Android4.0.3 显示系统深入理解

1. 简介      

        网上已经有很多兄弟对Android的显示系统做了深入解剖，很是佩服。可最近小弟在研究Android4.0时发现出入比较大，也许是Android4.0的修改比较多吧！因为小弟没有看Android4.0以前的代码。

       面对这么复杂一个Android显示系统，如何入手呢? 根据以前的经验，不管它有多么复杂，其功能不就是以下三步曲吗？

　　1）显示系统的创建及初始化

       2）画图

       3）销毁

       哪我的分析就从显示系统的创建及初始化开始吧！由于小弟对Java没有什么研究兴趣，所有重点就分析Native部分。当然Native的入口就在android_view_Surface.cpp中，此文件主要包含以下两部分给Java层调用：

       1）gSurfaceSessionMethods： 操作SurfaceSession的方法

       2）gSurfaceMethods：操作Surface的方法

2. android_view_Surface.cpp

 

2.1 SurfaceSession操作方法

 static JNINativeMethod gSurfaceSessionMethods[] = {
     {"init",     "()V",  (void*)SurfaceSession_init }, //创建SurfaceComposerClient
     {"destroy",  "()V",  (void*)SurfaceSession_destroy }, //直接销毁SurfaceComposerClient
     {"kill",     "()V",  (void*)SurfaceSession_kill },//先clear,再销毁SurfaceComposerClient
 };

 2.1.1 SurfaceSession_init

        其功能如下：

        1）创建SurfaceComposerClient对象

        2）调用SurfaceComposerClient::onFirstRef方法

        现在已经进入到SurfaceComposerClient的地盘，根据其名字含义，它应该是一个进行Surface合成的客户端，通过它发命令给SurfaceFlinger来进行需要的操作。其初始化流程如下图所示：

2.1.2 SurfaceComposerClient.cpp中的宝贝

        为了方便后面的理解，先看看SurfaceComposerClient中有些什么宝贝来完成这个任务。在其中定义了如下几个类：

2.1.2.1 ComposerService(获取SurfaceFlinger服务)

        一看到名字为Service，应该是用于从SurfaceFlinger中获取Service以建立连接关系<它是一个单实例，一个进程有且只有一个实例对象>，然后供后面进行相关的操作。其构造函数代码如下：      

 class ComposerService : public Singleton<ComposerService>
 {
     //实质为BpSurfaceComposer,通过它与SurfaceFlinger进行通信，
     //BnSurfaceComposer是SurfaceFlinger基类中的一个
     sp<ISurfaceComposer> mComposerService;

     //实质为BpMemoryHeap,它在SurfaceFlinger中对应为管理一个4096字节的
     //一个MemoryHeapBase对象，在SurfaceFlinger::readyToRun中创建
     sp<IMemoryHeap> mServerCblkMemory;

     //为MemoryHeapBase管理的内存在用户空间的基地址，通过mmap而来，
     //具体见MemoryHeapBase::mapfd
     surface_flinger_cblk_t volatile* mServerCblk;
     ComposerService();
     friend class Singleton<ComposerService>;
 public:
     static sp<ISurfaceComposer> getComposerService();
     static surface_flinger_cblk_t const volatile * getControlBlock();
 };

 ComposerService::ComposerService()
 : Singleton<ComposerService>() {
     const String16 name("SurfaceFlinger");
     //获取SurfaceFlinger服务，即BpSurfaceComposer对象
     while (getService(name, &mComposerService) != NO_ERROR) {
         usleep(250000);
     }
     //获取共享内存块
     mServerCblkMemory = mComposerService->getCblk();
     //获取共享内存块基地址
     mServerCblk = static_cast<surface_flinger_cblk_t volatile *>(
             mServerCblkMemory->getBase());
 }

      由此可见，ComposerService主要是获取SurfaceFlinger服务、获取在SurfaceFlinger::readyToRun中创建的共享内存块及其基地址。在Client中，谁要想与SurfaceFlinger通信，需要通过接口getComposerService来获取此BpSurfaceComposer。

     此ComposerService是在调用ComposerService::getInstance时进行有且只有一个的实例化，因为前面讲过，它是一个单实例。

2.1.2.2 Composer

      它也是一个单实例，管理并发送每个layer的ComposerState。其定义如下：

 struct ComposerState {
     sp<ISurfaceComposerClient> client;
     layer_state_t state;
     status_t    write(Parcel& output) const;
     status_t    read(const Parcel& input);
 };

 class Composer : public Singleton<Composer>
 {
     friend class Singleton<Composer>;

     mutable Mutex               mLock;
     //SurfaceComposerClient+SurfaceID与一个ComposerState一一对应
     SortedVector<ComposerState> mStates;
     int                         mOrientation;//整个屏幕的方向
     Composer() : Singleton<Composer>(),
         mOrientation(ISurfaceComposer::eOrientationUnchanged) { }
     //通过BpSurfaceComposer把mStates发送给SurfaceFlinger处理
     void closeGlobalTransactionImpl();

     //根据client和id从mStates中获取对应原ComposerState,从而获取对应的layer_state_t
     layer_state_t* getLayerStateLocked(
             const sp<SurfaceComposerClient>& client, SurfaceID id);

 public:
     //设置与client和id对应的layer_state_t中的位置信息,并保存在mStates中
     status_t setPosition(const sp<SurfaceComposerClient>& client, SurfaceID id,
             float x, float y);
     //设置与client和id对应的layer_state_t中的Size信息,并保存在mStates中
     status_t setSize(const sp<SurfaceComposerClient>& client, SurfaceID id,
             uint32_t w, uint32_t h);
     //设置与client和id对应的layer_state_t中的z-order信息,并保存在mStates中
     status_t setLayer(const sp<SurfaceComposerClient>& client, SurfaceID id,
             int32_t z);
     //设置与client和id对应的layer_state_t中的flags信息,并保存在mStates中
     status_t setFlags(const sp<SurfaceComposerClient>& client, SurfaceID id,
             uint32_t flags, uint32_t mask);
     //设置与client和id对应的layer_state_t中的透明区域信息,并保存在mStates中
     status_t setTransparentRegionHint(
             const sp<SurfaceComposerClient>& client, SurfaceID id,
             const Region& transparentRegion);
     //设置与client和id对应的layer_state_t中的alpha信息,并保存在mStates中
     status_t setAlpha(const sp<SurfaceComposerClient>& client, SurfaceID id,
             float alpha);
     //设置与client和id对应的layer_state_t中的矩阵信息,并保存在mStates中
     status_t setMatrix(const sp<SurfaceComposerClient>& client, SurfaceID id,
             float dsdx, float dtdx, float dsdy, float dtdy);
     //设置与client和id对应的layer_state_t中的位置信息,并保存在mStates中
     status_t setFreezeTint(
             const sp<SurfaceComposerClient>& client, SurfaceID id,
             uint32_t tint);
     //设置整个屏幕的方向
     status_t setOrientation(int orientation);
     //通过BpSurfaceComposer把mStates发送给SurfaceFlinger处理
     static void closeGlobalTransaction() {
         Composer::getInstance().closeGlobalTransactionImpl();
     }
 }

      把上面的comments看完就明白了，Composer管理每个SurfaceComposerClient中的每一个Surface的状态，并记录在ComposerState的layer_state_t中，然后调用者可以调用其closeGlobalTransaction方法把这些mStates发送给SurfaceFlinger处理（处理函数为：SurfaceFlinger::setTransactionState）。

      谁来调用它的方法设置层的属性及发送mStates呢? -----答案是由SurfaceComposerClient来调用。

  2.1.2.3  SurfaceComposerClient

       前面介绍的两个类一个用于获取SurfaceFlinger服务；一个用于记录每个Layer的状态，且可按要求把这些CoposerState发送给SurfaceFlinger。这个类是不是来使用前面两个类提供的服务呢? --答案是肯定的。其定义及详细注释如下：

 #define NUM_DISPLAY_MAX 4  //最多支持四个显示屏
 struct display_cblk_t //每个显示屏的配置参数
 {
     uint16_t    w;
     uint16_t    h;
     uint8_t     format;
     uint8_t     orientation;
     uint8_t     reserved[2];
     float       fps;
     float       density;
     float       xdpi;
     float       ydpi;
     uint32_t    pad[2];
 };
 //在SurfaceFlinger::readyToRun中创建的共享控制块
 struct surface_flinger_cblk_t   // 4KB max,管理系统中所有的显示屏
 {
     uint8_t         connected; //每一个bit表示一个显示屏
     uint8_t         reserved[3];
     uint32_t        pad[7];
     display_cblk_t  displays[NUM_DISPLAY_MAX];
 };

 class SurfaceComposerClient : public RefBase
 {
     friend class Composer;
 public:
     //获取Composer实例，并保存在mComposer中
                 SurfaceComposerClient();
     virtual     ~SurfaceComposerClient();

     //通过BpSurfaceComposerClient<mClient>创建Surface，
     //同时通过ISurfaceComposerClient::surface_data_t返回SurfaceID.然后创建一个SurfaceControl
     //并把返回的BpSurface和当前的SurfaceComposerClient保存在SurfaceControl中,
     //然后返回此SurfaceControl
     sp<SurfaceControl> createSurface(
             const String8& name,// name of the surface
             DisplayID display,  // Display to create this surface on
             uint32_t w,         // width in pixel
             uint32_t h,         // height in pixel
             PixelFormat format, // pixel-format desired
             uint32_t flags = 0  // usage flags
     );


     // Composer parameters <合成参数>
     //所有的合成参数必须在一个transaction中被修改，多个surface可在一个transaction中被更新，
     //所有的变化在关闭transaction时被一次性提交(通过调用closeGlobalTransaction来提交所有变化)。

     //什么都没有做
     static void openGlobalTransaction();

     //通过调用Composer::closeGlobalTransaction(),
     // 把Composer中记录的ComposerState(即mStates)发送给SurfaceFlinger
     static void closeGlobalTransaction();

     //什么都没做
     static status_t freezeDisplay(DisplayID dpy, uint32_t flags = 0);

     //什么都没做
     static status_t unfreezeDisplay(DisplayID dpy, uint32_t flags = 0);

     //把新的显示方向保存在Composer实例中
     static int setOrientation(DisplayID dpy, int orientation, uint32_t flags);

     //从surface_flinger_cblk_t.connected中获取显示屏个数
     static ssize_t getNumberOfDisplays();

     //获取显示屏的信息
     static status_t getDisplayInfo(DisplayID dpy, DisplayInfo* info);
     static ssize_t getDisplayWidth(DisplayID dpy);
     static ssize_t getDisplayHeight(DisplayID dpy);
     static ssize_t getDisplayOrientation(DisplayID dpy);

     //通过注册，当Binder异常退出时，可以获得通知
     status_t linkToComposerDeath(const sp<IBinder::DeathRecipient>& recipient,
             void* cookie = NULL, uint32_t flags = 0);

     //Start####: 以下函数都是把相应的修改状态记录在Composer的mStates中
     //调用Composer::setFlags来设置对应(client+id)的layer状态〈即ComposerState中的layer_state_t〉
     status_t    hide(SurfaceID id);
     status_t    show(SurfaceID id, int32_t layer = -1);
     status_t    freeze(SurfaceID id);
     status_t    unfreeze(SurfaceID id);
     status_t    setFlags(SurfaceID id, uint32_t flags, uint32_t mask);
     //调用Composer::setTransparentRegionHint
     status_t    setTransparentRegionHint(SurfaceID id, const Region& transparent);
     //调用Composer::setLayer
     status_t    setLayer(SurfaceID id, int32_t layer);
     //调用Composer::setAlpha
     status_t    setAlpha(SurfaceID id, float alpha=1.0f);
     //调用Composer::setFreezeTint
     status_t    setFreezeTint(SurfaceID id, uint32_t tint);
     //调用Composer::setMatrix
     status_t    setMatrix(SurfaceID id, float dsdx, float dtdx, float dsdy, float dtdy);
     //调用Composer::setPosition
     status_t    setPosition(SurfaceID id, float x, float y);
     //调用Composer::setSize
     status_t    setSize(SurfaceID id, uint32_t w, uint32_t h);
     //End####:
     status_t    destroySurface(SurfaceID sid);//通过BpSurfaceComposerClient销毁Surface

 private:
     //通过BpSurfaceComposer从SurfaceFlinger获取BpSurfaceComposerClient,
     //并把它保存在mClient中
     virtual void onFirstRef();
     Composer& getComposer();

     mutable     Mutex                       mLock;
                 status_t                    mStatus;
                 //实质为BpSurfaceComposerClient,与SurfaceFlinger.cpp中的Client相对应
                   sp<ISurfaceComposerClient>  mClient;
                 //Composer实例
                   Composer&                   mComposer;
 }

    其功能列表如下：

     1）获取BpSurfaceComposerClient（即mClient），在onFirstRef中实现

     2）通过BpSurfaceComposerClient（即mClient）创建和销毁Surface

     3）通过Composer来记录Surface和显示屏状态变化，及在Composer中通过BpSurfaceComposer把状态变化发给SurfaceFlinger处理

     至此，SurfaceComposerClient功能已经分析清楚。可是从这三个类中，我们已经看到三个Bp（BpSurfaceComposer，BpSurfaceComposerClient和BpSurface）及三个对应的接口。下面总结一下，每个接口的功能，在客户端由谁使用，在服务器端谁来实现。

 2.1.2.4 Surface相关接口总结

     

 2.2 Surface操作

其相关接口如下：

 static JNINativeMethod gSurfaceMethods[] = {
     {"nativeClassInit",     "()V",  (void*)nativeClassInit },
     {"init",                "(Landroid/view/SurfaceSession;ILjava/lang/String;IIIII)V",  (void*)Surface_init },
     {"init",                "(Landroid/os/Parcel;)V",  (void*)Surface_initParcel },
     {"initFromSurfaceTexture", "(Landroid/graphics/SurfaceTexture;)V", (void*)Surface_initFromSurfaceTexture },
     {"getIdentity",         "()I",  (void*)Surface_getIdentity },
     {"destroy",             "()V",  (void*)Surface_destroy },
     {"release",             "()V",  (void*)Surface_release },
     {"copyFrom",            "(Landroid/view/Surface;)V",  (void*)Surface_copyFrom },
     {"isValid",             "()Z",  (void*)Surface_isValid },
     {"lockCanvasNative",    "(Landroid/graphics/Rect;)Landroid/graphics/Canvas;",  (void*)Surface_lockCanvas },
     {"unlockCanvasAndPost", "(Landroid/graphics/Canvas;)V", (void*)Surface_unlockCanvasAndPost },
     {"unlockCanvas",        "(Landroid/graphics/Canvas;)V", (void*)Surface_unlockCanvas },
     {"openTransaction",     "()V",  (void*)Surface_openTransaction },
     {"closeTransaction",    "()V",  (void*)Surface_closeTransaction },
     {"setOrientation",      "(III)V", (void*)Surface_setOrientation },
     {"freezeDisplay",       "(I)V", (void*)Surface_freezeDisplay },
     {"unfreezeDisplay",     "(I)V", (void*)Surface_unfreezeDisplay },
     {"screenshot",          "(II)Landroid/graphics/Bitmap;", (void*)Surface_screenshotAll },
     {"screenshot",          "(IIII)Landroid/graphics/Bitmap;", (void*)Surface_screenshot },
     {"setLayer",            "(I)V", (void*)Surface_setLayer },
     {"setPosition",         "(FF)V",(void*)Surface_setPosition },
     {"setSize",             "(II)V",(void*)Surface_setSize },
     {"hide",                "()V",  (void*)Surface_hide },
     {"show",                "()V",  (void*)Surface_show },
     {"freeze",              "()V",  (void*)Surface_freeze },
     {"unfreeze",            "()V",  (void*)Surface_unfreeze },
     {"setFlags",            "(II)V",(void*)Surface_setFlags },
     {"setTransparentRegionHint","(Landroid/graphics/Region;)V", (void*)Surface_setTransparentRegion },
     {"setAlpha",            "(F)V", (void*)Surface_setAlpha },
     {"setMatrix",           "(FFFF)V",  (void*)Surface_setMatrix },
     {"setFreezeTint",       "(I)V",  (void*)Surface_setFreezeTint },
     {"readFromParcel",      "(Landroid/os/Parcel;)V", (void*)Surface_readFromParcel },
     {"writeToParcel",       "(Landroid/os/Parcel;I)V", (void*)Surface_writeToParcel },
 };

2.2.1 Surface_init调用流程

     在SurfaceFlinger端创建BSurface,在客户端返回SurfaceControl,同时在SurfaceControl中拥有了BpSurface用于与BSurface交互。

 

2.2.1.1 调用流程分析

     BpSurfaceComposerClient->createSurface返回BpSurface。且通过参数返回ISurfaceComposerClient::surface_data_t，其定义如下：

     其中token在SurfaceComposerClient的函数参数中，对应于SurfaceID。即在客户端，它就是SurfaceID。

     token: 加入到Client::mLayers中的序号，在Client中单调递增,初始值为:1，一个Layer创建一个BSurface

     identity: LayerBaseClient中的mIdentity,在所有的Layer中单调递增,初始值为:1

 struct surface_data_t {
     int32_t             token;  //加入到Client::mLayers中的序号，在Client中单调递增,初始值为:1
     int32_t             identity; //LayerBaseClient中的mIdentity,在所有的Layer中单调递增,初始值为:1
     status_t readFromParcel(const Parcel& parcel);
     status_t writeToParcel(Parcel* parcel) const;
 };

2.2.1.2 创建真正的Surface

      在Layer::createSurface中创建真正的BSurface,在SurfaceFlinger::createSurface中调用layer->getSurface时创建的。此BSurface定义如下：

 sp<ISurface> Layer::createSurface()
 {
     class BSurface : public BnSurface, public LayerCleaner {
         wp<const Layer> mOwner;
         virtual sp<ISurfaceTexture> getSurfaceTexture() const { //实现了ISurface的接口
             sp<ISurfaceTexture> res;
             sp<const Layer> that( mOwner.promote() );
             if (that != NULL) {
                 res = that->mSurfaceTexture;
             }
             return res;
         }
     public:
         BSurface(const sp<SurfaceFlinger>& flinger,
                 const sp<Layer>& layer)
             : LayerCleaner(flinger, layer), mOwner(layer) { }
     };
     sp<ISurface> sur(new BSurface(mFlinger, this));
     return sur;
 }

         在此BSurface中实现了ISurface的接口getSurfaceTexture，在此接口中返回Layer::mSurfaceTexture(类型为:SurfaceTextureLayer,它才是真正操作内存的东东),此成员在Layer::onFirstRef中创建，SurfaceTextureLayer是SurfaceTexture的派生类，代码如下：

 void Layer::onFirstRef()
 {
     LayerBaseClient::onFirstRef();

     struct FrameQueuedListener : public SurfaceTexture::FrameAvailableListener {
         FrameQueuedListener(Layer* layer) : mLayer(layer) { }
     private:
         wp<Layer> mLayer;
         virtual void onFrameAvailable() {
             sp<Layer> that(mLayer.promote());
             if (that != 0) {
                 that->onFrameQueued();
             }
         }
     };
     mSurfaceTexture = new SurfaceTextureLayer(mTextureName, this); //创建Layer中的mSurfaceTexture
     mSurfaceTexture->setFrameAvailableListener(new FrameQueuedListener(this));
     mSurfaceTexture->setSynchronousMode(true);
     mSurfaceTexture->setBufferCountServer(2);
 }

2.2.1.3 不得不说的SurfaceControl

       本来Surface_init调用SurfaceComposerClient::createSurface创建一个Surface,可却返回了一个SurfaceControl,下面看看SurfaceCotrol到底做了些什么，以及如何做的？

        相关数据结构如下图所示：

 

 SurfaceControl定义如下：

 class SurfaceControl : public RefBase
 {
 public:
     // release surface data from java
     void        clear();

     //调用SurfaceComposerClient中对应方法，把对应信息保存在
     //Composer的ComposerState中
     status_t    setLayer(int32_t layer);
     status_t    setPosition(int32_t x, int32_t y);
     status_t    setSize(uint32_t w, uint32_t h);
     status_t    hide();
     status_t    show(int32_t layer = -1);
     status_t    freeze();
     status_t    unfreeze();
     status_t    setFlags(uint32_t flags, uint32_t mask);
     status_t    setTransparentRegionHint(const Region& transparent);
     status_t    setAlpha(float alpha=1.0f);
     status_t    setMatrix(float dsdx, float dtdx, float dsdy, float dtdy);
     status_t    setFreezeTint(uint32_t tint);

     //把SurfaceControl中的mSurface和mIdentity写入parcel
     static status_t writeSurfaceToParcel(
             const sp<SurfaceControl>& control, Parcel* parcel);

     //以SurfaceControl为参数创建一个Surface返回，此Surface派生关系如下：
     //class Surface : public SurfaceTextureClient
     //class SurfaceTextureClient: public ANativeWindow, RefBase
     //struct ANativeWindow
     sp<Surface> getSurface() const;

 private:
     SurfaceControl(
             const sp<SurfaceComposerClient>& client,
             const sp<ISurface>& surface,
             const ISurfaceComposerClient::surface_data_t& data);

     ~SurfaceControl();

     void destroy();

     sp<SurfaceComposerClient>   mClient;
     sp<ISurface>                mSurface;
     SurfaceID                   mToken;  //对应SurfaceID,在Client中单调递增
     uint32_t                    mIdentity;  //Layer在系统中唯一的序列号,在系统中单调递增
     mutable Mutex               mLock;

     mutable sp<Surface>         mSurfaceData;
 }

       从其定义中可以看出，在getSurface中将有新花样，其它操作函数都是直接以mToken作为SurfaceID,直接调用SurfaceComposerClient中对应方法。 经过这样一分析，SurfaceControl也没什么神秘的了。但它的getSurface到有点神秘

 2.2.2 getSurface流程

      getSurface在客户端返回Surface(派生于SurfaceTextureClient)，并在Surface的mSurfaceTexture域中保存了BpSurfaceTexture。

      前面Surface初始化之后，就可以getSurface了。getSurface流程如下图所示：

           有了Surface，且在Surface中又有了BpSurfaceTexture，下一步就操作GraphicBuffer了。

 

 3. 画图流程

对于画图流程，可以从ViewRootImpl（ViewRootImpl.java）的draw函数看起，在画图之间，它要调用java层的surface.lockCanvas，画完图之后调用surface.unlockCanvasAndPost来提交显示。

surface.lockCanvas->

lockCanvasNative(Java)->

(C++)Surface_lockCanvas<android_view_Surface.cpp>

surface.unlockCanvasAndPost(Java)->

(C++)Surface_unlockCanvasAndPost<android_view_Surface.cpp>

本章主要分析这两个函数到底做了些什么>

 3.1 Surface_lockCanvas

       Android图形系统中一个重要的概念是surface。View及其子类（如TextView, Button）要画在surface上。每个surface创建一个Canvas对象（但属性时常改变），用来管理view在surface上的绘图操作，如画点画线。每个canvas对象对应一个bitmap，存储画在surface上的内容。

3.1.1 相关数据结构定义

 3.1.1.1 ANativeWindow_Buffer

 typedef struct ANativeWindow_Buffer {
     // The number of pixels that are show horizontally.
     int32_t width;

     // The number of pixels that are shown vertically.
     int32_t height;

     // The number of *pixels* that a line in the buffer takes in
     // memory.  This may be >= width.
     int32_t stride;

     // The format of the buffer.  One of WINDOW_FORMAT_*
     int32_t format;

     // The actual bits.
     void* bits;  //显示内存基地址，通过服务器端fd通过flat_binder_object传给客户端， 然后客户端通过mmap获取。

     // Do not touch.
     uint32_t reserved[6];
 } ANativeWindow_Buffer;

3.1.1.2 SurfaceInfo

 struct SurfaceInfo {
     uint32_t    w;
     uint32_t    h;
     uint32_t    s;
     uint32_t    usage;
     PixelFormat format;
     void*       bits;//显示内存基地址，通过服务器端fd通过flat_binder_object传给客户端， 然后客户端通过mmap获取。
     uint32_t    reserved[2];
 };

3.1.1.3 二者对应关系

 SurfaceInfo* other;
 ANativeWindow_Buffer outBuffer;
 other->w = uint32_t(outBuffer.width);
 other->h = uint32_t(outBuffer.height);
 other->s = uint32_t(outBuffer.stride);
 other->usage = GRALLOC_USAGE_SW_READ_OFTEN | GRALLOC_USAGE_SW_WRITE_OFTEN;
 other->format = uint32_t(outBuffer.format);
 other->bits = outBuffer.bits;

 3.1.1.4 GraphicBuffer

     在分析下面的流程时， 不得不对GraphicBuffer进行深入了解，特别是其Flattenable interface，这是实现画图buffer的关键。其相关定义如下：

 typedef struct native_handle
 {
     int version;        /* sizeof(native_handle_t) */
     int numFds;         /* number of file-descriptors at &data[0] */
     int numInts;        /* number of ints at &data[numFds] */
     int data[0];        /* numFds + numInts ints */
 } native_handle_t;

 typedef const native_handle_t* buffer_handle_t;

 class GraphicBuffer
     : public EGLNativeBase<
         ANativeWindowBuffer,
         GraphicBuffer,
         LightRefBase<GraphicBuffer> >, public Flattenable
 {
     ...
     // Flattenable interface
     size_t getFlattenedSize() const;
     size_t getFdCount() const;
     status_t flatten(void* buffer, size_t size,
             int fds[], size_t count) const;
     status_t unflatten(void const* buffer, size_t size,
             int fds[], size_t count);
     ...
     buffer_handle_t handle; //定义于基类ANativeWindowBuffer中
 };

3.1.1.5 Flattenable interface

下面看看每个Flattenable interface是如何实现的：

3.1.1.5.1 getFlattenedSize

 size_t GraphicBuffer::getFlattenedSize() const {
     return (8 + (handle ? handle->numInts : 0))*sizeof(int);
 }

3.1.1.5.2 getFdCount

 size_t GraphicBuffer::getFdCount() const {
     return handle ? handle->numFds : 0;
 }

3.1.1.5.3 flatten

 status_t GraphicBuffer::flatten(void* buffer, size_t size,
         int fds[], size_t count) const
 {
     size_t sizeNeeded = GraphicBuffer::getFlattenedSize();
     if (size < sizeNeeded) return NO_MEMORY;

     size_t fdCountNeeded = GraphicBuffer::getFdCount();
     if (count < fdCountNeeded) return NO_MEMORY;

     int* buf = static_cast<int*>(buffer);
     buf[0] = 'GBFR';
     buf[1] = width;
     buf[2] = height;
     buf[3] = stride;
     buf[4] = format;
     buf[5] = usage;
     buf[6] = 0;
     buf[7] = 0;

     if (handle) {
         buf[6] = handle->numFds;
         buf[7] = handle->numInts;
         native_handle_t const* const h = handle;
         memcpy(fds,     h->data,             h->numFds*sizeof(int));
         memcpy(&buf[8], h->data + h->numFds, h->numInts*sizeof(int));
     }

     return NO_ERROR;
 }

     把handle中的numFds拷贝到fds中，把handle中的numInts拷贝到buffer中。

 3.1.1.5.4 unflatten 

 status_t GraphicBuffer::unflatten(void const* buffer, size_t size,
         int fds[], size_t count)
 {
     if (size < 8*sizeof(int)) return NO_MEMORY;

     int const* buf = static_cast<int const*>(buffer);
     if (buf[0] != 'GBFR') return BAD_TYPE;

     const size_t numFds  = buf[6];
     const size_t numInts = buf[7];

     const size_t sizeNeeded = (8 + numInts) * sizeof(int);
     if (size < sizeNeeded) return NO_MEMORY;

     size_t fdCountNeeded = 0;
     if (count < fdCountNeeded) return NO_MEMORY;

     if (handle) {
         // free previous handle if any
         free_handle();
     }

     if (numFds || numInts) {
         width  = buf[1];
         height = buf[2];
         stride = buf[3];
         format = buf[4];
         usage  = buf[5];
         native_handle* h = native_handle_create(numFds, numInts);
         memcpy(h->data,          fds,     numFds*sizeof(int));
         memcpy(h->data + numFds, &buf[8], numInts*sizeof(int));
         handle = h;
     } else {
         width = height = stride = format = usage = 0;
         handle = NULL;
     }

     mOwner = ownHandle;

     if (handle != 0) {
         mBufferMapper.registerBuffer(handle);
     }

     return NO_ERROR;
 }

       把width,height,stride,format和usage保存到成员变量中，并创建一个native_handle，然后把numFds和numInts拷贝到handle的data中。同时把此handle注册到mBufferMapper中，mBufferMapper的注册函数实现代码如下：

 status_t GraphicBufferMapper::registerBuffer(buffer_handle_t handle)
 {
     status_t err;
     //gralloc_module_t const *mAllocMod;是一个硬件抽象层实现。通过hw_get_module(GRALLOC_HARDWARE_MODULE_ID, &module)方式获取
     err = mAllocMod->registerBuffer(mAllocMod, handle);

     LOGW_IF(err, "registerBuffer(%p) failed %d (%s)",
             handle, err, strerror(-err));
     return err;
 }

 GraphicBufferMapper::GraphicBufferMapper()
     : mAllocMod(0)
 {
     hw_module_t const* module;
     int err = hw_get_module(GRALLOC_HARDWARE_MODULE_ID, &module);
     LOGE_IF(err, "FATAL: can't find the %s module", GRALLOC_HARDWARE_MODULE_ID);
     if (err == 0) {
         mAllocMod = (gralloc_module_t const *)module;
     }
 }

 3.1.1.5.4  GRALLOC_HARDWARE_MODULE_ID实例

      对于GRALLOC_HARDWARE_MODULE_ID，以hardware/msm7k/libgralloc/gralloc.cpp为例进行分析。其registerBuffer实现函数：gralloc_register_buffer（hardware/msm7k/libgralloc/mapper.cpp)，其相关代码如下：

 int gralloc_register_buffer(gralloc_module_t const* module,
         buffer_handle_t handle)
 {
     if (private_handle_t::validate(handle) < 0)
         return -EINVAL;

     // if this handle was created in this process, then we keep it as is.
     int err = 0;
     private_handle_t* hnd = (private_handle_t*)handle;
     if (hnd->pid != getpid()) {
         hnd->base = NULL;
         if (!(hnd->flags & private_handle_t::PRIV_FLAGS_USES_GPU)) {
             void *vaddr;
             err = gralloc_map(module, handle, &vaddr);
         }
     }
     return err;
 }

 static int gralloc_map(gralloc_module_t const* module,
         buffer_handle_t handle,
         void** vaddr)
 {
     private_handle_t* hnd = (private_handle_t*)handle;
     if (!(hnd->flags & private_handle_t::PRIV_FLAGS_FRAMEBUFFER)) {
         size_t size = hnd->size;
 #if PMEM_HACK
         size += hnd->offset;
 #endif
         void* mappedAddress = mmap(0, size,
                 PROT_READ|PROT_WRITE, MAP_SHARED, hnd->fd, 0);
         if (mappedAddress == MAP_FAILED) {
             LOGE("Could not mmap handle %p, fd=%d (%s)",
                     handle, hnd->fd, strerror(errno));
             hnd->base = 0;
             return -errno;
         }
         hnd->base = intptr_t(mappedAddress) + hnd->offset;
         //LOGD("gralloc_map() succeeded fd=%d, off=%d, size=%d, vaddr=%p",
         //        hnd->fd, hnd->offset, hnd->size, mappedAddress);
     }
     *vaddr = (void*)hnd->base;
     return 0;
 }

    从gralloc_map可以看出，这个registerBuffer主要做了一件事：

    1）根据handle中传过来的fd和size进行mmap映射（把kernel中的内存映射到用户空间），映射之后的地址再加上hnd->offset便获得hnd->base供后面使用。

      从这里可以初步看出，这个图形buffer数据并不是真正的从client传递到server，而是在lock是从server把fd传递给client，由客户端进行mmap，然后进行使用。关于这个是怎么实现的，后面将详细分析其实现过程。

    对于如何从native_handle转换为private_handle_t，且在private_handle_t中可以获取fd和offset? 看一下其数据结构和flatten的实现方式就可以得知：
native_handle：

 typedef struct native_handle
 {
     int version;        /* sizeof(native_handle_t) */
     int numFds;         /* number of file-descriptors at &data[0] */
     int numInts;        /* number of ints at &data[numFds] */
     int data[0];        /* numFds + numInts ints */
 } native_handle_t;

     这个data[0]是关键，虽然分配了哪么多buffer,但实质上native_handle只占了3个int.其它的数据由包含它的数据结构来解析。

private_handle_t：

 struct private_handle_t {
     native_handle_t nativeHandle;
 #endif

     enum {
         PRIV_FLAGS_FRAMEBUFFER = 0x00000001,
         PRIV_FLAGS_USES_PMEM   = 0x00000002,
         PRIV_FLAGS_USES_GPU    = 0x00000004,
     };

     // file-descriptors
     int     fd;
     // ints
     int     magic;
     int     flags;
     int     size;
     int     offset;
     int     gpu_fd; // stored as an int, b/c we don't want it marshalled

     // FIXME: the attributes below should be out-of-line
     int     base;
     int     map_offset;
     int     pid;

 #ifdef __cplusplus
     static const int sNumInts = 8;  //numInts在这儿明确指定
     static const int sNumFds = 1;   //numFds在这儿明确指定
     static const int sMagic = 'gmsm';

     private_handle_t(int fd, int size, int flags) :
         fd(fd), magic(sMagic), flags(flags), size(size), offset(0),
         base(0), pid(getpid())
     {
         version = sizeof(native_handle);
         numInts = sNumInts;
         numFds = sNumFds;
     }
     ~private_handle_t() {
         magic = 0;
     }

     static int validate(const native_handle* h) {
         const private_handle_t* hnd = (const private_handle_t*)h;
         if (!h || h->version != sizeof(native_handle) ||
                 h->numInts != sNumInts || h->numFds != sNumFds ||
                 hnd->magic != sMagic)
         {
             LOGE("invalid gralloc handle (at %p)", h);
             return -EINVAL;
         }
         return 0;
     }
 #endif
 }

3.1.2 Surface_lockCanvas执行流程

查看高清大图

3.1.3 Surface_lockCanvas总结 

功能：Surface_lockCanvas获取显示buffer在本进程用户空间的地址，并据此创建一个SkBitmap给Java使用。
关键技术：BINDER_TYPE_FD类型的Binder、mmap、gralloc硬件抽象层

 

3.1.4 SurfaceTexture::dequeueBuffer如何创建GraphicBuffer

       相关代码如下：

 const sp<GraphicBuffer>& buffer(mSlots[buf].mGraphicBuffer);
 if ((buffer == NULL) ||
     (uint32_t(buffer->width)  != w) ||
     (uint32_t(buffer->height) != h) ||
     (uint32_t(buffer->format) != format) ||
     ((uint32_t(buffer->usage) & usage) != usage))
 {
     usage |= GraphicBuffer::USAGE_HW_TEXTURE;
     status_t error;
     sp<GraphicBuffer> graphicBuffer( //创建GraphicBuffer
             mGraphicBufferAlloc->createGraphicBuffer(
                     w, h, format, usage, &error));
     if (graphicBuffer == 0) {
         ST_LOGE("dequeueBuffer: SurfaceComposer::createGraphicBuffer "
                 "failed");
         return error;
     }
     if (updateFormat) {
         mPixelFormat = format;
     }
     mSlots[buf].mGraphicBuffer = graphicBuffer;
     mSlots[buf].mRequestBufferCalled = false;
     if (mSlots[buf].mEglImage != EGL_NO_IMAGE_KHR) {
         eglDestroyImageKHR(mSlots[buf].mEglDisplay, mSlots[buf].mEglImage);
         mSlots[buf].mEglImage = EGL_NO_IMAGE_KHR;
         mSlots[buf].mEglDisplay = EGL_NO_DISPLAY;
     }
     returnFlags |= ISurfaceTexture::BUFFER_NEEDS_REALLOCATION;
 }

     mGraphicBufferAlloc也是通过调用BpSurfaceComposer->createGraphicBufferAlloc而获取，它对应的服务器为SufaceFlinger中的GraphicBufferAlloc。

     mGraphicBufferAlloc实质为一个BpGraphicBufferAlloc,它真正创建GraphicBuffer的代码位于GraphicBufferAlloc::createGraphicBuffer中。代码关键调用流程如下：

     new GraphicBuffer(w, h, format, usage)->

       initSize(w, h, reqFormat, reqUsage)->

         GraphicBufferAllocator::get()->

            allocator.alloc(w, h, format, reqUsage, &handle, &stride)->

            返回handle,此handle为ANativeWindowBuffer成员，类型为native_handle。

              GraphicBufferAllocator::alloc->

                mAllocDev->alloc->

                 mAllocDev类型为alloc_device_t，它通过gralloc_open向

                 GRALLOC_HARDWARE_MODULE_ID获取，根据上面的实例msm7k，

                 它最终执行gralloc_device_open而获取gralloc_context_t.device.common，

                  alloc的实现函数为gralloc_alloc.

                   gralloc_alloc->

                    gralloc_alloc_buffer->

                     1）获取GPU内存(调用SimpleBestFitAllocator::allocate进行分配)

                     2）fd = open("/dev/null", O_RDONLY)获取fd

                     3）根据fd、size和flags创建private_handle_t，其相关代码如下：

 

 private_handle_t* hnd = new private_handle_t(fd, size, flags);
 if (base == NULL) {...
     }
 } else {
     private_module_t* m = reinterpret_cast<private_module_t*>(
             dev->common.module);
     hnd->offset = offset;
     hnd->base = int(base)+offset;
     hnd->gpu_fd = gpu_fd;
     hnd->map_offset = m->fb_map_offset;
     *pHandle = hnd;
 }