Android MediaRecorder系统结构

前面有分析过Camera的实现，现在来看看MediaRecorder的实现，这里我不会太去关注它的分层结构，我更关注它的逻辑！

APP层 /path/to/aosp/frameworks/base/media/java/android/media/MediaRecorder.java
JNI层 /path/to/aosp/frameworks/base/media/jni/android_media_MediaRecorder.cpp
调用NATIVE层的MediaRecorder(这里是BnMediaRecorderClient)
header /path/to/aosp/frameworks/av/include/media/mediarecorder.h
implementation /path/to/aosp/frameworks/av/media/libmedia/mediarecorder.cpp

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14

MediaRecorder::MediaRecorder() : mSurfaceMediaSource(NULL) {      ALOGV( "constructor" );        const sp<IMediaPlayerService>& service(getMediaPlayerService());      if (service != NULL) {          mMediaRecorder = service->createMediaRecorder(getpid());      }      if (mMediaRecorder != NULL) {          mCurrentState = MEDIA_RECORDER_IDLE;      }        doCleanUp(); }

getMediaPlayerService()这个方法位于/path/to/aosp/frameworks/av/include/media/IMediaDeathNotifier.h

获取到MediaPlayerService(这个是BpMediaPlayerService)之后
调用IMediaPlayerService当中的

?

1 2 3 4 5 6 7 8 9

sp<IMediaRecorder> MediaPlayerService::createMediaRecorder(pid_t pid) {      sp<MediaRecorderClient> recorder = new MediaRecorderClient( this , pid);      wp<MediaRecorderClient> w = recorder;      Mutex::Autolock lock(mLock);      mMediaRecorderClients.add(w);      ALOGV( "Create new media recorder client from pid %d" , pid);      return recorder; }

创建MediaRecorderClient(这里是BnMediaRecorder)

但是通过binder拿到的是BpMediaRecorder
因为有如下的interface_cast过程

?

1 2 3 4 5 6 7 8

virtual sp<IMediaRecorder> createMediaRecorder(pid_t pid) {      Parcel data, reply;      data.writeInterfaceToken(IMediaPlayerService::getInterfaceDescriptor());      data.writeInt32(pid);      remote()->transact(CREATE_MEDIA_RECORDER, data, &reply);      return interface_cast<IMediaRecorder>(reply.readStrongBinder()); }

而MediaRecorderClient当中又会创建StagefrightRecorder(MediaRecorderBase)，它位于
/path/to/aosp/frameworks/av/media/libmediaplayerservice/StagefrightRecorder.cpp

目前我们可以认为在APP/JNI/NATIVE这边是在一个进程当中，在MediaPlayerService当中的MediaRecorderClient/StagefrightRecorder是在另外一个进程当中，他们之间通过binder通信，而且Bp和Bn我们也都有拿到，后面我们将不再仔细区分Bp和Bn。

客户端这边
BnMediaRecorderClient
BpMediaRecorder
BpMediaPlayerService

服务端这边
BpMediaRecorderClient(如果需要通知客户端的话，它可以获得这个Bp)
BnMediaRecorder
BnMediaPlayerService

这有张图(点过去看原始大图)

我们以开始录影为例子，比如start()

在这里就兵分两路，一个CameraSource，一个MPEG4Writer(sp mWriter)
这两个class都位于/path/to/aosp/frameworks/av/media/libstagefright/当中

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

status_t StagefrightRecorder::startMPEG4Recording() {      int32_t totalBitRate;      status_t err = setupMPEG4Recording(              mOutputFd, mVideoWidth, mVideoHeight,              mVideoBitRate, &totalBitRate, &mWriter);      if (err != OK) {          return err;      }        int64_t startTimeUs = systemTime() / 1000 ;      sp<MetaData> meta = new MetaData;      setupMPEG4MetaData(startTimeUs, totalBitRate, &meta);        err = mWriter->start(meta.get());      if (err != OK) {          return err;      }        return OK; }

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

status_t StagefrightRecorder::setupMPEG4Recording(          int outputFd,          int32_t videoWidth, int32_t videoHeight,          int32_t videoBitRate,          int32_t *totalBitRate,          sp<MediaWriter> *mediaWriter) {      mediaWriter->clear();      *totalBitRate = 0 ;      status_t err = OK;      sp<MediaWriter> writer = new MPEG4Writer(outputFd);        if (mVideoSource < VIDEO_SOURCE_LIST_END) {            sp<MediaSource> mediaSource;          err = setupMediaSource(&mediaSource); // very important          if (err != OK) {              return err;          }            sp<MediaSource> encoder;          err = setupVideoEncoder(mediaSource, videoBitRate, &encoder); // very important          if (err != OK) {              return err;          }            writer->addSource(encoder);          *totalBitRate += videoBitRate;      }        // Audio source is added at the end if it exists.      // This help make sure that the "recoding" sound is suppressed for      // camcorder applications in the recorded files.      if (!mCaptureTimeLapse && (mAudioSource != AUDIO_SOURCE_CNT)) {          err = setupAudioEncoder(writer); // very important          if (err != OK) return err;          *totalBitRate += mAudioBitRate;      }        ...        writer->setListener(mListener);      *mediaWriter = writer;      return OK; }

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

// Set up the appropriate MediaSource depending on the chosen option status_t StagefrightRecorder::setupMediaSource(                        sp<MediaSource> *mediaSource) {      if (mVideoSource == VIDEO_SOURCE_DEFAULT              || mVideoSource == VIDEO_SOURCE_CAMERA) {          sp<CameraSource> cameraSource;          status_t err = setupCameraSource(&cameraSource);          if (err != OK) {              return err;          }          *mediaSource = cameraSource;      } else if (mVideoSource == VIDEO_SOURCE_GRALLOC_BUFFER) {          // If using GRAlloc buffers, setup surfacemediasource.          // Later a handle to that will be passed          // to the client side when queried          status_t err = setupSurfaceMediaSource();          if (err != OK) {              return err;          }          *mediaSource = mSurfaceMediaSource;      } else {          return INVALID_OPERATION;      }      return OK; }

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56

status_t StagefrightRecorder::setupCameraSource(          sp<CameraSource> *cameraSource) {      status_t err = OK;      if ((err = checkVideoEncoderCapabilities()) != OK) {          return err;      }      Size videoSize;      videoSize.width = mVideoWidth;      videoSize.height = mVideoHeight;      if (mCaptureTimeLapse) {          if (mTimeBetweenTimeLapseFrameCaptureUs < 0 ) {              ALOGE( "Invalid mTimeBetweenTimeLapseFrameCaptureUs value: %lld" ,                  mTimeBetweenTimeLapseFrameCaptureUs);              return BAD_VALUE;          }            mCameraSourceTimeLapse = CameraSourceTimeLapse::CreateFromCamera(                  mCamera, mCameraProxy, mCameraId,                  videoSize, mFrameRate, mPreviewSurface,                  mTimeBetweenTimeLapseFrameCaptureUs);          *cameraSource = mCameraSourceTimeLapse;      } else {          *cameraSource = CameraSource::CreateFromCamera(                  mCamera, mCameraProxy, mCameraId, videoSize, mFrameRate,                  mPreviewSurface, true /*storeMetaDataInVideoBuffers*/ );      }      mCamera.clear();      mCameraProxy.clear();      if (*cameraSource == NULL) {          return UNKNOWN_ERROR;      }        if ((*cameraSource)->initCheck() != OK) {          (*cameraSource).clear();          *cameraSource = NULL;          return NO_INIT;      }        // When frame rate is not set, the actual frame rate will be set to      // the current frame rate being used.      if (mFrameRate == - 1 ) {          int32_t frameRate = 0 ;          CHECK ((*cameraSource)->getFormat()->findInt32(                      kKeyFrameRate, &frameRate));          ALOGI( "Frame rate is not explicitly set. Use the current frame "               "rate (%d fps)" , frameRate);          mFrameRate = frameRate;      }        CHECK(mFrameRate != - 1 );        mIsMetaDataStoredInVideoBuffers =          (*cameraSource)->isMetaDataStoredInVideoBuffers();        return OK; }

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85

status_t StagefrightRecorder::setupVideoEncoder(          sp<MediaSource> cameraSource,          int32_t videoBitRate,          sp<MediaSource> *source) {      source->clear();        sp<MetaData> enc_meta = new MetaData;      enc_meta->setInt32(kKeyBitRate, videoBitRate);      enc_meta->setInt32(kKeyFrameRate, mFrameRate);        switch (mVideoEncoder) {          case VIDEO_ENCODER_H263:              enc_meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_VIDEO_H263);              break ;            case VIDEO_ENCODER_MPEG_4_SP:              enc_meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_VIDEO_MPEG4);              break ;            case VIDEO_ENCODER_H264:              enc_meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_VIDEO_AVC);              break ;            default :              CHECK(! "Should not be here, unsupported video encoding." );              break ;      }        sp<MetaData> meta = cameraSource->getFormat();        int32_t width, height, stride, sliceHeight, colorFormat;      CHECK(meta->findInt32(kKeyWidth, &width));      CHECK(meta->findInt32(kKeyHeight, &height));      CHECK(meta->findInt32(kKeyStride, &stride));      CHECK(meta->findInt32(kKeySliceHeight, &sliceHeight));      CHECK(meta->findInt32(kKeyColorFormat, &colorFormat));        enc_meta->setInt32(kKeyWidth, width);      enc_meta->setInt32(kKeyHeight, height);      enc_meta->setInt32(kKeyIFramesInterval, mIFramesIntervalSec);      enc_meta->setInt32(kKeyStride, stride);      enc_meta->setInt32(kKeySliceHeight, sliceHeight);      enc_meta->setInt32(kKeyColorFormat, colorFormat);      if (mVideoTimeScale > 0 ) {          enc_meta->setInt32(kKeyTimeScale, mVideoTimeScale);      }      if (mVideoEncoderProfile != - 1 ) {          enc_meta->setInt32(kKeyVideoProfile, mVideoEncoderProfile);      }      if (mVideoEncoderLevel != - 1 ) {          enc_meta->setInt32(kKeyVideoLevel, mVideoEncoderLevel);      }        OMXClient client;      CHECK_EQ(client.connect(), (status_t)OK);        uint32_t encoder_flags = 0 ;      if (mIsMetaDataStoredInVideoBuffers) {          encoder_flags |= OMXCodec::kStoreMetaDataInVideoBuffers;      }        // Do not wait for all the input buffers to become available.      // This give timelapse video recording faster response in      // receiving output from video encoder component.      if (mCaptureTimeLapse) {          encoder_flags |= OMXCodec::kOnlySubmitOneInputBufferAtOneTime;      }        sp<MediaSource> encoder = OMXCodec::Create(              client. interface (), enc_meta,              true /* createEncoder */ , cameraSource,              NULL, encoder_flags);      if (encoder == NULL) {          ALOGW( "Failed to create the encoder" );          // When the encoder fails to be created, we need          // release the camera source due to the camera's lock          // and unlock mechanism.          cameraSource->stop();          return UNKNOWN_ERROR;      }        *source = encoder;        return OK; }

这里和OMXCodec关联起来
有一个叫media_codecs.xml的配置文件来表明设备支持哪些codec

我们录制MPEG 4的时候还会有声音，所以后面还有个setupAudioEncoder，具体的方法就不展开了，总之就是把声音也作为一个Track加入到MPEG4Writer当中去。
这里插个题外话，Google说把setupAudioEncoder放到后面是为了避免开始录影的那一个提示声音也被录制进去，但是实际发现它这样做还是会有bug，在一些设备上还是会把那声录制进去，这个遇到的都是靠APP自己来播放声音来绕过这个问题的。

另外MPEG4Writer当中有个
start(MetaData*)
启动两个方法
a) startWriterThread

启动一个thread去写

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

void MPEG4Writer::threadFunc() {      ALOGV( "threadFunc" );        prctl(PR_SET_NAME, (unsigned long ) "MPEG4Writer" , 0 , 0 , 0 );        Mutex::Autolock autoLock(mLock);      while (!mDone) {          Chunk chunk;          bool chunkFound = false ;            while (!mDone && !(chunkFound = findChunkToWrite(&chunk))) {              mChunkReadyCondition.wait(mLock);          }            // Actual write without holding the lock in order to          // reduce the blocking time for media track threads.          if (chunkFound) {              mLock.unlock();              writeChunkToFile(&chunk);              mLock.lock();          }      }        writeAllChunks(); }

b) startTracks

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

status_t MPEG4Writer::startTracks(MetaData *params) {      for (List<Track *>::iterator it = mTracks.begin();           it != mTracks.end(); ++it) {          status_t err = (*it)->start(params);            if (err != OK) {              for (List<Track *>::iterator it2 = mTracks.begin();                   it2 != it; ++it2) {                  (*it2)->stop();              }                return err;          }      }      return OK; }

然后调用每个Track的start方法

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

status_t MPEG4Writer::Track::start(MetaData *params) {      ...        initTrackingProgressStatus(params);        ...        status_t err = mSource->start(meta.get()); // 这里会去执行CameraSource(start)，这两个是相互关联的        ...        pthread_create(&mThread, &attr, ThreadWrapper, this );      return OK; }   void *MPEG4Writer::Track::ThreadWrapper( void *me) {      Track *track = static_cast<Track *>(me);        status_t err = track->threadEntry();      return ( void *) err; }

通过status_t MPEG4Writer::Track::threadEntry()
是新启动另外一个thread，它里面会通过一个循环来不断读取CameraSource(read)里面的数据，CameraSource里面的数据当然是从driver返回过来的(可以参见CameraSourceListener，CameraSource用一个叫做mFrameReceived的List专门存放从driver过来的数据，如果收到数据会调用mFrameAvailableCondition.signal，若还没有开始录影，这个时候收到的数据是被丢弃的，当然MediaWriter先启动的是CameraSource的start方法，再启动写Track)，然后写到文件当中。
注意：准确来说这里MPEG4Writer读取的是OMXCodec里的数据，因为数据先到CameraSource，codec负责编码之后，MPEG4Writer才负责写到文件当中！关于数据在CameraSource/OMXCodec/MPEG4Writer之间是怎么传递的，可以参见http://guoh.org/lifelog/2013/06/interaction-between-stagefright-and-codec/当中讲Buffer的传输过程。

回头再来看，Stagefright做了什么事情？我更觉得它只是一个粘合剂(glue)的用处，它工作在MediaPlayerService这一层，把MediaSource，MediaWriter，Codec以及上层的MediaRecorder绑定在一起，这应该就是它最大的作用，Google用它来替换Opencore也是符合其一贯的工程派作风(相比复杂的学术派而言，虽然Google很多东西也很复杂，但是它一般都是以尽量简单的方式来解决问题)。
让大家觉得有点不习惯的是，它把MediaRecorder放在MediaPlayerService当中，这两个看起来是对立的事情，或者某一天它们会改名字，或者是两者分开，不知道～～

当然这只是个简单的大体介绍，Codec相关的后面争取专门来分析一下！

有些细节的东西在这里没有列出，需要的话会把一些注意点列出来：

1. 时光流逝录影
CameraSource对应的就是CameraSourceTimeLapse

具体做法就是在
dataCallbackTimestamp
当中有skipCurrentFrame

当然它是用些变量来记录和计算
mTimeBetweenTimeLapseVideoFramesUs(1E6/videoFrameRate) // 两个frame之间的间隔时间
记录上一个frame的(mLastTimeLapseFrameRealTimestampUs) // 上一个frame发生的时间
然后通过frame rate计算出两个frame之间的相距离时间，中间的都透过releaseOneRecordingFrame来drop掉
也就是说driver返回的东西都不变，只是在SW这层我们自己来处理掉

关于Time-lapse相关的可以参阅
https://en.wikipedia.org/wiki/Time-lapse_photography

2. 录影当中需要用到Camera的话是通过ICameraRecordingProxy，即Camera当中的RecordingProxy(这是一个BnCameraRecordingProxy)
当透过binder，将ICameraRecordingProxy传到服务端进程之后，它就变成了Bp，如下：

?

1 2 3 4 5 6 7 8 9

case SET_CAMERA: {      ALOGV( "SET_CAMERA" );      CHECK_INTERFACE(IMediaRecorder, data, reply);      sp<ICamera> camera = interface_cast<ICamera>(data.readStrongBinder());      sp<ICameraRecordingProxy> proxy =          interface_cast<ICameraRecordingProxy>(data.readStrongBinder());      reply->writeInt32(setCamera(camera, proxy));      return NO_ERROR; } break ;

在CameraSource当中会这样去使用

?

1 2 3 4 5 6 7 8

// We get the proxy from Camera, not ICamera. We need to get the proxy // to the remote Camera owned by the application. Here mCamera is a // local Camera object created by us. We cannot use the proxy from // mCamera here. mCamera = Camera::create(camera); if (mCamera == 0 ) return -EBUSY; mCameraRecordingProxy = proxy; mCameraFlags |= FLAGS_HOT_CAMERA;

疑问点：

CameraSource当中这个
List > mFramesBeingEncoded;
有什么用？
每编码完一个frame，CameraSource就会将其保存起来，Buffer被release的时候，会反过来release掉这些frame(s)，这种做法是为了效率么？为什么不编码完一个frame就将其release掉？
另外不得不再感叹下Google经常的delete this;行为，精妙，但是看起来反常！

原文地址：　http://guoh.org/lifelog/2013/06/android-mediarecorder-architecture/