGV("prepareAsync");    Mutex::Autolock _l(mLock);    return prepareAsync_l();}

它里面有一个互斥锁，之后调用prepareAsync_l()；

// must call with lock heldstatus_t MediaPlayer::prepareAsync_l(){    if ( (mPlayer != 0) && ( mCurrentState & ( MEDIA_PLAYER_INITIALIZED | MEDIA_PLAYER_STOPPED) ) ) {        mPlayer->setAudioStreamType(mStreamType);        mCurrentState = MEDIA_PLAYER_PREPARING;        return mPlayer->prepareAsync();    }    ALOGE("prepareAsync called in state %d", mCurrentState);    return INVALID_OPERATION;}

首先是mp->setAudioStreamType，在IMediaPlayer.cpp中对应的transact操作是SET_AUDIO_STREAM_TYPE

case SET_AUDIO_STREAM_TYPE: {            CHECK_INTERFACE(IMediaPlayer, data, reply);            reply->writeInt32(setAudioStreamType((audio_stream_type_t) data.readInt32()));            return NO_ERROR;        } break;

对应到BnMediaPlayer侧则是MediaPlayerService::Client::setAudioStreamType

status_t MediaPlayerService::Client::setAudioStreamType(audio_stream_type_t type){    ALOGV("[%d] setAudioStreamType(%d)", mConnId, type);    // TODO: for hardware output, call player instead    Mutex::Autolock l(mLock);    if (mAudioOutput != 0) mAudioOutput->setAudioStreamType(type);    return NO_ERROR;}

AudioOutput->setAudioStreamType

void            setAudioStreamType(audio_stream_type_t streamType) { mStreamType = streamType; }

这里讲一下AudiOutput，它是class AudioOutput : public MediaPlayerBase::AudioSink，这个AudioSink将在后面AwesomePlayer里用到，也就是调用的AudioOutput。

在setDataSource_pre中我们已经将mAudioOutput设置到AwesomePlayer中了。

sp<MediaPlayerBase> MediaPlayerService::Client::setDataSource_pre(        player_type playerType){    ALOGV("player type = %d", playerType);    // create the right type of player    sp<MediaPlayerBase> p = createPlayer(playerType);    if (p == NULL) {        return p;    }    if (!p->hardwareOutput()) {        mAudioOutput = new AudioOutput(mAudioSessionId, IPCThreadState::self()->getCallingUid());        static_cast<MediaPlayerInterface*>(p.get())->setAudioSink(mAudioOutput);    }    return p;}

接着我们继续看prepareAsync函数，mp->prepareAsync()对应的到BnMediaPlayer的操作是MediaPlayerService::Client::prepareAsync

        case PREPARE_ASYNC: {            CHECK_INTERFACE(IMediaPlayer, data, reply);            reply->writeInt32(prepareAsync());            return NO_ERROR;        } break;

MediaPlayerService::Client::prepareAsync

status_t MediaPlayerService::Client::prepareAsync(){    ALOGV("[%d] prepareAsync", mConnId);    sp<MediaPlayerBase> p = getPlayer();    if (p == 0) return UNKNOWN_ERROR;    status_t ret = p->prepareAsync();#if CALLBACK_ANTAGONIZER    ALOGD("start Antagonizer");    if (ret == NO_ERROR) mAntagonizer->start();#endif    return ret;}

这里调用了AwesomePlayer的prepareAsync函数

status_t AwesomePlayer::prepareAsync() {    ATRACE_CALL();    Mutex::Autolock autoLock(mLock);    if (mFlags & PREPARING) {        return UNKNOWN_ERROR;  // async prepare already pending    }    mIsAsyncPrepare = true;    return prepareAsync_l();}

先判断mFlags是不是preparing。初始化AwesomePlayer时将mFlags赋值为0，所以这个判断不会是true，继续看下面代码

AwesomePlayer::prepareAsync_l()

status_t AwesomePlayer::prepareAsync_l() {    if (mFlags & PREPARING) {        return UNKNOWN_ERROR;  // async prepare already pending    }    if (!mQueueStarted) {        mQueue.start();        mQueueStarted = true;    }    modifyFlags(PREPARING, SET);    mAsyncPrepareEvent = new AwesomeEvent(            this, &AwesomePlayer::onPrepareAsyncEvent);    mQueue.postEvent(mAsyncPrepareEvent);    return OK;}

首先还是判断mFlags，此时不是preparing。接着启动mQueue(类TimedEventQueue)。

之后修改mFlags的状态为PREPARING，表示现在正在准备处理文件的音视频流。

这里出现了一个AwesomeEvent，这是要闹什么？再看后面的代码，原来是定义了一个event对象，然后放到之前启动的mQueue中进行处理。
参见之前的TimedEventQueue分析文章，queue中处理的结果就是调用AwesomePlayer::onPrepareAsyncEvent函数。

继续看AwesomePlayer::onPrepareAsyncEvent

void AwesomePlayer::onPrepareAsyncEvent() {    Mutex::Autolock autoLock(mLock);    beginPrepareAsync_l();}

AwesomePlayer::beginPrepareAsync_l()

void AwesomePlayer::beginPrepareAsync_l() {    if (mFlags & PREPARE_CANCELLED) {        ALOGI("prepare was cancelled before doing anything");        abortPrepare(UNKNOWN_ERROR);        return;    }    if (mUri.size() > 0) {        status_t err = finishSetDataSource_l();        if (err != OK) {            abortPrepare(err);            return;        }    }    if (mVideoTrack != NULL && mVideoSource == NULL) {        status_t err = initVideoDecoder();        if (err != OK) {            abortPrepare(err);            return;        }    }    if (mAudioTrack != NULL && mAudioSource == NULL) {        status_t err = initAudioDecoder();        if (err != OK) {            abortPrepare(err);            return;        }    }    modifyFlags(PREPARING_CONNECTED, SET);    if (isStreamingHTTP()) {        postBufferingEvent_l();    } else {        finishAsyncPrepare_l();    }}

这里我们分析文件类型的，uri的流媒体以后再说。代码分别初始化audio和video。

初始化video解码器：

AwesomePlayer::initVideoDecoder

status_t AwesomePlayer::initVideoDecoder(uint32_t flags) {    ATRACE_CALL();    // Either the application or the DRM system can independently say    // that there must be a hardware-protected path to an external video sink.    // For now we always require a hardware-protected path to external video sink    // if content is DRMed, but eventually this could be optional per DRM agent.    // When the application wants protection, then    //   (USE_SURFACE_ALLOC && (mSurface != 0) &&    //   (mSurface->getFlags() & ISurfaceComposer::eProtectedByApp))    // will be true, but that part is already handled by SurfaceFlinger.#ifdef DEBUG_HDCP    // For debugging, we allow a system property to control the protected usage.    // In case of uninitialized or unexpected property, we default to "DRM only".    bool setProtectionBit = false;    char value[PROPERTY_VALUE_MAX];    if (property_get("persist.sys.hdcp_checking", value, NULL)) {        if (!strcmp(value, "never")) {            // nop        } else if (!strcmp(value, "always")) {            setProtectionBit = true;        } else if (!strcmp(value, "drm-only")) {            if (mDecryptHandle != NULL) {                setProtectionBit = true;            }        // property value is empty, or unexpected value        } else {            if (mDecryptHandle != NULL) {                setProtectionBit = true;            }        }    // can' read property value    } else {        if (mDecryptHandle != NULL) {            setProtectionBit = true;        }    }    // note that usage bit is already cleared, so no need to clear it in the "else" case    if (setProtectionBit) {        flags |= OMXCodec::kEnableGrallocUsageProtected;    }#else    if (mDecryptHandle != NULL) {        flags |= OMXCodec::kEnableGrallocUsageProtected;    }#endif    ALOGV("initVideoDecoder flags=0x%x", flags);    mVideoSource = OMXCodec::Create(            mClient.interface(), mVideoTrack->getFormat(),            false, // createEncoder            mVideoTrack,            NULL, flags, USE_SURFACE_ALLOC ? mNativeWindow : NULL);    if (mVideoSource != NULL) {        int64_t durationUs;        if (mVideoTrack->getFormat()->findInt64(kKeyDuration, &durationUs)) {            Mutex::Autolock autoLock(mMiscStateLock);            if (mDurationUs < 0 || durationUs > mDurationUs) {                mDurationUs = durationUs;            }        }        status_t err = mVideoSource->start();        if (err != OK) {            ALOGE("failed to start video source");            mVideoSource.clear();            return err;        }    }    if (mVideoSource != NULL) {        const char *componentName;        CHECK(mVideoSource->getFormat()                ->findCString(kKeyDecoderComponent, &componentName));        {            Mutex::Autolock autoLock(mStatsLock);            TrackStat *stat = &mStats.mTracks.editItemAt(mStats.mVideoTrackIndex);            stat->mDecoderName = componentName;        }        static const char *kPrefix = "OMX.Nvidia.";        static const char *kSuffix = ".decode";        static const size_t kSuffixLength = strlen(kSuffix);        size_t componentNameLength = strlen(componentName);        if (!strncmp(componentName, kPrefix, strlen(kPrefix))                && componentNameLength >= kSuffixLength                && !strcmp(&componentName[                    componentNameLength - kSuffixLength], kSuffix)) {            modifyFlags(SLOW_DECODER_HACK, SET);        }    }    return mVideoSource != NULL ? OK : UNKNOWN_ERROR;}

initVideoDecoder的核心就是OMXCodec::Create。

initAudioDecoder()

status_t AwesomePlayer::initAudioDecoder() {    ATRACE_CALL();    sp<MetaData> meta = mAudioTrack->getFormat();    const char *mime;    CHECK(meta->findCString(kKeyMIMEType, &mime));    // Check whether there is a hardware codec for this stream    // This doesn't guarantee that the hardware has a free stream    // but it avoids us attempting to open (and re-open) an offload    // stream to hardware that doesn't have the necessary codec    audio_stream_type_t streamType = AUDIO_STREAM_MUSIC;    if (mAudioSink != NULL) {        streamType = mAudioSink->getAudioStreamType();    }    mOffloadAudio = canOffloadStream(meta, (mVideoSource != NULL),                                     isStreamingHTTP(), streamType);    if (!strcasecmp(mime, MEDIA_MIMETYPE_AUDIO_RAW)) {        ALOGV("createAudioPlayer: bypass OMX (raw)");        mAudioSource = mAudioTrack;    } else {        // If offloading we still create a OMX decoder as a fall-back        // but we don't start it        mOmxSource = OMXCodec::Create(                mClient.interface(), mAudioTrack->getFormat(),                false, // createEncoder                mAudioTrack);        if (mOffloadAudio) {            ALOGV("createAudioPlayer: bypass OMX (offload)");            mAudioSource = mAudioTrack;        } else {            mAudioSource = mOmxSource;        }    }    if (mAudioSource != NULL) {        int64_t durationUs;        if (mAudioTrack->getFormat()->findInt64(kKeyDuration, &durationUs)) {            Mutex::Autolock autoLock(mMiscStateLock);            if (mDurationUs < 0 || durationUs > mDurationUs) {                mDurationUs = durationUs;            }        }        status_t err = mAudioSource->start();        if (err != OK) {            mAudioSource.clear();            mOmxSource.clear();            return err;        }    } else if (!strcasecmp(mime, MEDIA_MIMETYPE_AUDIO_QCELP)) {        // For legacy reasons we're simply going to ignore the absence        // of an audio decoder for QCELP instead of aborting playback        // altogether.        return OK;    }    if (mAudioSource != NULL) {        Mutex::Autolock autoLock(mStatsLock);        TrackStat *stat = &mStats.mTracks.editItemAt(mStats.mAudioTrackIndex);        const char *component;        if (!mAudioSource->getFormat()                ->findCString(kKeyDecoderComponent, &component)) {            component = "none";        }        stat->mDecoderName = component;    }    return mAudioSource != NULL ? OK : UNKNOWN_ERROR;}

initAudioDecoder的核心函数同样是OMXCodec::Create.

因为AwesomePlayer底层使用的是OMX的编解码器，所以在获取音视频信息是就需要使用OMX模块。

OMX部分下一篇将解释一下。

回到beginPrepareAsync_l。

void AwesomePlayer::beginPrepareAsync_l() {    if (mFlags & PREPARE_CANCELLED) {        ALOGI("prepare was cancelled before doing anything");        abortPrepare(UNKNOWN_ERROR);        return;    }    if (mUri.size() > 0) {        status_t err = finishSetDataSource_l();        if (err != OK) {            abortPrepare(err);            return;        }    }    if (mVideoTrack != NULL && mVideoSource == NULL) {        status_t err = initVideoDecoder();        if (err != OK) {            abortPrepare(err);            return;        }    }    if (mAudioTrack != NULL && mAudioSource == NULL) {        status_t err = initAudioDecoder();        if (err != OK) {            abortPrepare(err);            return;        }    }    modifyFlags(PREPARING_CONNECTED, SET);    if (isStreamingHTTP()) {        postBufferingEvent_l();    } else {        finishAsyncPrepare_l();    }}

因为我们是文件，所以继续走finishAsyncPrepare_l

finishAsyncPrepare_l

void AwesomePlayer::finishAsyncPrepare_l() {    if (mIsAsyncPrepare) {        if (mVideoSource == NULL) {            notifyListener_l(MEDIA_SET_VIDEO_SIZE, 0, 0);        } else {            notifyVideoSize_l();        }        notifyListener_l(MEDIA_PREPARED);    }    mPrepareResult = OK;    modifyFlags((PREPARING|PREPARE_CANCELLED|PREPARING_CONNECTED), CLEAR);    modifyFlags(PREPARED, SET);    mAsyncPrepareEvent = NULL;    mPreparedCondition.broadcast();    if (mAudioTearDown) {        if (mPrepareResult == OK) {            if (mExtractorFlags & MediaExtractor::CAN_SEEK) {                seekTo_l(mAudioTearDownPosition);            }            if (mAudioTearDownWasPlaying) {                modifyFlags(CACHE_UNDERRUN, CLEAR);                play_l();            }        }        mAudioTearDown = false;    }}

首先通知上层，发送MEDIA_SET_VIDEO_SIZE消息。

MediaPlayer.cpp

    case MEDIA_SET_VIDEO_SIZE:        ALOGV("New video size %d x %d", ext1, ext2);        mVideoWidth = ext1;        mVideoHeight = ext2;

MediaPlayer.java

    case MEDIA_SET_VIDEO_SIZE:        if (mOnVideoSizeChangedListener != null)             mOnVideoSizeChangedListener.onVideoSizeChanged(mMediaPlayer, msg.arg1, msg.arg2);

之后是通知MEDIA_PREPARED

MediaPlayer.cpp

    case MEDIA_PREPARED:        ALOGV("prepared");        mCurrentState = MEDIA_PLAYER_PREPARED;        if (mPrepareSync) {            ALOGV("signal application thread");            mPrepareSync = false;            mPrepareStatus = NO_ERROR;            mSignal.signal();        }        break;

MediaPlayer.java

    case MEDIA_PREPARED:        scanInternalSubtitleTracks();        if (mOnPreparedListener != null)            mOnPreparedListener.onPrepared(mMediaPlayer);        return;

接着清楚AwesomePlayer的状态，并设置为PREPARED。

然后通过mPreparedCondition.broadcast();，发送广播。通知等待这个条件的地方。

status_t AwesomePlayer::prepare_l() {    if (mFlags & PREPARED) {        return OK;    }    if (mFlags & PREPARING) {        return UNKNOWN_ERROR;    }    mIsAsyncPrepare = false;    status_t err = prepareAsync_l();    if (err != OK) {        return err;    }    while (mFlags & PREPARING) {        mPreparedCondition.wait(mLock);    }    return mPrepareResult;}

原来在prepare_l中函数一直等待这个条件的执行。prepareAsync_l中启动的mQueue消息队列使得执行进入另一个pthread线程，如果希望两个线程同步，那么可以使用condition方式进行通知。

prepare_l函数最后返回mPrepareResult，mPrepareResult在finishAsyncPrepare_l中mPrepareResult = OK;。

至此prepare的流程已经分析完成。下一篇将分析OMX模块。