Android录音下————AudioRecord源码分析

落日映苍穹つ 2022-03-17 09:14 1354阅读 0赞

# Android录音下————AudioRecord源码分析 #

### 文章目录 ###

*  Android录音下————AudioRecord源码分析
 *   *   *  一.概述
         *   *  1.主要分析点
             *  2.储备知识
         *  二.getMinBufferSize
         *   *  1.AR.getMinBufferSize(java)
             *  2.AR.getMinBufferSize(c++)
         *  三.new AudioRecord
         *   *  1.AR.new AudioRecord(java)
             *  2.AR.android\_media\_AudioRecord\_setup(c++)
             *  3.AR.set(c++)
             *  4.AS.get\_audio\_flinger(c++)
             *  5.IAF.openRecord(c++)
             *  6.AF.openRecord(c++)
             *  7.流程梳理
         *  四. startRecording
         *   *  1.AR.startRecording(java)
             *  2.AR.startRecording(c++)
             *  3.AR.start(c++)
             *  4.RT.start(c++)
             *  5.AS.startInput(c++)
             *  6.APS.startInput(c++)
             *  7.APMB.startInput(c++)
             *  8. APS.setParameters(c++)
             *  9.AS.setParameters(c++)
             *  10.AF.setParameters(c++)
             *  11.流程梳理
         *  五.read
         *   *  1.android\_media\_AudioRecord\_readInByteArray
             *  2.AR.read
         *  六.stop
         *  七.参考资料

### 一.概述 ###

在上一篇博客中，主要看了AudioRecord的应用。接下来让我们看一看AudioRecord源码的分析。

注：Native层源码版本：android5.1.0

#### 1.主要分析点 ####

分析的时候以上一篇博客AudioRecord使用过程中涉及到的方法来进行分析。主要分析下面几个方法：

*  getMinBufferSize:获取AudioRecord对象所需的最小缓冲区大小
 *  new AudioRecord：创建一个AudioRecord对象
 *  startRecording：启动AudioRecord进行录音
 *  read：读取录音数据
 *  stop：停止录音

#### 2.储备知识 ####

Android架构主要分为三层

*  底层 ：linux内核
 *  中间层：主要由c++层实现
 *  应用层：主要由java开发的应用程序

应用层和中间层的关联主要是由JNI来实现的，

所以大致过程就是，java应用产生的操作–>JNI调用中间层–>如果中间层需要硬件或者低层支持，则调用底层，否则处理完返回给应用层。

在音频通路建立的过程中，需要AndroidBinder相关的概念。  
详细内容可见我之前的博客：[Android之IPC4————Bander1 概述与Bander驱动][Android_IPC4_Bander1 _Bander]

### 二.getMinBufferSize ###

getMinBufferSize:获取AudioRecord对象所需的最小缓冲区大小。

#### 1.AR.getMinBufferSize(java) ####

static public int getMinBufferSize(int sampleRateInHz, int channelConfig, int audioFormat) {
        
            int channelCount = 0;
            switch (channelConfig) {
        
            //单声道
            case AudioFormat.CHANNEL_IN_DEFAULT: // AudioFormat.CHANNEL_CONFIGURATION_DEFAULT
            case AudioFormat.CHANNEL_IN_MONO:
            case AudioFormat.CHANNEL_CONFIGURATION_MONO:
                channelCount = 1;
                break;
            //立体声    
            case AudioFormat.CHANNEL_IN_STEREO:
            case AudioFormat.CHANNEL_CONFIGURATION_STEREO:
            case (AudioFormat.CHANNEL_IN_FRONT | AudioFormat.CHANNEL_IN_BACK):
                channelCount = 2;
                break;
            case AudioFormat.CHANNEL_INVALID:
            default:
                loge("getMinBufferSize(): Invalid channel configuration.");
                return ERROR_BAD_VALUE;
            }
    
    		//调用的jni中的native_get_min_buff_size函数
            int size = native_get_min_buff_size(sampleRateInHz, channelCount, audioFormat);
            if (size == 0) {
        
                return ERROR_BAD_VALUE;
            }
            else if (size == -1) {
        
                return ERROR;
            }
            else {
        
                return size;
            }
        }

#### 2.AR.getMinBufferSize(c++) ####

查看源码可知native\_get\_min\_buff\_size实现，是通过JNI函数进入framework/base/core/jni/android\_media\_AudioRecord.cpp的函数

native\_get\_min\_buff\_size函数到cpp中函数关联是通过android\_media\_AudioRecord.cpp中的函数数组来查看的

路径：framework/base/core/jni/android\_media\_AudioRecord.cpp

static const JNINativeMethod gMethods[] = {
        
        // name,               signature,  funcPtr
        {
        "native_start",         "(II)I",    (void *)android_media_AudioRecord_start},
        {
        "native_stop",          "()V",    (void *)android_media_AudioRecord_stop},
        {
        "native_setup",         "(Ljava/lang/Object;Ljava/lang/Object;[IIIII[ILjava/lang/String;J)I",
                                          (void *)android_media_AudioRecord_setup},
        {
        "native_finalize",      "()V",    (void *)android_media_AudioRecord_finalize},
        {
        "native_release",       "()V",    (void *)android_media_AudioRecord_release},
        {
        "native_read_in_byte_array",
                                 "([BIIZ)I",
                                         (void *)android_media_AudioRecord_readInArray<jbyteArray>},
        {
        "native_read_in_short_array",
                                 "([SIIZ)I",
                                         (void *)android_media_AudioRecord_readInArray<jshortArray>},
        {
        "native_read_in_float_array",
                                 "([FIIZ)I",
                                         (void *)android_media_AudioRecord_readInArray<jfloatArray>},
        {
        "native_read_in_direct_buffer","(Ljava/lang/Object;IZ)I",
                                           (void *)android_media_AudioRecord_readInDirectBuffer},
        {
        "native_get_buffer_size_in_frames",
                                 "()I", (void *)android_media_AudioRecord_get_buffer_size_in_frames},
        {
        "native_set_marker_pos","(I)I",   (void *)android_media_AudioRecord_set_marker_pos},
        {
        "native_get_marker_pos","()I",    (void *)android_media_AudioRecord_get_marker_pos},
        {
        "native_set_pos_update_period",
                                 "(I)I",   (void *)android_media_AudioRecord_set_pos_update_period},
        {
        "native_get_pos_update_period",
                                 "()I",    (void *)android_media_AudioRecord_get_pos_update_period},
    //native_get_min_buff_size-->android_media_AudioRecord_get_min_buff_size
        {
        "native_get_min_buff_size",
                                 "(III)I",   (void *)android_media_AudioRecord_get_min_buff_size},
        {
        "native_setInputDevice", "(I)Z", (void *)android_media_AudioRecord_setInputDevice},
        {
        "native_getRoutedDeviceId", "()I", (void *)android_media_AudioRecord_getRoutedDeviceId},
        {
        "native_enableDeviceCallback", "()V", (void *)android_media_AudioRecord_enableDeviceCallback},
        {
        "native_disableDeviceCallback", "()V",
                                            (void *)android_media_AudioRecord_disableDeviceCallback},
        {
        "native_get_timestamp", "(Landroid/media/AudioTimestamp;I)I",
                                           (void *)android_media_AudioRecord_get_timestamp},
    };

路径：framework/base/core/jni/android\_media\_AudioRecord.cpp

static jint android_media_AudioRecord_get_min_buff_size(JNIEnv *env,  jobject thiz,
        jint sampleRateInHertz, jint channelCount, jint audioFormat) {
        
    
        ALOGV(">> android_media_AudioRecord_get_min_buff_size(%d, %d, %d)",
              sampleRateInHertz, channelCount, audioFormat);
    
        size_t frameCount = 0;
        audio_format_t format = audioFormatToNative(audioFormat);
        //以传址的方式获取formatCount值。
        status_t result = AudioRecord::getMinFrameCount(&frameCount,
                sampleRateInHertz,
                format,
                audio_channel_in_mask_from_count(channelCount));
    
        if (result == BAD_VALUE) {
        
            return 0;
        }
        if (result != NO_ERROR) {
        
            return -1;
        }
        return frameCount * channelCount * audio_bytes_per_sample(format);
    }

根据最小的frameCount计算的buffersize。frameCount即一秒中有多少个frame。  
frame是音频中一个常见单位：表示音乐帧，即采样位数（每个采样数据占用字节大小）\*通道数（声音通道的个数）  
所以最小buffer= 每秒钟帧数 \* 通道数 \*采样位数（每个采样数据所占用的字节大小）

### 三.new AudioRecord ###

#### 1.AR.new AudioRecord(java) ####

public AudioRecord(int audioSource, int sampleRateInHz, int channelConfig, int audioFormat,
                int bufferSizeInBytes)
        throws IllegalArgumentException {
        
            this((new AudioAttributes.Builder())
                        .setInternalCapturePreset(audioSource)
                        .build(),
                    (new AudioFormat.Builder())
                        .setChannelMask(getChannelMaskFromLegacyConfig(channelConfig,
                                            true/*allow legacy configurations*/))
                        .setEncoding(audioFormat)
                        .setSampleRate(sampleRateInHz)
                        .build(),
                    bufferSizeInBytes,
                    AudioManager.AUDIO_SESSION_ID_GENERATE);
        }

可以看出这里调用了另一个构造方法  
下面的代码中省略了细节

public AudioRecord(AudioAttributes attributes, AudioFormat format, int bufferSizeInBytes,
                int sessionId) throws IllegalArgumentException {
        
            mRecordingState = RECORDSTATE_STOPPED;
    
    		//获取主线程的looper
            // remember which looper is associated with the AudioRecord instanciation
            if ((mInitializationLooper = Looper.myLooper()) == null) {
        
                mInitializationLooper = Looper.getMainLooper();
            }
    
           .....//各种属性的初始化
    
        
             //调用native层的native_setup，把自己的WeakReference(软引用)传进去
            //TODO: update native initialization when information about hardware init failure
            //      due to capture device already open is available.
            int initResult = native_setup( new WeakReference<AudioRecord>(this),
                    mAudioAttributes, sampleRate, mChannelMask, mChannelIndexMask,
                    mAudioFormat, mNativeBufferSizeInBytes,
                    session, ActivityThread.currentOpPackageName(), 0 /*nativeRecordInJavaObj*/);
            if (initResult != SUCCESS) {
        
                loge("Error code "+initResult+" when initializing native AudioRecord object.");
                return; // with mState == STATE_UNINITIALIZED
            }
    
            mSampleRate = sampleRate[0];
            mSessionId = session[0];
    
            mState = STATE_INITIALIZED;
        }

#### 2.AR.android\_media\_AudioRecord\_setup(c++) ####

目录：framework/base/core/jni/android\_media\_AudioRecord.cpp

android_media_AudioRecord_setup(JNIEnv *env, jobject thiz, jobject weak_this,
            jobject jaa, jint sampleRateInHertz, jint channelMask,
                    // Java channel masks map directly to the native definition
            jint audioFormat, jint buffSizeInBytes, jintArray jSession)
    {
        
      ...
      
        i
    
        // create an uninitialized AudioRecord object
        //创建一个未初始化的AudioRecord对象
        sp<AudioRecord> lpRecorder = new AudioRecord();
    
       ...
        const status_t status = lpRecorder->set(paa->source,
            sampleRateInHertz,
            format,        // word length, PCM
            channelMask,
            frameCount,
            recorderCallback,// callback_t
            lpCallbackData,// void* user
            0,             // notificationFrames,
            true,          // threadCanCallJava
            sessionId,
            AudioRecord::TRANSFER_DEFAULT,
            flags,
            paa);
    
        if (status != NO_ERROR) {
        
    		//初始化检查失败
            ALOGE("Error creating AudioRecord instance: initialization check failed with status %d.",
                    status);
            goto native_init_failure;
        }
    
       ...
    
     
        // save our newly created C++ AudioRecord in the "nativeRecorderInJavaObj" field
        // of the Java object
        //将我们新创建的c++ AudioRecord保存在Java对象的“nativeRecorderInJavaObj”字段中
        setAudioRecord(env, thiz, lpRecorder);
    
        // save our newly created callback information in the "nativeCallbackCookie" field
        // of the Java object (in mNativeCallbackCookie) so we can free the memory in finalize()
    	//将我们新创建的回调信息保存在Java对象的“nativeCallbackCookie”字段中(在mNativeCallbackCookie中)，
    	//这样我们就可以在finalize()中释放内存
    	env->SetLongField(thiz, javaAudioRecordFields.nativeCallbackCookie, (jlong)lpCallbackData);
    
        return (jint) AUDIO_JAVA_SUCCESS;
    
        // failure:
    native_init_failure:
        env->DeleteGlobalRef(lpCallbackData->audioRecord_class);
        env->DeleteGlobalRef(lpCallbackData->audioRecord_ref);
        delete lpCallbackData;
        env->SetLongField(thiz, javaAudioRecordFields.nativeCallbackCookie, 0);
    
        return (jint) AUDIORECORD_ERROR_SETUP_NATIVEINITFAILED;
    }

在这个函数里出来大部分的初始化，比较核心的内容主要有：

*  lpRecorder->set
 *  把刚创建的AudioRecord对象保存在Java层（“nativeRecorderInJavaObj”），通过getAudioRecord函数再获取
 *  回调信息保存在Java对象的“nativeCallbackCookie”字段中

#### 3.AR.set(c++) ####

android5.1版本AudioRecord  
目录：\\frameworks\\av\\media\\libmedia\\AudioRecord.cpp

status_t AudioRecord::set(
            audio_source_t inputSource,
            uint32_t sampleRate,
            audio_format_t format,
            audio_channel_mask_t channelMask,
            size_t frameCount,
            callback_t cbf,
            void* user,
            uint32_t notificationFrames,
            bool threadCanCallJava,
            int sessionId,
            transfer_type transferType,
            audio_input_flags_t flags,
            const audio_attributes_t* pAttributes)
    {
        
        ALOGV("set(): inputSource %d, sampleRate %u, format %#x, channelMask %#x, frameCount %zu, "
              "notificationFrames %u, sessionId %d, transferType %d, flags %#x",
              inputSource, sampleRate, format, channelMask, frameCount, notificationFrames,
              sessionId, transferType, flags);
    
       ....
       
    
        mNotificationFramesReq = notificationFrames;
        // mNotificationFramesAct is initialized in openRecord_l
      
    
        if (sessionId == AUDIO_SESSION_ALLOCATE) {
        
            mSessionId = AudioSystem::newAudioUniqueId();
        } else {
        
            mSessionId = sessionId;
        }
        ALOGV("set(): mSessionId %d", mSessionId);
    
        mFlags = flags;
        mCbf = cbf;
    
        if (cbf != NULL) {
        
            mAudioRecordThread = new AudioRecordThread(*this, threadCanCallJava);
            mAudioRecordThread->run("AudioRecord", ANDROID_PRIORITY_AUDIO);
        }
    
        // create the IAudioRecord
        //创建IAudioRecord
        status_t status = openRecord_l(0 /*epoch*/);
    
        if (status != NO_ERROR) {
        
            if (mAudioRecordThread != 0) {
        
                mAudioRecordThread->requestExit();   // see comment in AudioRecord.h
                mAudioRecordThread->requestExitAndWait();
                mAudioRecordThread.clear();
            }
            return status;
        }
    
        mStatus = NO_ERROR;
        mActive = false;
        mUserData = user;
        // TODO: add audio hardware input latency here
        mLatency = (1000*mFrameCount) / sampleRate;
        mMarkerPosition = 0;
        mMarkerReached = false;
        mNewPosition = 0;
        mUpdatePeriod = 0;
        AudioSystem::acquireAudioSessionId(mSessionId, -1);
        mSequence = 1;
        mObservedSequence = mSequence;
        mInOverrun = false;
    
        return NO_ERROR;
    }

继续追踪openRecord\_l  
目录：\\frameworks\\av\\media\\libmedia\\AudioRecord.cpp

// must be called with mLock held
    status_t AudioRecord::openRecord_l(size_t epoch)
    {
        
        status_t status;
        const sp<IAudioFlinger>& audioFlinger = AudioSystem::get_audio_flinger();
       ...
        audio_io_handle_t input;
        status = AudioSystem::getInputForAttr(&mAttributes, &input, (audio_session_t)mSessionId,
                                            mSampleRate, mFormat, mChannelMask, mFlags);
    
        if (status != NO_ERROR) {
        
            ALOGE("Could not get audio input for record source %d, sample rate %u, format %#x, "
                  "channel mask %#x, session %d, flags %#x",
                  mAttributes.source, mSampleRate, mFormat, mChannelMask, mSessionId, mFlags);
            return BAD_VALUE;
        }
        {
        
        // Now that we have a reference to an I/O handle and have not yet handed it off to AudioFlinger,
        // we must release it ourselves if anything goes wrong.
    
    	//现在我们有了一个I/O句柄的引用，但还没有将它交给AudioFlinger，
    	//如果出了什么问题，我们必须自己解决。
        size_t frameCount = mReqFrameCount;
        size_t temp = frameCount;   // temp may be replaced by a revised value of frameCount,
                                    // but we will still need the original value also
    								//temp可以替换为修改后的frameCount值，
    								//但是我们仍然需要原始值
    								
        int originalSessionId = mSessionId;
    
        // The notification frame count is the period between callbacks, as suggested by the server.
    	//通知帧计数是服务器建议的回调之间的周期。
    	size_t notificationFrames = mNotificationFramesReq;
    
        sp<IMemory> iMem;           // for cblk
        sp<IMemory> bufferMem;
        sp<IAudioRecord> record = audioFlinger->openRecord(input,
                                                           mSampleRate, mFormat,
                                                           mChannelMask,
                                                           &temp,
                                                           &trackFlags,
                                                           tid,
                                                           &mSessionId,
                                                           &notificationFrames,
                                                           iMem,
                                                           bufferMem,
                                                           &status);
        ALOGE_IF(originalSessionId != AUDIO_SESSION_ALLOCATE && mSessionId != originalSessionId,
                "session ID changed from %d to %d", originalSessionId, mSessionId);
    
      ...
    
     
    
        // Starting address of buffers in shared memory.
        // The buffers are either immediately after the control block,
        // or in a separate area at discretion of server.
        //共享内存中缓冲区的起始地址。
    	//缓冲区要么紧接在控制块之后，
    	//或在另一个由服务器决定的区域。
        void *buffers;
        if (bufferMem == 0) {
        
            buffers = cblk + 1;
        } else {
        
            buffers = bufferMem->pointer();
            if (buffers == NULL) {
        
                ALOGE("Could not get buffer pointer");
                return NO_INIT;
            }
        }
    
        // invariant that mAudioRecord != 0 is true only after set() returns successfully
        //只有在set()成功返回后，mAudioRecord != 0才为true的不变式
        if (mAudioRecord != 0) {
        
            mAudioRecord->asBinder()->unlinkToDeath(mDeathNotifier, this);
            mDeathNotifier.clear();
        }
        mAudioRecord = record;
        mCblkMemory = iMem;
        mBufferMemory = bufferMem;
        IPCThreadState::self()->flushCommands();
    
        mCblk = cblk;
        // note that temp is the (possibly revised) value of frameCount
        //注意，temp是frameCount的值(可能经过了修改)
        if (temp < frameCount || (frameCount == 0 && temp == 0)) {
        o00
            ALOGW("Requested frameCount %zu but received frameCount %zu", frameCount, temp);
        }
        frameCount = temp;
    
        mAwaitBoost = false;
        if (mFlags & AUDIO_INPUT_FLAG_FAST) {
        
            if (trackFlags & IAudioFlinger::TRACK_FAST) {
        
                ALOGV("AUDIO_INPUT_FLAG_FAST successful; frameCount %zu", frameCount);
                mAwaitBoost = true;
            } else {
        
                ALOGV("AUDIO_INPUT_FLAG_FAST denied by server; frameCount %zu", frameCount);
                // once denied, do not request again if IAudioRecord is re-created
                //一旦被拒绝，如果重新创建IAudioRecord，则不要再次请求
                mFlags = (audio_input_flags_t) (mFlags & ~AUDIO_INPUT_FLAG_FAST);
            }
        }
    
        // Make sure that application is notified with sufficient margin before overrun
        //确保在超限前通知应用程序有足够的余量
        if (notificationFrames == 0 || notificationFrames > frameCount) {
        
            ALOGW("Received notificationFrames %zu for frameCount %zu", notificationFrames, frameCount);
        }
        mNotificationFramesAct = notificationFrames;
    
        // We retain a copy of the I/O handle, but don't own the reference
        //我们保留了I/O句柄的副本，但不拥有引用
        mInput = input;
        mRefreshRemaining = true;
    
        mFrameCount = frameCount;
        // If IAudioRecord is re-created, don't let the requested frameCount
        // decrease.  This can confuse clients that cache frameCount().
        //如果IAudioRecord被重新创建，不要让请求的frameCount减少。这可能会使缓存frameCount()的客户机感到困惑。
        if (frameCount > mReqFrameCount) {
        
            mReqFrameCount = frameCount;
        }
    
        // update proxy
        mProxy = new AudioRecordClientProxy(cblk, buffers, mFrameCount, mFrameSize);
        mProxy->setEpoch(epoch);
        mProxy->setMinimum(mNotificationFramesAct);
    
        mDeathNotifier = new DeathNotifier(this);
        mAudioRecord->asBinder()->linkToDeath(mDeathNotifier, this);
    
        return NO_ERROR;
        }
    
    release:
        AudioSystem::releaseInput(input, (audio_session_t)mSessionId);
        if (status == NO_ERROR) {
        
            status = NO_INIT;
        }
        return status;
    }

比较重要的几行代码

*  audioFlinger = AudioSystem::get\_audio\_flinger();
 *  record = audioFlinger->openRecord
 *  mAudioR09ecord = record;

下面主要分析AudioSystem::get\_audio\_flinger()和audioFlinger->openRecord

#### 4.AS.get\_audio\_flinger(c++) ####

目录：\\frameworks\\av\\media\\libmedia\\AudioSystem

const sp<IAudioFlinger> AudioSystem::get_audio_flinger()
    {
        
        sp<IAudioFlinger> af;
        sp<AudioFlingerClient> afc;
        {
        
            Mutex::Autolock _l(gLock);
            if (gAudioFlinger == 0) {
        
                sp<IServiceManager> sm = defaultServiceManager();
                sp<IBinder> binder;
                do {
        
                    binder = sm->getService(String16("media.audio_flinger"));
                    if (binder != 0)
                        break;
                    ALOGW("AudioFlinger not published, waiting...");
                    usleep(500000); // 0.5 s
                } while (true);
                if (gAudioFlingerClient == NULL) {
        
                    gAudioFlingerClient = new AudioFlingerClient();
                } else {
        
                    if (gAudioErrorCallback) {
        
                        gAudioErrorCallback(NO_ERROR);
                    }
                }
                binder->linkToDeath(gAudioFlingerClient);
                gAudioFlinger = interface_cast<IAudioFlinger>(binder);
                LOG_ALWAYS_FATAL_IF(gAudioFlinger == 0);
                afc = gAudioFlingerClient;
            }
            af = gAudioFlinger;
        }
        if (afc != 0) {
        
            af->registerClient(afc);
        }
        return af;
    }

上面是一段很典型的利用binder的ipc

*  sm = defaultServiceManager()获取SM
 *  binder = sm->getService(String16(“media.audio\_flinger”));获取对应的服务
 *  gAudioFlinger = interface\_cast(binder)获取IAF对象，并将binder传入其中。

接下来我们继续追踪IAF.openRecord

#### 5.IAF.openRecord(c++) ####

目录：\\frameworks\\av\\media\\libmedia\\IAudioFlinger

virtual sp<IAudioRecord> openRecord(
                                    audio_io_handle_t input,
                                    uint32_t sampleRate,
                                    audio_format_t format,
                                    audio_channel_mask_t channelMask,
                                    size_t *pFrameCount,
                                    track_flags_t *flags,
                                    pid_t tid,
                                    int *sessionId,
                                    size_t *notificationFrames,
                                    sp<IMemory>& cblk,
                                    sp<IMemory>& buffers,
                                    status_t *status)
        {
        
            Parcel data, reply;
            sp<IAudioRecord> record;
            data.writeInterfaceToken(IAudioFlinger::getInterfaceDescriptor());
            ...data传输数据
            data.writeInt64(notificationFrames != NULL ? *notificationFrames : 0);
            cblk.clear();
            buffers.clear();
            status_t lStatus = remote()->transact(OPEN_RECORD, data, &reply);
            if (lStatus != NO_ERROR) {
        
                ALOGE("openRecord error: %s", strerror(-lStatus));
            } else {
        
               ...
                record = interface_cast<IAudioRecord>(reply.readStrongBinder());
                cblk = interface_cast<IMemory>(reply.readStrongBinder());
               ...
            }
            
            return record;
        }

重点分析

*  data传输数据
 *  remote()->transact调用远端Service进行ipc
 *  record = interface\_cast 获取远端返回的结果

#### 6.AF.openRecord(c++) ####

目录：frameworks/av/services/audioflinger/AudioFlinger.cpp

sp<IAudioRecord> AudioFlinger::openRecord(
            audio_io_handle_t input,
            uint32_t sampleRate,
            audio_format_t format,
            audio_channel_mask_t channelMask,
            size_t *frameCount,
            IAudioFlinger::track_flags_t *flags,
            pid_t tid,
            int *sessionId,
            size_t *notificationFrames,
            sp<IMemory>& cblk,
            sp<IMemory>& buffers,
            status_t *status)
    {
        
        sp<RecordThread::RecordTrack> recordTrack;
        sp<RecordHandle> recordHandle;
        sp<Client> client;
        status_t lStatus;
        int lSessionId;
    
        cblk.clear();
        buffers.clear();
    
       ...
    
        {
        
            Mutex::Autolock _l(mLock);
            RecordThread *thread = checkRecordThread_l(input);
            if (thread == NULL) {
        
                ALOGE("openRecord() checkRecordThread_l failed");
                lStatus = BAD_VALUE;
                goto Exit;
            }
    
            pid_t pid = IPCThreadState::self()->getCallingPid();
           ...
            // TODO: the uid should be passed in as a parameter to openRecord
            recordTrack = thread->createRecordTrack_l(client, sampleRate, format, channelMask,
                                                      frameCount, lSessionId, notificationFrames,
                                                      IPCThreadState::self()->getCallingUid(),
                                                      flags, tid, &lStatus);
            LOG_ALWAYS_FATAL_IF((lStatus == NO_ERROR) && (recordTrack == 0));
    
           
        }
    
       ...
        cblk = recordTrack->getCblk();
        buffers = recordTrack->getBuffers();
    
        // return handle to client
        recordHandle = new RecordHandle(recordTrack);
    
    Exit:
        *status = lStatus;
        return recordHandle;
    }

重点分析：

*  RecordThread \*thread = checkRecordThread\_l(input)
 *  recordTrack = thread->createRecordTrack\_l（…）
 *  recordHandle = new RecordHandle(recordTrack)

所以最终返回的RecordHandle里面包含RecordTrack，然后在其中有包含RecordThread

#### 7.流程梳理 ####

*  java new AR–>AR…android\_media\_AudioRecord\_setup–>AR.set–>AR.openRecord\_l
 *  在AR.openRecord\_l中调用AS.get\_audio\_flinger获取AF的binder
 *  在AR.openRecord\_l通过AF.openRecord获得一个sp，包含RecordHandle–>RecordTrack–>RecordThread
 *  将上面的sp赋值给mAudioRecord

### 四. startRecording ###

上面是创建AudioRecord对象的大部分源码，接下来我们看看startRecording过程

#### 1.AR.startRecording(java) ####

public void startRecording()
        throws IllegalStateException {
        
            if (mState != STATE_INITIALIZED) {
        
                throw new IllegalStateException("startRecording() called on an "
                        + "uninitialized AudioRecord.");
            }
    
            // start recording
            synchronized(mRecordingStateLock) {
        
                if (native_start(MediaSyncEvent.SYNC_EVENT_NONE, 0) == SUCCESS) {
        
                    handleFullVolumeRec(true);
                    mRecordingState = RECORDSTATE_RECORDING;
                }
            }
        }

这里没什么说的直接调用的Native层的代码

#### 2.AR.startRecording(c++) ####

目录：\\frameworks\\av\\media\\libmedia\\AudioRecord.cpp

android_media_AudioRecord_start(JNIEnv *env, jobject thiz, jint event, jint triggerSession)
    {
        
        sp<AudioRecord> lpRecorder = getAudioRecord(env, thiz);
        if (lpRecorder == NULL ) {
        
            jniThrowException(env, "java/lang/IllegalStateException", NULL);
            return (jint) AUDIO_JAVA_ERROR;
        }
    
        return nativeToJavaStatus(
                lpRecorder->start((AudioSystem::sync_event_t)event, triggerSession));
    }

#### 3.AR.start(c++) ####

目录：\\frameworks\\av\\media\\libmedia\\AudioRecord.cpp

status_t AudioRecord::start(AudioSystem::sync_event_t event, int triggerSession)
    {
        
        ALOGV("start, sync event %d trigger session %d", event, triggerSession);
    
        AutoMutex lock(mLock);
        if (mActive) {
        
            return NO_ERROR;
        }
    
        // reset current position as seen by client to 0
        //将客户端看到的当前位置重置为0
        mProxy->setEpoch(mProxy->getEpoch() - mProxy->getPosition());
        //使用processAudioBuffer()强制刷新剩余帧，因为在停止之前的最后一次读取可能是局部的。
        // force refresh of remaining frames by processAudioBuffer() as last
        // read before stop could be partial.
        mRefreshRemaining = true;
    
        mNewPosition = mProxy->getPosition() + mUpdatePeriod;
        int32_t flags = android_atomic_acquire_load(&mCblk->mFlags);
    
        status_t status = NO_ERROR;
        if (!(flags & CBLK_INVALID)) {
        
            ALOGV("mAudioRecord->start()");
            status = mAudioRecord->start(event, triggerSession);
            if (status == DEAD_OBJECT) {
        
                flags |= CBLK_INVALID;
            }
        }
        if (flags & CBLK_INVALID) {
        
            status = restoreRecord_l("start");
        }
    
        if (status != NO_ERROR) {
        
            ALOGE("start() status %d", status);
        } else {
        
            mActive = true;
            sp<AudioRecordThread> t = mAudioRecordThread;
            if (t != 0) {
        
                t->resume();
            } else {
        
                mPreviousPriority = getpriority(PRIO_PROCESS, 0);
                get_sched_policy(0, &mPreviousSchedulingGroup);
                androidSetThreadPriority(0, ANDROID_PRIORITY_AUDIO);
            }
        }
    
        return status;
    }

在这里最关键的一行代码 status = mAudioRecord->start(event, triggerSession);  
关于mAudioRecord变量在上一节我们有提过

所以： mAudioRecord.start–>RecordHandle.start–>RecordTrack.start–>RecordThread.start

#### 4.RT.start(c++) ####

目录：frameworks/av/services/audioflinger/Threads.cpp

status_t AudioFlinger::RecordThread::start(RecordThread::RecordTrack* recordTrack,
                                               AudioSystem::sync_event_t event,
                                               int triggerSession)
    {
        
        ALOGV("RecordThread::start event %d, triggerSession %d", event, triggerSession);
        sp<ThreadBase> strongMe = this;
        status_t status = NO_ERROR;
    
        if (event == AudioSystem::SYNC_EVENT_NONE) {
        
            recordTrack->clearSyncStartEvent();
        } else if (event != AudioSystem::SYNC_EVENT_SAME) {
        
            recordTrack->mSyncStartEvent = mAudioFlinger->createSyncEvent(event,
                                           triggerSession,
                                           recordTrack->sessionId(),
                                           syncStartEventCallback,
                                           recordTrack);
            // Sync event can be cancelled by the trigger session if the track is not in a
            // compatible state in which case we start record immediately
            if (recordTrack->mSyncStartEvent->isCancelled()) {
        
                recordTrack->clearSyncStartEvent();
            } else {
        
                // do not wait for the event for more than AudioSystem::kSyncRecordStartTimeOutMs
                recordTrack->mFramesToDrop = -
                        ((AudioSystem::kSyncRecordStartTimeOutMs * recordTrack->mSampleRate) / 1000);
            }
        }
    
        {
        
            // This section is a rendezvous between binder thread executing start() and RecordThread
            AutoMutex lock(mLock);
            if (mActiveTracks.indexOf(recordTrack) >= 0) {
        
                if (recordTrack->mState == TrackBase::PAUSING) {
        
                    ALOGV("active record track PAUSING -> ACTIVE");
                    recordTrack->mState = TrackBase::ACTIVE;
                } else {
        
                    ALOGV("active record track state %d", recordTrack->mState);
                }
                return status;
            }
    
            // TODO consider other ways of handling this, such as changing the state to :STARTING and
            //      adding the track to mActiveTracks after returning from AudioSystem::startInput(),
            //      or using a separate command thread
            recordTrack->mState = TrackBase::STARTING_1;
            mActiveTracks.add(recordTrack);
            mActiveTracksGen++;
            status_t status = NO_ERROR;
            if (recordTrack->isExternalTrack()) {
        
                mLock.unlock();
                status = AudioSystem::startInput(mId, (audio_session_t)recordTrack->sessionId());
                mLock.lock();
                // FIXME should verify that recordTrack is still in mActiveTracks
                if (status != NO_ERROR) {
        
                    mActiveTracks.remove(recordTrack);
                    mActiveTracksGen++;
                    recordTrack->clearSyncStartEvent();
                    ALOGV("RecordThread::start error %d", status);
                    return status;
                }
            }
            // Catch up with current buffer indices if thread is already running.
            // This is what makes a new client discard all buffered data.  If the track's mRsmpInFront
            // was initialized to some value closer to the thread's mRsmpInFront, then the track could
            // see previously buffered data before it called start(), but with greater risk of overrun.
    
            recordTrack->mRsmpInFront = mRsmpInRear;
            recordTrack->mRsmpInUnrel = 0;
            // FIXME why reset?
            if (recordTrack->mResampler != NULL) {
        
                recordTrack->mResampler->reset();
            }
            recordTrack->mState = TrackBase::STARTING_2;
            // signal thread to start
            mWaitWorkCV.broadcast();
            if (mActiveTracks.indexOf(recordTrack) < 0) {
        
                ALOGV("Record failed to start");
                status = BAD_VALUE;
                goto startError;
            }
            return status;
        }
    
    startError:
        if (recordTrack->isExternalTrack()) {
        
            AudioSystem::stopInput(mId, (audio_session_t)recordTrack->sessionId());
        }
        recordTrack->clearSyncStartEvent();
        // FIXME I wonder why we do not reset the state here?
        return status;
    }

继续追踪AudioSystem::startInput

#### 5.AS.startInput(c++) ####

目录：\\frameworks\\av\\media\\libmedia\\AudioSystem

status_t AudioSystem::startInput(audio_io_handle_t input,
                                     audio_session_t session)
    {
        
        const sp<IAudioPolicyService>& aps = AudioSystem::get_audio_policy_service();
        if (aps == 0) return PERMISSION_DENIED;
        return aps->startInput(input, session);
    }

AudioSystem::get\_audio\_policy\_service()的过程和上面的二.4中的代码内容差不多。都是获取一个IAudioPolicyService对象，在这个对象中有binder。

IAudioPolicyService.startInput会去通过binder调用AudioPolicyService.startInput，所以我们继续看AudioPolicyService.startInput

#### 6.APS.startInput(c++) ####

在AudioPolicyService.cpp,并没有找到startInput,然后在同目录下的AudioPolicyIntefaceImpl.cpp,找到其实现。

目录: frameworks/av/services/audiopolicy/AudioPolicyIntefaceImpl.cpp

status_t AudioPolicyService::startInput(audio_io_handle_t input,
                                            audio_session_t session)
    {
        
        if (mAudioPolicyManager == NULL) {
        
            return NO_INIT;
        }
        Mutex::Autolock _l(mLock);
    
        return mAudioPolicyManager->startInput(input, session);
    }

在AudioPolicyService.h中找到mAudioPolicyManager的变量类型

> AudioPolicyInterface \*mAudioPolicyManager;

而AudioPolicyInterface实际上由不同的平台各自实现的。当然也有默认的。  
这个类的作用是：

*  跟踪当前系统状态（可移动设备连接、电话状态、用户请求…）。系统状态更改和用户操作将通过audiopolicyInterface的方法通知音频策略管理器。
 *  创建audioTrack对象时接收的process getoutput（）
 *  同样处理从音频记录对象接收的getinput（）和putinput（）查询，并配置音频输入。

在这里我们看看Android默认实现的类

ps:这一块的源码真不好理，加上我半吊子的c++水平…

#### 7.APMB.startInput(c++) ####

目录：hardware/libhardware\_legacy/audio/AudioPolicyManagerBase.cpp

status_t AudioPolicyManagerBase::startInput(audio_io_handle_t input)
    {
        
        ALOGV("startInput() input %d", input);
        ssize_t index = mInputs.indexOfKey(input);
        if (index < 0) {
        
            ALOGW("startInput() unknown input %d", input);
            return BAD_VALUE;
        }
        AudioInputDescriptor *inputDesc = mInputs.valueAt(index);
    
    #ifdef AUDIO_POLICY_TEST
        if (mTestInput == 0)
    #endif //AUDIO_POLICY_TEST
        {
        
            // refuse 2 active AudioRecord clients at the same time except if the active input
            // uses AUDIO_SOURCE_HOTWORD in which case it is closed.
            audio_io_handle_t activeInput = getActiveInput();
            if (!isVirtualInputDevice(inputDesc->mDevice) && activeInput != 0) {
        
                AudioInputDescriptor *activeDesc = mInputs.valueFor(activeInput);
                if (activeDesc->mInputSource == AUDIO_SOURCE_HOTWORD) {
        
                    ALOGW("startInput() preempting already started low-priority input %d", activeInput);
                    stopInput(activeInput);
                    releaseInput(activeInput);
                } else {
        
                    ALOGW("startInput() input %d failed: other input already started", input);
                    return INVALID_OPERATION;
                }
            }
        }
    
        audio_devices_t newDevice = getDeviceForInputSource(inputDesc->mInputSource);
        if ((newDevice != AUDIO_DEVICE_NONE) && (newDevice != inputDesc->mDevice)) {
        
            inputDesc->mDevice = newDevice;
        }
    
        // automatically enable the remote submix output when input is started
        //启动输入时自动启用远程子混合输出
        if (audio_is_remote_submix_device(inputDesc->mDevice)) {
        
            setDeviceConnectionState(AUDIO_DEVICE_OUT_REMOTE_SUBMIX,
                    AudioSystem::DEVICE_STATE_AVAILABLE, AUDIO_REMOTE_SUBMIX_DEVICE_ADDRESS);
        }
    
        AudioParameter param = AudioParameter();
        param.addInt(String8(AudioParameter::keyRouting), (int)inputDesc->mDevice);
    
        int aliasSource = (inputDesc->mInputSource == AUDIO_SOURCE_HOTWORD) ?
                                            AUDIO_SOURCE_VOICE_RECOGNITION : inputDesc->mInputSource;
    
        param.addInt(String8(AudioParameter::keyInputSource), aliasSource);
        ALOGV("AudioPolicyManager::startInput() input source = %d", inputDesc->mInputSource);
    
        mpClientInterface->setParameters(input, param.toString());
    
        inputDesc->mRefCount = 1;
        return NO_ERROR;
    }

好了，我们看的最后面调用的mpClientInterface->setParameters(input, param.toString());

mpClientInterface的是 AudioPolicyClientInterface

所以继续追踪APC.setParameters  
目录：frameworks/av/services/audiopolicy/AudioPolicyClientInterface

void AudioPolicyService::AudioPolicyClient::setParameters(audio_io_handle_t io_handle,
                       const String8& keyValuePairs,
                       int delay_ms)
    {
        
        mAudioPolicyService->setParameters(io_handle, keyValuePairs.string(), delay_ms);
    }

继续追踪APS.setParameters

#### 8. APS.setParameters(c++) ####

目录：frameworks/av/services/audiopolicy/AudioPolicyService.cpp

void AudioPolicyService::setParameters(audio_io_handle_t ioHandle,
                                           const char *keyValuePairs,
                                           int delayMs)
    {
        
        mAudioCommandThread->parametersCommand(ioHandle, keyValuePairs,
                                               delayMs);
    }

追踪 mAudioCommandThread->parametersCommand  
还是在这个文件下

status_t AudioPolicyService::AudioCommandThread::parametersCommand(audio_io_handle_t ioHandle,
                                                                       const char *keyValuePairs,
                                                                       int delayMs)
    {
        
        sp<AudioCommand> command = new AudioCommand();
        command->mCommand = SET_PARAMETERS;
        sp<ParametersData> data = new ParametersData();
        data->mIO = ioHandle;
        data->mKeyValuePairs = String8(keyValuePairs);
        command->mParam = data;
        command->mWaitStatus = true;
        ALOGV("AudioCommandThread() adding set parameter string %s, io %d ,delay %d",
                keyValuePairs, ioHandle, delayMs);
        return sendCommand(command, delayMs);
    }

可以看到，在这里启动线程，并将SET\_PARAMETERS指令传过去

继续追踪sendCommand  
还是在这个文件下

status_t AudioPolicyService::AudioCommandThread::sendCommand(sp<AudioCommand>& command, int delayMs)
    {
        
        {
        
            Mutex::Autolock _l(mLock);
            insertCommand_l(command, delayMs);
            mWaitWorkCV.signal();
        }
        Mutex::Autolock _l(command->mLock);
        while (command->mWaitStatus) {
        
            nsecs_t timeOutNs = kAudioCommandTimeoutNs + milliseconds(delayMs);
            if (command->mCond.waitRelative(command->mLock, timeOutNs) != NO_ERROR) {
        
                command->mStatus = TIMED_OUT;
                command->mWaitStatus = false;
            }
        }
        return command->mStatus;
    }

在这里面进行了加锁操作  
同时调用了 insertCommand\_l（）函数  
继续追踪

/ insertCommand_l() must be called with mLock held
    void AudioPolicyService::AudioCommandThread::insertCommand_l(sp<AudioCommand>& command, int delayMs)
    {
        
        ssize_t i;  // not size_t because i will count down to -1
        Vector < sp<AudioCommand> > removedCommands;
        command->mTime = systemTime() + milliseconds(delayMs);
    
        // acquire wake lock to make sure delayed commands are processed
      	//获取唤醒锁以确保已处理延迟的命令   
        if (mAudioCommands.isEmpty()) {
        
            acquire_wake_lock(PARTIAL_WAKE_LOCK, mName.string());
        }
    
        // check same pending commands with later time stamps and eliminate them
        //用以后的时间戳检查相同的挂起命令并消除它们
        for (i = mAudioCommands.size()-1; i >= 0; i--) {
        
            sp<AudioCommand> command2 = mAudioCommands[i];
            // commands are sorted by increasing time stamp: no need to scan the rest of mAudioCommands
            if (command2->mTime <= command->mTime) break;
    
            // create audio patch or release audio patch commands are equivalent
            // with regard to filtering
            if ((command->mCommand == CREATE_AUDIO_PATCH) ||
                    (command->mCommand == RELEASE_AUDIO_PATCH)) {
        
                if ((command2->mCommand != CREATE_AUDIO_PATCH) &&
                        (command2->mCommand != RELEASE_AUDIO_PATCH)) {
        
                    continue;
                }
            } else if (command2->mCommand != command->mCommand) continue;
    
            switch (command->mCommand) {
        
            case SET_PARAMETERS: {
        
                ParametersData *data = (ParametersData *)command->mParam.get();
                ParametersData *data2 = (ParametersData *)command2->mParam.get();
                if (data->mIO != data2->mIO) break;
                ALOGV("Comparing parameter command %s to new command %s",
                        data2->mKeyValuePairs.string(), data->mKeyValuePairs.string());
                AudioParameter param = AudioParameter(data->mKeyValuePairs);
                AudioParameter param2 = AudioParameter(data2->mKeyValuePairs);
                for (size_t j = 0; j < param.size(); j++) {
        
                    String8 key;
                    String8 value;
                    param.getAt(j, key, value);
                    for (size_t k = 0; k < param2.size(); k++) {
        
                        String8 key2;
                        String8 value2;
                        param2.getAt(k, key2, value2);
                        if (key2 == key) {
        
                            param2.remove(key2);
                            ALOGV("Filtering out parameter %s", key2.string());
                            break;
                        }
                    }
                }
                // if all keys have been filtered out, remove the command.
                // otherwise, update the key value pairs
                ///如果已筛选出所有键，则删除该命令。
                //否则，更新键值对
                if (param2.size() == 0) {
        
                    removedCommands.add(command2);
                } else {
        
                    data2->mKeyValuePairs = param2.toString();
                }
                command->mTime = command2->mTime;
                // force delayMs to non 0 so that code below does not request to wait for
                // command status as the command is now delayed
                delayMs = 1;
            } break;
    
            ...其他指令...
        }
    
        // remove filtered commands
        for (size_t j = 0; j < removedCommands.size(); j++) {
        
            // removed commands always have time stamps greater than current command
            for (size_t k = i + 1; k < mAudioCommands.size(); k++) {
        
                if (mAudioCommands[k].get() == removedCommands[j].get()) {
        
                    ALOGV("suppressing command: %d", mAudioCommands[k]->mCommand);
                    mAudioCommands.removeAt(k);
                    break;
                }
            }
        }
        removedCommands.clear();
    
        // Disable wait for status if delay is not 0.
        // Except for create audio patch command because the returned patch handle
        // is needed by audio policy manager
        if (delayMs != 0 && command->mCommand != CREATE_AUDIO_PATCH) {
        
            command->mWaitStatus = false;
        }
    
        // insert command at the right place according to its time stamp
        ALOGV("inserting command: %d at index %zd, num commands %zu",
                command->mCommand, i+1, mAudioCommands.size());
        mAudioCommands.insertAt(command, i + 1);
    }

在这个函数里，将SET\_PARAMETERS指令放入SET\_PARAMETERS。我们继续来看，他是在什么地方执行该指令的

在threadLoop函数中

ool AudioPolicyService::AudioCommandThread::threadLoop()
    {
        
        nsecs_t waitTime = INT64_MAX;
    
        mLock.lock();
        while (!exitPending())
        {
        
            sp<AudioPolicyService> svc;
            while (!mAudioCommands.isEmpty() && !exitPending()) {
        
                nsecs_t curTime = systemTime();
                // commands are sorted by increasing time stamp: execute them from index 0 and up
                if (mAudioCommands[0]->mTime <= curTime) {
        
                    sp<AudioCommand> command = mAudioCommands[0];
                    mAudioCommands.removeAt(0);
                    mLastCommand = command;
    
                    switch (command->mCommand) {
        
                  ...其他指令
                    case SET_PARAMETERS: {
        
                        ParametersData *data = (ParametersData *)command->mParam.get();
                        ALOGV("AudioCommandThread() processing set parameters string %s, io %d",
                                data->mKeyValuePairs.string(), data->mIO);
                        command->mStatus = AudioSystem::setParameters(data->mIO, data->mKeyValuePairs);
                        }break;
                 ...其他指令
            }
            // release mLock before releasing strong reference on the service as
            // AudioPolicyService destructor calls AudioCommandThread::exit() which acquires mLock.
            mLock.unlock();
            svc.clear();
            mLock.lock();
            if (!exitPending() && mAudioCommands.isEmpty()) {
        
                // release delayed commands wake lock
                release_wake_lock(mName.string());
                ALOGV("AudioCommandThread() going to sleep");
                mWaitWorkCV.waitRelative(mLock, waitTime);
                ALOGV("AudioCommandThread() waking up");
            }
        }
        // release delayed commands wake lock before quitting
        if (!mAudioCommands.isEmpty()) {
        
            release_wake_lock(mName.string());
        }
        mLock.unlock();
        return false;
    }

好的，在这里我们看到调用了AudioSystem::setParameters方法，继续追踪

#### 9.AS.setParameters(c++) ####

目录：frameworks\\av\\media\\libmedia\\AudioSystem

status_t AudioSystem::setParameters(audio_io_handle_t ioHandle, const String8& keyValuePairs)
    {
        
        const sp<IAudioFlinger>& af = AudioSystem::get_audio_flinger();
        if (af == 0) return PERMISSION_DENIED;
        return af->setParameters(ioHandle, keyValuePairs);
    }

af = AudioSystem::get\_audio\_flinger(); 这行代码就很熟悉了，获取IAudioFlinger，其中包含bindr。

#### 10.AF.setParameters(c++) ####

status_t AudioFlinger::setParameters(audio_io_handle_t ioHandle, const String8& keyValuePairs)
    {
        
        ALOGV("setParameters(): io %d, keyvalue %s, calling pid %d",
                ioHandle, keyValuePairs.string(), IPCThreadState::self()->getCallingPid());
    
        // check calling permissions
        if (!settingsAllowed()) {
        
            return PERMISSION_DENIED;
        }
    
        // AUDIO_IO_HANDLE_NONE means the parameters are global to the audio hardware interface
        if (ioHandle == AUDIO_IO_HANDLE_NONE) {
        
            Mutex::Autolock _l(mLock);
            status_t final_result = NO_ERROR;
            {
        
                AutoMutex lock(mHardwareLock);
                mHardwareStatus = AUDIO_HW_SET_PARAMETER;
                for (size_t i = 0; i < mAudioHwDevs.size(); i++) {
        
                    audio_hw_device_t *dev = mAudioHwDevs.valueAt(i)->hwDevice();
                    status_t result = dev->set_parameters(dev, keyValuePairs.string());
                    final_result = result ?: final_result;
                }
                mHardwareStatus = AUDIO_HW_IDLE;
            }
            // disable AEC and NS if the device is a BT SCO headset supporting those pre processings
            AudioParameter param = AudioParameter(keyValuePairs);
            String8 value;
            if (param.get(String8(AUDIO_PARAMETER_KEY_BT_NREC), value) == NO_ERROR) {
        
                bool btNrecIsOff = (value == AUDIO_PARAMETER_VALUE_OFF);
                if (mBtNrecIsOff != btNrecIsOff) {
        
                    for (size_t i = 0; i < mRecordThreads.size(); i++) {
        
                        sp<RecordThread> thread = mRecordThreads.valueAt(i);
                        audio_devices_t device = thread->inDevice();
                        bool suspend = audio_is_bluetooth_sco_device(device) && btNrecIsOff;
                        // collect all of the thread's session IDs
                        KeyedVector<int, bool> ids = thread->sessionIds();
                        // suspend effects associated with those session IDs
                        for (size_t j = 0; j < ids.size(); ++j) {
        
                            int sessionId = ids.keyAt(j);
                            thread->setEffectSuspended(FX_IID_AEC,
                                                       suspend,
                                                       sessionId);
                            thread->setEffectSuspended(FX_IID_NS,
                                                       suspend,
                                                       sessionId);
                        }
                    }
                    mBtNrecIsOff = btNrecIsOff;
                }
            }
            String8 screenState;
            if (param.get(String8(AudioParameter::keyScreenState), screenState) == NO_ERROR) {
        
                bool isOff = screenState == "off";
                if (isOff != (AudioFlinger::mScreenState & 1)) {
        
                    AudioFlinger::mScreenState = ((AudioFlinger::mScreenState & ~1) + 2) | isOff;
                }
            }
            return final_result;
        }
    
        // hold a strong ref on thread in case closeOutput() or closeInput() is called
        // and the thread is exited once the lock is released
        sp<ThreadBase> thread;
        {
        
            Mutex::Autolock _l(mLock);
            thread = checkPlaybackThread_l(ioHandle);
            if (thread == 0) {
        
                thread = checkRecordThread_l(ioHandle);
            } else if (thread == primaryPlaybackThread_l()) {
        
                // indicate output device change to all input threads for pre processing
                AudioParameter param = AudioParameter(keyValuePairs);
                int value;
                if ((param.getInt(String8(AudioParameter::keyRouting), value) == NO_ERROR) &&
                        (value != 0)) {
        
                    for (size_t i = 0; i < mRecordThreads.size(); i++) {
        
                        mRecordThreads.valueAt(i)->setParameters(keyValuePairs);
                    }
                }
            }
        }
        if (thread != 0) {
        
            return thread->setParameters(keyValuePairs);
        }
        return BAD_VALUE;
    }

在AF.setParameters()的两个参数里，AudioParameter::keyRouting是音频策略管理器和AudioParameter::keyInputSource选定的输入设备参数。

此时通过队列发送命令，交给另一个线程处理。之后交回调用硬件适配层（HAL）的音频硬件。即接口AudioHardwareInterface的子类，由硬件平台厂家实现.

#### 11.流程梳理 ####

AR.startRecord  
–>AR.startRecording(c++)  
–>AR.start–>RecordHandle::start  
–>RecordTrack::start  
–> RecordHandle::start

–>RecordTrack::start  
–>RecordThread::start

–>AudioSystem::startInput  
–>AudioPolicyService::startInput  
–>AudioPolicyManagerBase(或厂家自己的管理器)::startInput  
–>AudioFlinger::setParameters  
–>AudioPolicyService::setParameters  
–>异步（通过向队列发送命令，然后由另一个线程去处理）

–>硬件适配层（HAL）的音频硬件（接口 AudioHardwareInterface的子类，由硬件平台厂家实现）

AudioRecord运行在应用程序进程中，AudioFlinger和AudioPolicyService运行在mediaserver进程中。应用程序调用到AudioFlinger，然后由AudioPolicyService确定输入设备，并将其作为设置到音频硬件中，进行相应的切换。

上面的切换完成后.  
在应用程序侧的AudioRecord的start函数中会让其线程ClientRecordThread开始工作。

....
     androidSetThreadPriority(0, ANDROID_PRIORITY_AUDIO);

在AudioFlinger中，录音线程的start函数（RecordThread::start）需要发送信号唤醒正在mWaitWorkCV上等待的线程循环，开始录音工作。录音线程循环进入睡眠等待的代码片段如下：

mWaitWorkCV.wait(mLock);//等待
    LOGV(“RecordThread: loop starting”);
    continue;
    
    
    LOGV(“Signal record thread”);
    mWaitWorkCV.signal();//在切换完输入设备后，唤醒线程

录音线程RecordThread的线程循环函数threadLoop比较复杂，要考虑单声道和双声道，要考虑16位还是8位的PCM数据，还要考虑是否需要重采样等因素。但总体原理是，使用HAL中的音频输入流AudioStreamIn的read函数读取音频数据数据到缓冲区mRsmpInBuffer：

// otherwise use the HAL / AudioStreamIn directly
    ssize_t bytesRead = mInput->stream->read(mInput->stream, &mRsmpInBuffer[rear * mChannelCount], mBufferSize);

### 五.read ###

#### 1.android\_media\_AudioRecord\_readInByteArray ####

前面的没什么可说的，  
read → native\_read\_in\_short\_array → android\_media\_AudioRecord\_readInByteArray:  
直接看android\_media\_AudioRecord\_readInByteArray中的实现  
目录:目录：framework/base/core/jni/android\_media\_AudioRecord.cpp

static jint android_media_AudioRecord_readInByteArray(JNIEnv *env,  jobject thiz,
                                                            jbyteArray javaAudioData,
                                                            jint offsetInBytes, jint sizeInBytes) {
        
        jbyte* recordBuff = NULL;
        // get the audio recorder from which we'll read new audio samples
        //拿起录音机，我们将从里面读取新的音频样本
        sp<AudioRecord> lpRecorder = getAudioRecord(env, thiz);
        if (lpRecorder == NULL) {
        
            ALOGE("Unable to retrieve AudioRecord object, can't record");
            return 0;
        }
    
        if (!javaAudioData) {
        
            ALOGE("Invalid Java array to store recorded audio, can't record");
            return 0;
        }
    
        // get the pointer to where we'll record the audio
        // NOTE: We may use GetPrimitiveArrayCritical() when the JNI implementation changes in such
        // a way that it becomes much more efficient. When doing so, we will have to prevent the
        // AudioSystem callback to be called while in critical section (in case of media server
        // process crash for instance)
        
        把指针指向我们录制音频的地方
    	//注意:当JNI实现以一种更有效的方式发生变化时，我们可能会使用getprimartivearraycritical()。
    	//这样做时，我们必须防止在关键部分调用AudioSystem回调(例如在媒体服务器进程崩溃的情况下)
        recordBuff = (jbyte *)env->GetByteArrayElements(javaAudioData, NULL);
    
        if (recordBuff == NULL) {
        
            ALOGE("Error retrieving destination for recorded audio data, can't record");
            return 0;
        }
    
        // read the new audio data from the native AudioRecord object
        //从本机AudioRecord对象读取新的音频数据
        ssize_t recorderBuffSize = lpRecorder->frameCount()*lpRecorder->frameSize();
        ssize_t readSize = lpRecorder->read(recordBuff + offsetInBytes,
                                            sizeInBytes > (jint)recorderBuffSize ?
                                                (jint)recorderBuffSize : sizeInBytes );
        env->ReleaseByteArrayElements(javaAudioData, recordBuff, 0);
    
        if (readSize < 0) {
        
            readSize = (jint)AUDIO_JAVA_INVALID_OPERATION;
        }
        return (jint) readSize;
    }

#### 2.AR.read ####

ssize_t AudioRecord::read(void* buffer, size_t userSize)
    {
        
        if (mTransfer != TRANSFER_SYNC) {
        
            return INVALID_OPERATION;
        }
    
        if (ssize_t(userSize) < 0 || (buffer == NULL && userSize != 0)) {
        
    		//检查。用户最有可能传递错误代码，这将使返回值不明确(实际上是size vs error)。
            // sanity-check. user is most-likely passing an error code, and it would
            // make the return value ambiguous (actualSize vs error).
            ALOGE("AudioRecord::read(buffer=%p, size=%zu (%zu)", buffer, userSize, userSize);
            return BAD_VALUE;
        }
    
        ssize_t read = 0;
        Buffer audioBuffer;
    
        while (userSize >= mFrameSize) {
        
            audioBuffer.frameCount = userSize / mFrameSize;
    
            status_t err = obtainBuffer(&audioBuffer, &ClientProxy::kForever);
            if (err < 0) {
        
                if (read > 0) {
        
                    break;
                }
                return ssize_t(err);
            }
    
            size_t bytesRead = audioBuffer.size;
            memcpy(buffer, audioBuffer.i8, bytesRead);
            buffer = ((char *) buffer) + bytesRead;
            userSize -= bytesRead;
            read += bytesRead;
    
            releaseBuffer(&audioBuffer);
        }
    
        return read;
    }

### 六.stop ###

stop → native\_stop → android\_media\_AudioRecord\_stop → lpRecorder->stop() :

void AudioRecord::stop()
    {
        
        AutoMutex lock(mLock);
        if (!mActive) {
        
            return;
        }
    
        mActive = false;
        mProxy->interrupt();
        mAudioRecord->stop();
        // the record head position will reset to 0, so if a marker is set, we need
        // to activate it again
        mMarkerReached = false;
        sp<AudioRecordThread> t = mAudioRecordThread;
        if (t != 0) {
        
            t->pause();
        } else {
        
            setpriority(PRIO_PROCESS, 0, mPreviousPriority);
            set_sched_policy(0, mPreviousSchedulingGroup);
        }
    }

停止的过程较为简单，应用程序侧跨进程调用AudioFlinger侧的stop，从而调用AudioSystem::stopInput，再到AudioPolicyService::stopInput，再到音频策略管理器（如AudioPolicyManagerBase）的stopInput，接着再使用AudioFlinger再到HAL音频硬件的setParameters将AudioParameter::keyRouting设置为0。在停止完输入设备后，应用程序侧需要停止其回调函数线程。

关于录音数据的采集，以及停止。由于时间原因，我就没有仔细的跟下去。有兴趣的可以自己跟一下。

### 七.参考资料 ###

[Android源码分析：录音AudioRecording][Android_AudioRecording]  
[AudioRecord 流程及源码分析][AudioRecord]

[Android_IPC4_Bander1 _Bander]: https://blog.csdn.net/qq_38499859/article/details/83833539
[Android_AudioRecording]: https://blog.csdn.net/brightming/article/details/50267589
[AudioRecord]: https://blog.csdn.net/qq_29028177/article/details/72726067