使用camera的流程: openCamera() -> applySettings() -> setPreviewTexture() -> startPreview() ->autoFocus() -> takePicture()。
打开camera设备的大致过程:
1, 实例化CameraModule对象,即mCurrentModule表示当前的module,默认是photoModule。
2, 显示第一次运行的对话框FirstRunDialog,在dialog正常结束后,执行resume。
3, 根据mCurrentModule的类型,实际执行的PhotoModule.java中的resume,间接调用CameraProvider接口的实现类CameraController中的方法requestCamera,如果当期是api2,就会通过AndroidCamera2AgentImpl.java的实例,调用openCamera()$AndroidCamera2AgentImpl.java,实际调用的是父类CameraAgent.java中的方法openCamera。接下来会异步的方式执行打开camera的过程,具体就是CameraActions.OPEN_CAMERA消息的处理,这个消息的处理过程中调用Cameramanager.java的openCamera。
4, 通过CameraManager.java的openCamera,开启打开camera的过程。同时实例化CameraDevice.StateCallback类型的回调mCameraDeviceStateCallback,以便在camera打开后执行其onOpened方法,这个变量是在AndroidCamera2AgentImpl.java中定义的,在这个onOpened回调中,又会CameraOpenCallbackopenCallback的回调onCameraOpened,而这个onCameraOpened的实现在CameraController.java中,从CameraController通过onCameraOpened把camera打开成功的消息传递到CameraActivity,进一步传递到 PhotoModule.java中的onCameraAvailable,开启预览。
创建CameraDeviceImpl.java实例。
5, CameraManager.java中的openCameraDeviceUserAsync,打开一个到camera设备的connection,先是获取CameraService句柄,然后通过cameraService的connectDevice实现到camera hal层连接。
在openCameraDeviceUserAsync函数的最后,调用了deviceImpl.setRemoteDevice(cameraUser);同时携带了打开的camera客户端作为参数,指定到这里说明camera成功打开了,所以会执行onOpened# CameraDevice.StateCallback,以及StateCallbackKK的onOpened。
6, CameraService中的connectDevice,会通过makeClient创建CameraDeviceClient实例,这个实例对应了cameraservice.java中的BasicClient类型。
CameraDeviceClient的继承关系:
classCameraDeviceClient :
publicCamera2ClientBase<CameraDeviceClientBase>,
publiccamera2::FrameProcessorBase::FilteredListener
继承了Camera2ClientBase,CameraDeviceClientBase,实现了监听:FrameProcessorBase::FilteredListener。
Camera2ClientBase是一个模板类,其中的TClientBase是指CameraDeviceClientBase。
CameraDeviceClientBase又继承了CameraService::BasicClient,camera2::BnCameraDeviceUser。其中继承BnCameraDeviceUser使其具有了跨进程通信的能力。
所以实例化CameraDeviceClient时,这一系列类的构造函数都会被调用。
7, 在Camera2ClientBase的构造函数中创建了Camera3Device实例sp<CameraDeviceBase> mDevice;
CameraDeviceClient完成实例化后,执行其initialize方法,一方面调用Camera2ClientBase的initializeImpl,执行权限检查操作实际调用的是CameraService::BasicClient::startCameraOps()方法。
另一方面调用Camera3Device的initialize,打开hal设备,执行Hal层的初始化,这个过程中会创建Camera3BufferManager,启动RequestThread,拍照的request,预览的request都会在这个线程的threadLoop中得到处理。
在CameraDeviceClient的initializeImpl中,创建一个FrameProcessorBase实例,这是一个输出帧元数据处理线程,当设备由新的frames可用时,就会调用onResultAvailable方法。
8,在camera成功打开后,接着应用设置项,设置显示纹理,然后才是开启预览界面,其中在设置显示PreviewTexture时,创建了CameraCaptureSession,这是后期发送预览、拍照请求的基础。在CameraCaptureSession成功创建后,会回调CameraCaptureSession.StateCallback(相关实例在AndroidCamera2AgentImpl.java)的onConfigured,同时传回创建的CameraCaptureSession session对象供预览、拍照使用,然后把camerastate改成AndroidCamera2StateHolder.CAMERA_PREVIEW_READY,表示可以预览了。
(以上是Android O版本的调用流程)
接上一篇继续分析,下面就开始获取CameraService的句柄,调用CameraService中的connectDevice函数。
frameworks/base/core/java/android/hardware/camera2/CameraManager.java
private CameraDeviceopenCameraDeviceUserAsync(String cameraId,
CameraDevice.StateCallback callback, Handler handler, final int uid)
throws CameraAccessException@ CameraManager.java {
//首先是获取CameraService的句柄
ICameraServicecameraService = CameraManagerGlobal.get().getCameraService();
//向cameraService发送连接请求
cameraUser= cameraService.connectDevice(callbacks, id,
mContext.getOpPackageName(),uid);
}
先看获取CameraService的过程:
frameworks/base/core/java/android/hardware/camera2/CameraManager.java
CameraManagerGlobal.get().getCameraService();
public ICameraService getCameraService() {
connectCameraServiceLocked();
}
private void connectCameraServiceLocked() {
//这里通过serviceManager来查询cameraservice的句柄,对应的servicename是
// private static final String CAMERA_SERVICE_BINDER_NAME ="media.camera";跟前面提到的
//cameraservice的启动过程中注册cameraservice到servicemanager时,设置的服务名是一
//样的,所以这里得到的就是CameraService的句柄。
IBinder cameraServiceBinder =
ServiceManager.getService(CAMERA_SERVICE_BINDER_NAME);
//接下来是把查询到的cameraservice句柄这个Ibinder转成ICameraService,在注册
//cameraservice时,是把ICameraService转成Ibinder保存的,这里反向转化,由此可以推断//CameraService.cpp一定继承自IBinder,
ICameraService cameraService = ICameraService.Stub.asInterface(cameraServiceBinder);
//这里注册一个监听ICameraServiceListener,当一个新的camera可用时,有相应的回调
cameraService.addListener(this);
}
接着看下CameraService.cpp是不是继承自IBinder,
CameraService.h
class CameraService : public::android::hardware::BnCameraService,
其余省略,CameraService继承自BnCameraService,
BnCameraService.h
//对应的命名空间:android::hardware::
namespace android {
namespace hardware {
class BnCameraService : public::android::BnInterface<ICameraService>
}
}
这里的ICameraService是有ICameraService.aidl进过aidl工具自动生成的,ICameraService.aidl文件经转化后生成了ICameraService.java,ICameraService.h,ICameraService.cpp文件,早期版本aidl文件转化后只有.java文件生成。如果*.aidl文件被添加到的Android.mk,它的build Target是库,比如:include $(BUILD_SHARED_LIBRARY),那么就会自动生成.h,.cpp文件。
继续看BnInterface是不是根IBinder有关系:
IInterface.h
template<typename INTERFACE>
class BnInterface : public INTERFACE, public BBinder
可以看到BnInterface是一个模板类,这里的INTERFACE就是ICameraService,并且其继承自BBinder,
Frameworks/native/include/binder/Binder.h
class BBinder : public IBinder{}
从这里可以看出BBinder继承自IBinder。
从以上继承关系,可以知道connectDevice的调用流程:
CameraManager.java
private CameraDevice openCameraDeviceUserAsync(String cameraId,
CameraDevice.StateCallback callback, Handler handler, final int uid)
throws CameraAccessException {
......
try {
if (supportsCamera2ApiLocked(cameraId)) {
// Use cameraservice's cameradeviceclient implementation for HAL3.2+ devices
ICameraService cameraService = CameraManagerGlobal.get().getCameraService();
if (cameraService == null) {
throw new ServiceSpecificException(
ICameraService.ERROR_DISCONNECTED,
"Camera service is currently unavailable");
}
cameraUser = cameraService.connectDevice(callbacks, cameraId,
mContext.getOpPackageName(), uid);
} else {
}
} catch (ServiceSpecificException e) {
} catch (RemoteException e) {
}
}
return device;
}
out/target/common/obj/java_libraries/framework_intermediates/.../ICameraService.java
这是由ICameraService.aidl自动生成的.java文件
public interface ICameraService extends android.os.IInterface
{
/** Local-side IPC implementation stub class. */
public static abstract class Stub extends android.os.Binder implements android.hardware.ICameraService
{
private static class Proxy implements android.hardware.ICameraService
{
/**
* Open a camera device through the new camera API
* Only supported for device HAL versions >= 3.2
*/
@Override public android.hardware.camera2.ICameraDeviceUser connectDevice(android.hardware.camera2.ICameraDeviceCallbacks callbacks, java.lang.String cameraId, java.lang.String opPackageName, int clientUid) throws android.os.RemoteException
{
android.os.Parcel _data = android.os.Parcel.obtain();
android.os.Parcel _reply = android.os.Parcel.obtain();
android.hardware.camera2.ICameraDeviceUser _result;
try {
_data.writeInterfaceToken(DESCRIPTOR);
_data.writeStrongBinder((((callbacks!=null))?(callbacks.asBinder()):(null)));
_data.writeString(cameraId);
_data.writeString(opPackageName);
_data.writeInt(clientUid);
mRemote.transact(Stub.TRANSACTION_connectDevice, _data, _reply, 0);
_reply.readException();
_result = android.hardware.camera2.ICameraDeviceUser.Stub.asInterface(_reply.readStrongBinder());
}
return _result;
}
}
}
}
由mRemote.transact()开启跨进程的通信,经由IBinder,BpBinder,IPCThreadState把请求发到Binder驱动,由Binder驱动把请求发到cameraService服务端,针对同一个请求,client和server端的业务码是一致的。
CameraService.cpp的onTransact()方法会被调用:
status_t CameraService::onTransact(uint32_tcode, const Parcel& data, Parcel* reply, uint32_t flags) {
return BnCameraService::onTransact(code, data, reply, flags);à
}
ICameraService.cpp
::android::status_tBnCameraService::onTransact(uint32_t _aidl_code, const ::android::Parcel&_aidl_data, ::android::Parcel* _aidl_reply, uint32_t _aidl_flags) {
caseCall::CONNECTDEVICE:
::android::sp<::android::hardware::camera2::ICameraDeviceUser>_aidl_return;
//cameraManager.java中发起connectDevice时是带四个参数,这里加了一个ICameraDeviceUser类型的参数,并把这个出参作为reply的一部分返回给client端,这里的connectDevice才是真正调用到CameraService.cpp中的connectDevice方法。
::android::binder::Status_aidl_status(connectDevice(in_callbacks, in_cameraId, in_opPackageName,in_clientUid, &_aidl_return));
//把_aidl_return写入到返回的数据结构中
_aidl_ret_status=_aidl_reply->writeStrongBinder(::android::hardware::camera2::ICameraDeviceUser::asBinder(_aidl_return));
}
下面先看下怎么返回_aidl_return到client端的,cameraservice先把结果写到_aidl_reply这个parcel中,然后由Binder驱动在发到client端,其中的细节是client端发起请求后会进入睡眠,等server端有了处理结果,把这个结果写到了binder驱动后,client会被Binder驱动唤醒执行读取操作。这里接收结果的客户端是:
ICameraService.java中的Proxy:
public interface ICameraService:: publicstatic abstract class Stub:: private static class Proxy{
@Overridepublic android.hardware.camera2.ICameraDeviceUserconnectDevice(android.hardware.camera2.ICameraDeviceCallbacks callbacks, intcameraId, java.lang.String opPackageName, int clientUid) throwsandroid.os.RemoteException{
//这句代码是发送请求的开始
mRemote.transact(Stub.TRANSACTION_connectDevice,_data, _reply, 0);
//这句就是服务端处理后返回的结果,通过_reply.readStrongBinder()从parcel中读取结果,然后返回值给cameraManager。
_result= android.hardware.camera2.ICameraDeviceUser.Stub.asInterface(_reply.readStrongBinder());
return_result;
}
}
接着看CameraService.cpp中connectDevice都做了什么操作:
CameraService.cpp
Status CameraService::connectDevice(
constsp<hardware::camera2::ICameraDeviceCallbacks>& cameraCb,
intcameraId, onst String16& clientPackageName, int clientUid,
/*out*/sp<hardware::camera2::ICameraDeviceUser>*device){
//这里的device是出参,类型是ICameraDeviceUser,也是有ICameraDeviceUser.aidl自动生成的,这个对象跟CameraDeviceClient的实例client对应,CameraDeviceClient继承了BnCameraDeviceUser进而继承了ICameraDeviceUser,
sp<CameraDeviceClient>client = nullptr;
// connectHelper的定义在CameraService.h中
ret=connectHelper<hardware::camera2::ICameraDeviceCallbacks,CameraDeviceClient>
(cameraCb, id, CAMERA_HAL_API_VERSION_UNSPECIFIED,clientPackageName,
clientUid, USE_CALLING_PID, API_2, /*legacyMode*/ false,/*shimUpdateOnly*/ false,
/*out*/client);
*device= client;
}
CameraService.h
在O版本上,connectHelper的函数实现又被放在了frameworks/av/services/camera/libcameraservice/cameraservice.cppz中
//这是一个模板方法,CALLBACK是hardware::camera2::ICameraDeviceCallbacks,
//CLIENT是CameraDeviceClient。这个方法主要作用是生成CameraClient实例,并调用其inittialize方法。
template<class CALLBACK, classCLIENT>
binder::StatusCameraService::connectHelper(const sp<CALLBACK>& cameraCb, constString8& cameraId, int halVersion, const String16& clientPackageName,int clientUid, int clientPid, apiLevel effectiveApiLevel, bool legacyMode, boolshimUpdateOnly, /*out*/sp<CLIENT>& device) {
ret= makeClient(this, cameraCb, clientPackageName, id, facing, clientPid,
clientUid, getpid(),legacyMode, halVersion, deviceVersion, effectiveApiLevel,
/*out*/&tmp)
client= static_cast<CLIENT*>(tmp.get());
err= client->initialize(mModule)
}
CameraService.cpp
Status CameraService::makeClient(constsp<CameraService>& cameraService,
const sp<IInterface>& cameraCb, const String16& packageName,int cameraId,
int facing, int clientPid, uid_t clientUid, int servicePid, boollegacyMode,
int halVersion, int deviceVersion, apiLevel effectiveApiLevel,
/*out*/sp<BasicClient>* client){
//根据apiversion的不同,创建不同的CameraClient实例,这里创建CameraDeviceClient实例。
*client= new CameraDeviceClient(cameraService, tmp, packageName, cameraId,
facing, clientPid,clientUid, servicePid);
}
看下CameraDeviceClient的继承关系,
CameraDeviceClient.h
class CameraDeviceClient :
public Camera2ClientBase<CameraDeviceClientBase>,
public camera2::FrameProcessorBase::FilteredListener{}
struct CameraDeviceClientBase :
public CameraService::BasicClient,
public hardware::camera2::BnCameraDeviceUser{}
可以看到CameraDeviceClient继承了CameraService::BasicClient,并且实现了ICameraDeviceUser的这个Binder的api,同时还实现了帧处理线程的监听。
接着看CameraDeviceClient的构造函数:
CameraDeviceClient.cpp
CameraDeviceClient::CameraDeviceClient(constsp<CameraService>& cameraService,
constsp<hardware::camera2::ICameraDeviceCallbacks>& remoteCallback,
constString16& clientPackageName, int cameraId, int cameraFacing, int clientPid,
uid_tclientUid, int servicePid) :
Camera2ClientBase(cameraService,remoteCallback, clientPackageName,
cameraId,cameraFacing, clientPid, clientUid, servicePid),
主要是在参数初始化列表中调用了父类Camera2ClientBase的构造函数。
Camera2ClientBase.cpp
Camera2ClientBase是一个模板类,这里的TClientBase是CameraDeviceClientBase,可以从CameraDeviceClient的继承关系看出。除了调用父类TClientBase(CameraDeviceClientBase)的构造函数外,还创建Camera3Device实例。
template <typename TClientBase>
Camera2ClientBase<TClientBase>::Camera2ClientBase(
constsp<CameraService>& cameraService, constsp<TCamCallbacks>& remoteCallback,
constString16& clientPackageName, int cameraId, int cameraFacing, int clientPid,
uid_tclientUid, int servicePid):
TClientBase(cameraService, remoteCallback,clientPackageName,
cameraId, cameraFacing, clientPid, clientUid, servicePid),{
mDevice= new Camera3Device(cameraId);
}
接着把继承的构造函数看完,CameraDeviceClientBase又调用了父类CameraService::BasicClient,的构造函数,BasicClient的构造函数实现代码在CameraService中,主要做的事情是应用权限相关的,这块权限的处理不是很了解。
这一系列构造函数的执行,最重要的还是Camera2ClientBase中的Camera3Device实例的创建及紧接着的initialize方法的调用。
下面看initialize方法的调用流程:
CameraDeviceClient.cpp
status_tCameraDeviceClient::initialize(CameraModule *module){
res= Camera2ClientBase::initialize(module);
//这里注册了一个监听,mFrameProcessor是一个Thread,是一个输出帧元数据处理线程,
//处理预览回调相关的事情,这个线程会等待camera设备新的帧,然后调用监听接口的方法onResultAvailable,
//这个方法:CameraDeviceClient::onResultAvailable,又会执行回调:
// remoteCb->onResultReceived(result.mMetadata,result.mResultExtras);这个remoteCb是
//hardware::camera2::ICameraDeviceCallbacks类型的,这个callback实例是在
//Cameramanager.java中执行打开camera设备时创建的,然后由CameraService的connectDevice方法一路传递到CameraDeviceClient这里,所以这个回调实际的实现代码是:
// CameraDeviceImpl.java中的内部类CameraDeviceCallbacks的方法:onResultReceived。
mFrameProcessor->registerListener(FRAME_PROCESSOR_LISTENER_MIN_ID,
FRAME_PROCESSOR_LISTENER_MAX_ID,/*listener*/this, /*sendPartials*/true);
}
其中的CameraDeviceImpl.java中的内部类CameraDeviceCallbacks,在它的被回调方法onResultReceived中,通过mCaptureCallbackMap取出执行开启预览、拍照时传入的callback(CameraDeviceImpl.CaptureCallback),这个CameraDeviceImpl.CaptureCallback实际是对应用程序端传过来的CameraCaptureSession.CaptureCallback的封装,具体封装是通过createCaptureCallbackProxy方法实现的,所以当有一帧远数据可用时,最终层层回调会执行CameraCaptureSession.CaptureCallback的onCaptureProgressed,onCaptureCompleted方法,将元数据传给应用端。
Camera2ClientBase.cpp
status_t Camera2ClientBase<TClientBase>::initialize(CameraModule*module){
//这里的mDevice是Camera3Device类的实例。
res= mDevice->initialize(module);
}
Camera3Device.cpp
status_tCamera3Device::initialize(CameraModule *module){
//调用CameraModule的open方法打开HAL设备,从这里开始就进入到了HAL层,HAL设备对应的结构体类型是camera3_device_t,module就是CameraModule的实例,这个实例的创建是在CameraService第一次被引用时在其void CameraService::onFirstRef()函数中,mModule = new CameraModule(rawModule);这部分跟CameraService的启动有关系。
res= module->open(deviceName.string(),
reinterpret_cast<hw_device_t**>(&device));
//初始化HAL层设备,
res= device->ops->initialize(device, this);
//创建Buffer管理器。
mBufferManager= new Camera3BufferManager();
res= find_camera_metadata_ro_entry(info.static_camera_characteristics,
ANDROID_CONTROL_AE_LOCK_AVAILABLE,&aeLockAvailableEntry);
//开启一个请求队列线程,run方法后它的threadLoop方法就会执行。
mRequestThread= new RequestThread(this, mStatusTracker, device, aeLockAvailable);
res= mRequestThread->run(String8::format("C3Dev-%d-ReqQueue",mId).string());
//创建准备流的线程,但是并没有马上运行这个线程,而是等到调用Camera3Device的prepare方法时,根据需要开启线程,什么时候调用了Camera3Device的prepare方法呢?这个我没打log跟,一种可能的情况是当创建一个session时,预分配缓存时调用。
mPreparerThread= new PreparerThread();
}
在O版本,打开hal设备是通过CameraProviderManager来完成的。
-------------------------------------------------------------------------------------------------------------------
到这里Camera设备的打开就完成了。紧接着的就是开启预览。
再回到应用层,CaptureModule.java
上面camera设备打开的过程是从openCameraAndStartPreview中open方法开始的,当camera成功打开后,会回调onCameraOpened,在这个回调中通过camera.startPreview启动预览。
private void openCameraAndStartPreview() {
mOneCameraOpener.open(cameraId,captureSetting, mCameraHandler, mainThread,
imageRotationCalculator,mBurstController, mSoundPlayer,
newOpenCallback() {
@Override
publicvoid onCameraOpened(@Nonnull final OneCamera camera) {
mCamera= camera;
updatePreviewBufferDimension();
updatePreviewBufferSize();
camera.startPreview(newSurface(getPreviewSurfaceTexture()),
newCaptureReadyCallback() {
@Override
publicvoid onReadyForCapture() {
//开启预览,要先创建拍照session,如果session成功创建,会回调到这里,说明预览已经准备好了,可以准备拍照了,
mMainThread.execute(newRunnable() {
public void run() {
onPreviewStarted();
onReadyStateChanged(true);
}
}
}
});
}
},,);
}
OneCameraImpl.java
public void startPreview(SurfacepreviewSurface, CaptureReadyCallback listener) {
setupAsync(mPreviewSurface,listener);
}
开启异步拍照session
private void setupAsync(final SurfacepreviewSurface, final CaptureReadyCallback listener) {
mCameraHandler.post(new Runnable() {
@Override
public void run() {
setup(previewSurface,listener);
}
});
}
private void setup(Surface previewSurface,final CaptureReadyCallback listener) {
mDevice.createCaptureSession(outputSurfaces,new CameraCaptureSession.StateCallback() {
public void onConfigured(CameraCaptureSessionsession) {
mCaptureSession = session;
boolean success =repeatingPreview(null);
if (success) {
listener.onReadyForCapture();
}
}
}
Session的创建是调用到CameraDeviceImpl.java中的createCaptureSession,进而调用configureStreamsChecked配置流,所谓session创建是否成功,就是是否成功配置了输入输出流,如果成功了设备会block进入idle,并且回调StateCallbackKK. onIdle();配置可能会失败,比如格式大小不支持,这时回调StateCallbackKK. onUnconfigured()。不管这个配置成功与否,都会new一个CameraCaptureSessionImpl实例,如果配置是成功的,就会回调上面CameraCaptureSession.StateCallback()中的onConfigured,同时把CameraCaptureSessionImpl实例作为onConfigured的参数传到OneCameraImpl.java中,就是mCaptureSession= session,也就是只有配置成功了,才会接着发出预览的request即repeatingPreview。
OneCameraImpl.java
private boolean repeatingPreview(Objecttag) {
CaptureRequest.Builderbuilder = mDevice.
createCaptureRequest(CameraDevice.TEMPLATE_PREVIEW);
mCaptureSession.setRepeatingRequest(builder.build(),mCaptureCallback,
mCameraHandler);
}
如果request成功build,就可以准备拍照了。
CameraCaptureSessionImpl.java
public synchronized int setRepeatingRequest(CaptureRequest request, CaptureCallback callback,
Handlerhandler) throws CameraAccessException {
//这里会把提交的request的requestID入队,因为session的创建要配置camera设备的内部管道,要分配内存缓冲区,很耗时,所以capture request提交后会先入队,等session ready就开始执行capture
return addPendingSequence(mDeviceImpl.setRepeatingRequest(request,
createCaptureCallbackProxy(handler,callback), mDeviceHandler));
}
其中的参数createCaptureCallbackProxy(handler, callback)是指定了回调从CameraDeviceImpl.CaptureCallback到CameraCaptureSession.CaptureCallback的,比较重要的一个方法是它的onCaptureCompleted。其中callback是OneCameraImpl.java中的mCaptureCallback。
继续看Request的创建提交。
CameraDeviceImpl.java
public int setRepeatingRequest(CaptureRequest request, CaptureCallback callback,
Handlerhandler) throws CameraAccessException {
return submitCaptureRequest(requestList, callback, handler, /*streaming*/true);
}
private int submitCaptureRequest(List<CaptureRequest> requestList, CaptureCallbackcallback,
Handlerhandler, boolean repeating) throws CameraAccessException {
requestInfo= mRemoteDevice.submitRequestList(requestArray, repeating);
if(callback != null) {
mCaptureCallbackMap.put(requestInfo.getRequestId(),
new CaptureCallbackHolder(
callback, requestList,handler, repeating, mNextSessionId - 1));
}
}
通过ICameraDeviceUserWrapper的实例mRemoteDevice提交request,
ICameraDeviceUserWrapper.java
public SubmitInfosubmitRequest(CaptureRequest request, boolean streaming){
return mRemoteDevice.submitRequest(request, streaming);
}
这里的mRemoteDevice类型是ICameraDeviceUser,这个实例是通过cameraService的connectDevice方法返回的。
前面我们说过ICameraDeviceUser对应了CameraDeviceClient,CameraDeviceClient对应了CameraService的内部类Client。
ICameraDeviceUser.java,ICameraDeviceUser.cpp都是aidl文件自动生成的。
这样request就借助aidl的跨进程从ICameraDeviceUser.java到了CameraDeviceClient.cpp这边,进而跟cameraservice建立了联系。
继续看submitCaptureRequest对callback的处理,把callback做了一个包装放在了mCaptureCallbackMap中跟requestID做了关联,那么什么时候回调了这个callback呢?
前面在说CameraDeviceClient.cpp的初始化时提到,mFrameProcessor是一个输出帧元数据处理线程,处理预览回调相关的事情,这个线程会等待caemra设备新的帧,然后然后调用监听接口的方法onResultAvailable,这个方法:CameraDeviceClient::onResultAvailable,又会执行回调:remoteCb->onResultReceived(result.mMetadata,result.mResultExtras);这个remoteCb是hardware::camera2::ICameraDeviceCallbacks类型的,这个callback实例是在Cameramanager.java中执行打开camera设备时创建的,然后由CameraService的connectDevice方法一路传递到CameraDeviceClient这里,所以这个回调实际的实现代码是:
CameraDeviceImpl.java中的内部类CameraDeviceCallbacks的方法:onResultReceived。
我们看CameraDeviceImpl.java的onResultReceived方法:
CameraDeviceImpl.java
public voidonResultReceived(CameraMetadataNative result,
CaptureResultExtrasresultExtras) throws RemoteException {
//根据requestId,取得holder
intrequestId = resultExtras.getRequestId();
finalCaptureCallbackHolder holder =
CameraDeviceImpl.this.mCaptureCallbackMap.get(requestId);
finalCaptureRequest request = holder.getRequest(resultExtras.getSubsequenceId());
//通过holder得到callback执行回调,同时传入数据resultAsCapture。
holder.getCallback().onCaptureProgressed(CameraDeviceImpl.this,
request, resultAsCapture);
holder.getCallback().onCaptureCompleted(CameraDeviceImpl.this,
request, resultAsCapture);
}
这样就把底层的数据传到Framework,进一步传到了应用层。
另外获取元数据,也可以通过ImageReader。
在实例化Camera实例时,获取ImageReader对象,同时设置它的监听,当有一张新的图片可用时,回调其onImageAvailable接口,在这个onImageAvailable接口中,读取、存储元数据。
OneCameraImpl.java
private final ImageReader mCaptureImageReader;
ImageReader.OnImageAvailableListener mCaptureImageListener =
new ImageReader.OnImageAvailableListener() {
@Override
public void onImageAvailable(ImageReader reader) {
// Add the image data to the latest in-flight capture.
// If all the data for that capture is complete, store the
// image data.
InFlightCapture capture = null;
synchronized (mCaptureQueue) {
if (mCaptureQueue.getFirst().setImage(reader.acquireLatestImage())
.isCaptureComplete()) {
capture = mCaptureQueue.removeFirst();
}
}
if (capture != null) {
onCaptureCompleted(capture);
}
}
};
获取ImageReader实例,设置监听。
OneCameraImpl(CameraDevice device, CameraCharacteristics characteristics, Size pictureSize) {
mCaptureImageReader = ImageReader.newInstance(pictureSize.getWidth(),
pictureSize.getHeight(),
sCaptureImageFormat, 2);
mCaptureImageReader.setOnImageAvailableListener(mCaptureImageListener,
mCameraHandler);
}
拍照完成时,会回调onCaptureCompleted。
oneCameraImpl.java
private void onCaptureCompleted(InFlightCapture capture) {
// Experimental support for writing RAW. We do not have a usable JPEG
// here, so we don't use the usual capture session mechanism and instead
// just store the RAW file in its own directory.
// TODO: If we make this a real feature we should probably put the DNGs
// into the Camera directly.
//可以存储元数据
if (sCaptureImageFormat == ImageFormat.RAW_SENSOR) {
if (!RAW_DIRECTORY.exists()) {
if (!RAW_DIRECTORY.mkdirs()) {
throw new RuntimeException("Could not create RAW directory.");
}
}
File dngFile = new File(RAW_DIRECTORY, capture.session.getTitle() + ".dng");
writeDngBytesAndClose(capture.image, capture.totalCaptureResult,
mCharacteristics, dngFile);
} else {
//也可以存储jpg。
// Since this is not an HDR+ session, we will just save the
// result.
byte[] imageBytes = acquireJpegBytesAndClose(capture.image);
saveJpegPicture(imageBytes, capture.parameters, capture.session,
capture.totalCaptureResult);
}
broadcastReadyState(true);
capture.parameters.callback.onPictureTaken(capture.session);
}
调用writeDngBytesAndClose存储元数据,
private static void writeDngBytesAndClose(Image image, TotalCaptureResult captureResult,
CameraCharacteristics characteristics, File dngFile) {
try (DngCreator dngCreator = new DngCreator(characteristics, captureResult);
FileOutputStream outputStream = new FileOutputStream(dngFile)) {
// TODO: Add DngCreator#setThumbnail and add the DNG to the normal
// filmstrip.
dngCreator.writeImage(outputStream, image);
outputStream.close();
image.close();
} catch (IOException e) {
Log.e(TAG, "Could not store DNG file", e);
return;
}
Log.i(TAG, "Successfully stored DNG file: " + dngFile.getAbsolutePath());
}
来源:oschina
链接:https://my.oschina.net/u/920274/blog/4319663