Camera_API2分析

《Camera_API2分析》由会员分享，可在线阅读，更多相关《Camera_API2分析（54页珍藏版）》请在装配图网上搜索。

1、. .一. android6.0源码分析之Camera API2.0简介前面几篇主要分析的是Android Camera API1.0的架构以 及初始化流程，而google在android5.0(Lollipop)开场对Camera的架构进展了调整，为了适应HAL3，新添加实现了 CameraDeviceClient，而Camera API1.0已经被deprecated(即可能在更新的版本里会不支持此API).1、Camera API2.0的架构图Camera API2.0下的Camera架构与API1.0有所区别，下面将给出Camera API2.0以及Camera HAL3.2+下的Ca

2、mera的总体架构图：由图可知，Java层要想与C+层的CameraService层进展通信，都是通过Java层的IPC Binder进制进展的，主要包括ICameraService.aidl以及ICameraDeviceUser.aidl两个接口来实现，其会在Java层维护一个CameraDeviceImpl即Camera设备的代理，而CameraService以及CameraDeviceImpl的初始化会在此文的第二，第三节进展分析。而Java层对Camera的具体操作的操作流程大致为，Java层通过Device代理发送一个CaptureRequest，而C+层进展相应的处理，再调用相应的

3、回调来通知Java相应的处理结果，并将相应的Capture数据保存在Surface Buffer里，这样Java层在回调函数中可以对数据进展相应的处理。而对于具体操作流程的分析，请参考文章开场时的Camera2相关文章的连接。2、Java层的CameraService的实现和应用从Camera API2开场，Camera的实现方式有所不同，最主要的区别是不再使用JNI来调用本地代码，从而获得本地CameraService，并实现其C/S 模式的通信，而是直接在Java层通过Java层的IPC Binder机制来获取Java层的CameraService的代理对象，从而直接在Java层获取本地的

4、CameraService与Camera Device进展相应的通信。相应的代码及目录：ICameraService.aidl:frameworks/base/core/java/android/hardware CameraService.cpp:frameworks/av/services/camera/libcameraservice CameraManager.java:frameworks/base/core/java/android/hardware/camera2获取CameraService的核心代码如下：/CameraManager.javaprvate void conne

5、ctCameraServiceLocked() if(mCameraService != null)return; /获取Binder IBinder cameraServiceBinder = ServiceManager.getService(CAMERA_SERVICE_BINDER_NAME); if(cameraServiceBinder = null) return; try cameraServiceBinder.linkToDeath(this,/*flags*/ 0); catch(RemoteException e) return; ICameraService camer

6、aServiceRaw = ICameraService.Stub.asInterface(cameraServiceBinder); /根据cameraServiceRaw 创立CameraService实例 ICameraService cameraService = CameraServiceDecorator.newInstance(cameraServiceRaw); . try /添加监听 cameraService.addListener(this); /赋值给mCameraService的全局变量 mCameraService = cameraService; catch(Ca

7、meraRuntimeException e) . 由代码可知，通过Java层的Binder从ServiceManager里获取了一个Java层的CameraService实例，在翻开Camera的流程中，会通过此CameraService(Native的CameraService)与Camera通信，而其中的通信通过ICameraDeviceUser来实现，接下来分析ICameraDeviceUser的实现。3、ICameraDeviceUser.aidl的通信实现Java层与C+ CameraService层之间的通信，通过封装了一个CameraDeviceUser来实现，它只是在Java

8、层使用了AIDL技术来实现Client，即在Java层维护了一个CameraDevice，这样的好处是响应速度更快，因为这样不需要通过每次进入Native层来完成通信，而可以通过Java层的IPC Binder机制即可完成。即API2.0通过AIDL实现一个接口ICameraDeviceUser，从而在Java层维护一个Camera proxy，之后的通信都是通过此代理CameraDeviceImpl来实现。相关代码及目录：ICameraDeviceUser.aidl:frameworks/base/core/java/android/hardware/camera2 ICameraDevic

9、eUser.cpp:frameworks/av/camera/camera2 CameraDeviceImpl.java:frameworks/base/core/java/android/hardware/camera2/impl获取Camera Device的Java层代理的核心代码如下：/CameraManager.javaprivate CameraDevice openCameraDeviceUserAsync(.) /初始化Camera Java层代理对象 CameraDevice device = null; try synchronized(mLock) /初始化ICamera

10、DeviceUser ICameraDeviceUser cameraUser = null; /初始化具体的CameraDevice代理 android.hardware.camera2.impl.CameraDeviceImpl deviceImpl = new android.hardware. camera2.impl.CameraDeviceImpl(cameraId,callback,handler,characteristics); BinderHolder holder = new BinderHolder(); ICameraDeviceCallbacks callbacks

11、 = deviceImpl.getCallbacks(); . try /如果支持HAL3.2+的devices if(supportsCamera2ApiLocked(cameraId) /获取CameraService ICameraService cameraService = CameraManagerGlobal.get().getCameraService(); . /连接设备 cameraService.connectDevice(callbacks,id,mContextgetOpPackageName() ,USE_CALLING_UID,holder); /通过Binder

12、获得翻开的Camera设备返回的Camera代理 cameraUser = ICameraDeviceUser.Stub.asInterface(holder.getBinder(); else/否那么用遗产API cameraUser = CameraDeviceUserShim.connectBinderShim(callbacks,id); catch(.) . /包装代理对象 deviceImpl.setRemoteDevice(cameraUser); device = deviceImpl; catch(.) . 返回Camera代理 return device;由代码可知，首先获

13、取CameraService，然后通过它来开启Camera，而开启成功后，C+层会返回一个Camera device代理对象，此处即为ICameraDeviceUser，所以在Java层对其进展相应的封装，变成一个CameraDeviceImpl对象，此后，只要需要对Camera进展操作，都会调用CameraDeviceImpl对象的相关方法，并通过ICameraDeviceUser以及Java IPC Binder来与本地的Camera device进展通信，至此，Camera API2.0的框架就分析完毕了，具体的操作，如Camera的初始化，preview,capture等流程的分析，请

14、看文章开场时，所列出的分析。二. android6.0源码分析之Camera2 HAL分析在上一篇文章对Camera API2.0的框架进展了简单的介绍，其中Camera HAL屏蔽了底层的实现细节，并且为上层提供了相应的接口，具体的HAL的原理，个人觉得老罗的文章Android硬件抽象层HAL概要介绍和学习方案分析的很详细，这里不做分析，本文将只分析Camera HAL的初始化等相关流程。1、Camera HAL的初始化Camera HAL的初始加载是在Native的CameraService初始化流程中的，而CameraService初始化是在Main_mediaServer.cpp的ma

15、in方法开场的：/Main_mediaServer.cppint main(int argc _unused, char* argv) sp proc(ProcessState:self(); /获取ServieManager sp sm = defaultServiceManager(); ALOGI(ServiceManager: %p, sm.get(); AudioFlinger:instantiate(); /初始化media效劳 MediaPlayerService:instantiate(); /初始化资源管理效劳 ResourceManagerService:instantia

16、te(); /初始化Camera效劳 CameraService:instantiate(); /初始化音频效劳 AudioPolicyService:instantiate(); SoundTriggerHwService:instantiate(); /初始化Radio效劳 RadioService:instantiate(); registerExtensions(); /开场线程池 ProcessState:self()-startThreadPool(); IPCThreadState:self()-joinThreadPool();其中，CameraService继承自Binder

17、Service，instantiate也是在BinderService中定义的，此方法就是调用publish方法，所以来看publish方法：/ BinderService.hstatic status_t publish(bool allowIsolated = false) sp sm(defaultServiceManager(); /将效劳添加到ServiceManager return sm-addService(String16(SERVICE:getServiceName(),new SERVICE(), allowIsolated);这里，将会把CameraService效劳参

18、加到ServiceManager进展管理。 而在前面的文章android6.0源码分析之Camera API2.0简介中，需要通过Java层的IPC Binder来获取此CameraService对象，在此过程中会初始CameraService的sp类型的对象，而对于sp，此处不做过多的分析，具体的可以查看深入理解Android卷中的第五章中的相关容。此处，在CameraService的构造时，会调用CameraService的onFirstRef方法：/CameraService.cppvoid CameraService:onFirstRef() BnCameraService:onFir

19、stRef(); . camera_module_t *rawModule; /根据CAMERA_HARDWARE_MODULE_ID(字符串camera)来获取camera_module_t对象 int err = hw_get_module(CAMERA_HARDWARE_MODULE_ID, (const hw_module_t *)&rawModule); /创立CameraModule对象 mModule = new CameraModule(rawModule); /模块初始化 err = mModule-init(); . /通过Module获取Camera的数量 mNumber

20、OfCameras = mModule-getNumberOfCameras(); mNumberOfNormalCameras = mNumberOfCameras; /初始化闪光灯 mFlashlight = new CameraFlashlight(*mModule, *this); status_t res = mFlashlight-findFlashUnits(); int latestStrangeCameraId = INT_MAX; for (int i = 0; i getCameraInfo(i, &info); . /如果Module版本高于2.4，找出冲突的设备参数

21、if (mModule-getModuleApiVersion() = CAMERA_MODULE_API_VERSION_2_4 & haveInfo) cost = info.resource_cost; conflicting_devices = info.conflicting_devices; conflicting_devices_length = info.conflicting_devices_length; /将冲突设备参加冲突set集中 std:set conflicting; for (size_t i = 0; i getModuleApiVersion() = CAM

22、ERA_MODULE_API_VERSION_2_1) mModule-setCallbacks(this); /假设大于2.2，那么设置供应商的Tag if (mModule-getModuleApiVersion() = CAMERA_MODULE_API_VERSION_2_2) setUpVendorTags(); /将此效劳注册到CameraDeviceFactory CameraDeviceFactory:registerService(this); CameraService:pingCameraServiceProxy();onFirstRef方法中，首先会通过HAL框架的hw

23、_get_module来获取CameraModule对象，然后会对其进展相应的初始化，并会进展一些参数的设置，如camera的数量，闪光灯的初始化，以及回调函数的设置等，到这里，Camera2 HAL的模块就初始化完毕了，下面给出初始化时序图：2、Camera HAL的open流程分析通过阅读android6.0源码发现，它提供了高通的Camera实现，并且提供了高通的Camera库，也实现了高通的Camera HAL的相应接口，对于高通的Camera，它在后台会有一个守护进程daemon，daemon是介于应用和驱动之间翻译ioctl的中间层(委托处理)。本节将以Camera中的open流程

24、为例，来分析Camera HAL的工作过程，在应用对硬件发出open请求后，会通过Camera HAL来发起open请求，而Camera HAL的open入口在QCamera2Hal.cpp进展了定义：/QCamera2Hal.cppcamera_module_t HAL_MODULE_INFO_SYM = /它里面包含模块的公共方法信息 common: camera_common, get_number_of_cameras: qcamera:QCamera2Factory:get_number_of_cameras, get_camera_info: qcamera:QCamera2Fac

25、tory:get_camera_info, set_callbacks: qcamera:QCamera2Factory:set_callbacks, get_vendor_tag_ops: qcamera:QCamera3VendorTags:get_vendor_tag_ops, open_legacy: qcamera:QCamera2Factory:open_legacy, set_torch_mode: NULL, init : NULL, reserved: 0;static hw_module_t camera_common = tag: HARDWARE_MODULE_TAG,

26、 module_api_version: CAMERA_MODULE_API_VERSION_2_3, hal_api_version: HARDWARE_HAL_API_VERSION, id: CAMERA_HARDWARE_MODULE_ID, name: QCamera Module, author: Qualcomm Innovation Center Inc, /它的方法数组里绑定了open接口 methods: &qcamera:QCamera2Factory:mModuleMethods, dso: NULL, reserved: 0;struct hw_module_meth

27、ods_t QCamera2Factory:mModuleMethods = /open方法的绑定 open: QCamera2Factory:camera_device_open,;Camera HAL层的open入口其实就是camera_device_open方法：/ QCamera2Factory.cppint QCamera2Factory:camera_device_open(const struct hw_module_t *module, const char *id, struct hw_device_t *hw_device) . return gQCamera2Factor

28、y-cameraDeviceOpen(atoi(id), hw_device);它调用了cameraDeviceOpen方法，而其中的hw_device就是最后要返回给应用层的CameraDeviceImpl在Camera HAL层的对象，继续分析cameraDeviceOpen方法：/ QCamera2Factory.cppint QCamera2Factory:cameraDeviceOpen(int camera_id, struct hw_device_t *hw_device) . /Camera2采用的Camera HAL版本为HAL3.0 if ( mHalDescriptors

29、camera_id.device_version = CAMERA_DEVICE_API_VERSION_3_0 ) /初始化QCamera3HardwareInterface对象，这里构造函数里将会进展configure_streams以及 /process_capture_result等的绑定 QCamera3HardwareInterface *hw = new QCamera3HardwareInterface( mHalDescriptorscamera_id.cameraId, mCallbacks); /通过QCamera3HardwareInterface来翻开Camera r

30、c = hw-openCamera(hw_device); . else if (mHalDescriptorscamera_id.device_version = CAMERA_DEVICE_API_VERSION_1_0) /HAL API为2.0 QCamera2HardwareInterface *hw = new QCamera2HardwareInterface(uint32_t)camera_id); rc = hw-openCamera(hw_device); . else . return rc;此方法有两个关键点：一个是QCamera3HardwareInterface对象

31、的创立，它是用户空间与核空间进展交互的接口；另一个是调用它的openCamera方法来翻开Camera，下面将分别进展分析。2.1 QCamera3HardwareInterface构造函数分析在它的构造函数里面有一个关键的初始化，即mCameraDevice.ops = &mCameraOps，它会定义Device操作的接口：/QCamera3HWI.cppcamera3_device_ops_t QCamera3HardwareInterface:mCameraOps = initialize: QCamera3HardwareInterface:initialize, /配置流数据的相关

32、处理 configure_streams: QCamera3HardwareInterface:configure_streams, register_stream_buffers: NULL, construct_default_request_settings: QCamera3HardwareInterface:construct_default_request_settings, /处理结果的接口 process_capture_request: QCamera3HardwareInterface:process_capture_request, get_metadata_vendor

33、_tag_ops: NULL, dump: QCamera3HardwareInterface:dump, flush: QCamera3HardwareInterface:flush, reserved: 0,; 其中，会在configure_streams中配置好流的处理handle：/QCamera3HWI.cppint QCamera3HardwareInterface:configure_streams(const struct camera3_device *device, camera3_stream_configuration_t *stream_list) /获得QCamer

34、a3HardwareInterface对象 QCamera3HardwareInterface *hw =reinterpret_cast(device-priv); . /调用它的configureStreams进展配置 int rc = hw-configureStreams(stream_list); . return rc;继续追踪configureStream方法：/QCamera3HWI.cppint QCamera3HardwareInterface:configureStreams(camera3_stream_configuration_t *streamList) . /初

35、始化Camera版本 al_version = CAM_HAL_V3; . /开场配置stream . /初始化相关Channel为NULL if (mMetadataChannel) delete mMetadataChannel; mMetadataChannel = NULL; if (mSupportChannel) delete mSupportChannel; mSupportChannel = NULL; if (mAnalysisChannel) delete mAnalysisChannel; mAnalysisChannel = NULL; /创立Metadata Chan

36、nel，并对其进展初始化 mMetadataChannel = new QCamera3MetadataChannel(mCameraHandle-camera_handle, mCameraHandle-ops, captureResultCb,&gCamCapabilitymCameraId-padding_info, CAM_QCOM_FEATURE_NONE, this); . /初始化 rc = mMetadataChannel-initialize(IS_TYPE_NONE); . /如果h/w support可用，那么创立分析stream的Channel if (gCamCapa

37、bilitymCameraId-hw_analysis_supported) mAnalysisChannel = new QCamera3SupportChannel(mCameraHandle-camera_handle, mCameraHandle-ops,&gCamCapabilitymCameraId-padding_info, CAM_QCOM_FEATURE_PP_SUPERSET_HAL3,CAM_STREAM_TYPE_ANALYSIS, &gCamCapabilitymCameraId-analysis_recommended_res,this); . bool isRaw

38、StreamRequested = false; /清空stream配置信息 memset(&mStreamConfigInfo, 0, sizeof(cam_stream_size_info_t); /为requested stream分配/QCamera3HWI.cppint QCamera3HardwareInterface:configureStreams(camera3_stream_configuration_t *streamList) . /初始化Ca相关的channel对象 for (size_t i = 0; i num_streams; i+) camera3_strea

39、m_t *newStream = streamList-streamsi; uint32_t stream_usage = newStream-usage; mStreamConfigInfo.stream_sizesmStreamConfigInfo.num_streams.width = (int32_t)newStream- width; mStreamConfigInfo.stream_sizesmStreamConfigInfo.num_streams.height = (int32_t)newStream- height; if (newStream-stream_type = C

40、AMERA3_STREAM_BIDIRECTIONAL|newStream-usage & GRALLOC_USAGE_HW_CAMERA_ZSL) &newStream-format = HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED & jpegStream) mStreamConfigInfo.typemStreamConfigInfo.num_streams = CAM_STREAM_TYPE_SNAPSHOT; mStreamConfigInfo.postprocess_maskmStreamConfigInfo.num_streams = CAM_Q

41、COM_FEATURE_NONE; else if(newStream-stream_type = CAMERA3_STREAM_INPUT) else switch (newStream-format) /为非zsl streams查找他们的format . if (newStream-priv = NULL) /为新的stream构造Channel switch (newStream-stream_type) /分类型构造 case CAMERA3_STREAM_INPUT: newStream-usage |= GRALLOC_USAGE_HW_CAMERA_READ; newStrea

42、m-usage |= GRALLOC_USAGE_HW_CAMERA_WRITE;/WR for inplace algos break; case CAMERA3_STREAM_BIDIRECTIONAL: . break; case CAMERA3_STREAM_OUTPUT: . break; default: break; /根据前面的得到的stream的参数类型以及format分别对各类型的channel进展构造 if (newStream-stream_type = CAMERA3_STREAM_OUTPUT | newStream-stream_type = CAMERA3_ST

43、REAM_BIDIRECTIONAL) QCamera3Channel *channel = NULL; switch (newStream-format) case HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED: /* use higher number of buffers for HFR mode */ . /创立Regular Channel channel = new QCamera3RegularChannel(mCameraHandle-camera_handle, mCameraHandle-ops, captureResultCb,&gCam

44、CapabilitymCameraId- padding_info,this,newStream,(cam_stream_type_t)mStreamConfigInfo.type mStreamConfigInfo.num_streams,mStreamConfigInfo.postprocess_mask mStreamConfigInfo.num_streams,mMetadataChannel,numBuffers); . mPendingRequestInfo newStream-max_buffers = channel-getNumBuffers(); newStream-pri

45、v = channel; break; case HAL_PIXEL_FORMAT_YCbCr_420_888: /创立YWV Channel . break; case HAL_PIXEL_FORMAT_RAW_OPAQUE: case HAL_PIXEL_FORMAT_RAW16: case HAL_PIXEL_FORMAT_RAW10: /创立Raw Channel . break; case HAL_PIXEL_FORMAT_BLOB: /创立QCamera3PicChannel . break; default: break; else if (newStream-stream_ty

46、pe = CAMERA3_STREAM_INPUT) newStream-max_buffers = MAX_INFLIGHT_REPROCESS_REQUESTS; else for (List:iterator it=mStreamInfo.begin();it != mStreamInfo.end(); it+) if (*it)-stream = newStream) (*it)-channel = (QCamera3Channel*) newStream-priv; break; else if (newStream-stream_type != CAMERA3_STREAM_INP

47、UT) mStreamConfigInfo.num_streams+; if (isZsl) if (mPictureChannel) mPictureChannel-overrideYuvSize(zslStream-width, zslStream-height); else if (mPictureChannel & m_bIs4KVideo) mPictureChannel-overrideYuvSize(videoWidth, videoHeight); /RAW DUMP channel if (mEnableRawDump & isRawStreamRequested = fal

48、se) cam_dimension_t rawDumpSize; rawDumpSize = getMaxRawSize(mCameraId); mRawDumpChannel = new QCamera3RawDumpChannel(mCameraHandle-camera_handle, mCameraHandle-ops,rawDumpSize,&gCamCapabilitymCameraId-padding_info, this, CAM_QCOM_FEATURE_NONE); . /进展相关Channel的配置 . /* Initialize mPendingRequestInfo

49、and mPendnigBuffersMap */ for (List:iterator i = mPendingRequestsList.begin(); i != mPendingRequestsList.end(); i+) clearInputBuffer(i-input_buffer); i = mPendingRequestsList.erase(i); mPendingFrameDropList.clear(); / Initialize/Reset the pending buffers list mPendingBuffersMap.num_buffers = 0; mPen

50、dingBuffersMap.mPendingBufferList.clear(); mPendingReprocessResultList.clear(); return rc;此方法容比较多，只抽取其中核心的代码进展说明，它首先会根据HAL的版本来对stream进展相应的配置初始化，然后再根据stream类型对stream_list的stream创立相应的Channel，主要有QCamera3MetadataChannel，QCamera3SupportChannel等，然后再进展相应的配置，其中QCamera3MetadataChannel在后面的处理capture request的时候会用到，这里就不做分析，而Camerametadata那么是Java层和CameraService之间传递的元数据，见android6.0源码分析之Camera API2.0简介中的Camera2架构图，至此，QCamera3HardwareInterface构造完毕，与本文相关的就是配置了mCameraDevice.ops。2.2 openCamera分析本节主要分析Module是如何翻开Camera的，openCamera的代码如下：/QCamera3HWI