sheldon_blogs

Android 显示系统:SurfaceFlinger详解

一、Android系统启动

  Android设备从按下开机键到桌面显示画面,大致过程如下图流程:

  开机显示桌面、从桌面点击 App 图标到 Activity显示在屏幕上的过程又是怎样的呢?下面介绍Android系统中的“画家” - SurfaceFlinger.
  SurfaceFlinger 启动过程:

 

 二、SurfaceFlinger代码剖析[Android 11]

  代码路径:/frameworks/native/services/surfaceflinger/
  SurfaceFlinger二进制分成surfaceflinger可执行文件(main入口)和libsurfaceflinger.so库文件(功能实现),由main_surfaceflinger.cpp文件编译而成,Android.bp代码模块编译配置如下:

 1.【执行文件-surfaceflinger】

...
cc_binary { name:
"surfaceflinger", defaults: ["surfaceflinger_defaults"], init_rc: ["surfaceflinger.rc"], srcs: ["main_surfaceflinger.cpp"], whole_static_libs: [ "libsigchain", ], shared_libs: [ "android.frameworks.displayservice@1.0", "android.hardware.configstore-utils", "android.hardware.configstore@1.0", "android.hardware.graphics.allocator@2.0", "libbinder", "libcutils", "libdisplayservicehidl", "libhidlbase", "libhidltransport", "liblayers_proto", "liblog", "libsurfaceflinger", "libtimestats_proto", "libutils", ], static_libs: [ "libserviceutils", "libtrace_proto", ], ldflags: ["-Wl,--export-dynamic"], // TODO(b/71715793): These version-scripts are required due to the use of // whole_static_libs to pull in libsigchain. To work, the files had to be // locally duplicated from their original location // $ANDROID_ROOT/art/sigchainlib/ multilib: { lib32: { version_script: "version-script32.txt", }, lib64: { version_script: "version-script64.txt", }, }, }
...

  SurfaceFlinger可执行二进制文件surfaceflinger由main_surfaceflinger.cpp文件独立编译而成,主要负责搭建进程启动环境:

int main(int, char**) {
    signal(SIGPIPE, SIG_IGN);

    // 从8.0开始,Android提供了hidl机制,将原先直接由JNI->Native->HAL的接口调用形式,统一规范成hidl service/client交互形式。
    // 该方式从一方面规范和统一了Android Framework和HAL的调用机制,但实际从项目维度,这种调用方式对性能上开销,将比直接调用的方式要花费更多的时间。
    hardware::configureRpcThreadpool(1 /* maxThreads */,
            false /* callerWillJoin */);

    startGraphicsAllocatorService();

    // When SF is launched in its own process, limit the number of
    // binder threads to 4.
    ProcessState::self()->setThreadPoolMaxThreadCount(4);

    // start the thread pool
    sp<ProcessState> ps(ProcessState::self());
    ps->startThreadPool();

    // 创建SurfaceFlinger对象,由强指针指向。
    // SurfaceFlinger继承RefBase类,所以此处一旦new出对象赋给sp指针后,将立刻出发SurfaceFlinger类的onFirstRef方法的调用。
    // instantiate surfaceflinger
    sp<SurfaceFlinger> flinger = surfaceflinger::createSurfaceFlinger();

    setpriority(PRIO_PROCESS, 0, PRIORITY_URGENT_DISPLAY);

    set_sched_policy(0, SP_FOREGROUND);

    // Put most SurfaceFlinger threads in the system-background cpuset
    // Keeps us from unnecessarily using big cores
    // Do this after the binder thread pool init
    if (cpusets_enabled()) set_cpuset_policy(0, SP_SYSTEM);

    // SurfaceFlinger类正式初始化
    // initialize before clients can connect
    flinger->init();

    // SurfaceFlinger向ServiceManager注册Binder服务,
    // 这样在其他进程中可以通过getService+SERVICE_NAME来获取SurfaceFlinger服务,继而可以和SurfaceFlinger类进行Binder通信。
    // publish surface flinger
    sp<IServiceManager> sm(defaultServiceManager());
    sm->addService(String16(SurfaceFlinger::getServiceName()), flinger, false,
                   IServiceManager::DUMP_FLAG_PRIORITY_CRITICAL | IServiceManager::DUMP_FLAG_PROTO);

    //里面的new DisplayService()方法调用HIDL定义接口 Return<sp<IDisplayEventReceiver >> getEventReceiver() override;
    startDisplayService(); // dependency on SF getting registered above

    if (SurfaceFlinger::setSchedFifo(true) != NO_ERROR) {
        ALOGW("Couldn't set to SCHED_FIFO: %s", strerror(errno));
    }

    // SurfaceFlinger类进入主循环(此处注意SurfaceFlinger类未继承Threads类,不遵循Threads类的接口执行顺序)
    // run surface flinger in this thread
    flinger->run();

    return 0;
}

  HIDL接口介绍可以参考:https://source.android.google.cn/reference/hidl/

 2.【动态库-libsurfaceflinger.so】

Android.bp代码模块编译配置如下:

...
cc_library_shared { name:
"libsurfaceflinger", defaults: ["libsurfaceflinger_defaults"], cflags: [ "-fvisibility=hidden", "-Werror=format", "-DREDUCE_VIDEO_WORKLOAD", "-DUSE_AML_HW_ACTIVE_MODE", ], srcs: [ ":libsurfaceflinger_sources", ], logtags: ["EventLog/EventLogTags.logtags"], include_dirs: [ "frameworks/native/vulkan/vkjson", "frameworks/native/vulkan/include", "hardware/amlogic/gralloc/amlogic", "hardware/amlogic/hwcomposer/tvp", "hardware/amlogic/gralloc", ], static_libs: [ "libomxutils_static@2", "libamgralloc_ext_static@2", ], cppflags: [ "-fwhole-program-vtables", // requires ThinLTO ], lto: { thin: true, }, }
...

  上面提到的createSurfaceFlinger()中会调用new SurfaceFlinger(),然后会执行到:onFirstRef():

void SurfaceFlinger::onFirstRef()
{
    mEventQueue.init(this);
}

  onFirstRef() 中会创建 Handler 并初始化: /frameworks/native/services/surfaceflinger/Scheduler/MessageQueue.cpp

//MessageQueue.cpp
void MessageQueue::init(const sp<SurfaceFlinger>& flinger)
{
    mFlinger = flinger;
    mLooper = new Looper(true);
    mHandler = new Handler(*this);
}

  然后会执行到 SurfaceFlinger::init(),该方法主要功能是:

  • 初始化 EGL
  • 创建 HWComposer
  • 初始化非虚拟显示屏
  • 启动 EventThread 线程
  • 启动开机动画
// Do not call property_set on main thread which will be blocked by init
// Use StartPropertySetThread instead.
void SurfaceFlinger::init() {
    ALOGI(  "SurfaceFlinger's main thread ready to run. "
            "Initializing graphics H/W...");
    Mutex::Autolock _l(mStateLock);

// 对于CompositionEngine 属性进行设置, 创建RenderEngine对象
// Get a RenderEngine for the given display / config (can't fail) // TODO(b/77156734): We need to stop casting and use HAL types when possible. // Sending maxFrameBufferAcquiredBuffers as the cache size is tightly tuned to single-display. mCompositionEngine->setRenderEngine(renderengine::RenderEngine::create( renderengine::RenderEngineCreationArgs::Builder() .setPixelFormat(static_cast<int32_t>(defaultCompositionPixelFormat)) .setImageCacheSize(maxFrameBufferAcquiredBuffers) .setUseColorManagerment(useColorManagement) .setEnableProtectedContext(enable_protected_contents(false)) .setPrecacheToneMapperShaderOnly(false) .setSupportsBackgroundBlur(mSupportsBlur) .setContextPriority(useContextPriority ? renderengine::RenderEngine::ContextPriority::HIGH : renderengine::RenderEngine::ContextPriority::MEDIUM) .build())); mCompositionEngine->setTimeStats(mTimeStats); LOG_ALWAYS_FATAL_IF(mVrFlingerRequestsDisplay, "Starting with vr flinger active is not currently supported.");

  //创建HWComposer对象并传入一个name属性,再通过mCompositionEngine->setHwComposer设置对象属性。 mCompositionEngine
->setHwComposer(getFactory().createHWComposer(getBE().mHwcServiceName)); mCompositionEngine->getHwComposer().setConfiguration(this, getBE().mComposerSequenceId);

  //processDisplayHotplugEventsLocked(); 处理 任何初始热插拔和显示更改的结果
  //在此方法中主要有调用 initScheduler(displayId);
// Process any initial hotplug and resulting display changes. processDisplayHotplugEventsLocked(); const auto display = getDefaultDisplayDeviceLocked(); LOG_ALWAYS_FATAL_IF(!display, "Missing internal display after registering composer callback."); LOG_ALWAYS_FATAL_IF(!getHwComposer().isConnected(*display->getId()), "Internal display is disconnected."); if (useVrFlinger) { auto vrFlingerRequestDisplayCallback = [this](bool requestDisplay) { // This callback is called from the vr flinger dispatch thread. We // need to call signalTransaction(), which requires holding // mStateLock when we're not on the main thread. Acquiring // mStateLock from the vr flinger dispatch thread might trigger a // deadlock in surface flinger (see b/66916578), so post a message // to be handled on the main thread instead. static_cast<void>(schedule([=] { ALOGI("VR request display mode: requestDisplay=%d", requestDisplay); mVrFlingerRequestsDisplay = requestDisplay; signalTransaction(); })); }; mVrFlinger = dvr::VrFlinger::Create(getHwComposer().getComposer(), getHwComposer() .fromPhysicalDisplayId(*display->getId()) .value_or(0), vrFlingerRequestDisplayCallback); if (!mVrFlinger) { ALOGE("Failed to start vrflinger"); } } // initialize our drawing state mDrawingState = mCurrentState; // set initial conditions (e.g. unblank default device) initializeDisplays(); char primeShaderCache[PROPERTY_VALUE_MAX]; property_get("service.sf.prime_shader_cache", primeShaderCache, "1"); if (atoi(primeShaderCache)) { getRenderEngine().primeCache(); } // Inform native graphics APIs whether the present timestamp is supported: const bool presentFenceReliable = !getHwComposer().hasCapability(hal::Capability::PRESENT_FENCE_IS_NOT_RELIABLE); mStartPropertySetThread = getFactory().createStartPropertySetThread(presentFenceReliable); if (mStartPropertySetThread->Start() != NO_ERROR) { ALOGE("Run StartPropertySetThread failed!"); } ALOGV("Done initializing"); }

  首先看下如何创建 HWComposer:frameworks/native/services/surfaceflinger/SurfaceFlingerDefaultFactory.cpp 

//make_unique 相当于 new,(能够取代new 而且无需 delete pointer,有助于代码管理)。
std::unique_ptr<HWComposer> DefaultFactory::createHWComposer(const std::string& serviceName) {
    return std::make_unique<android::impl::HWComposer>(serviceName);
}

  对于CompositionEngine进行初始化:

std::unique_ptr<compositionengine::CompositionEngine> DefaultFactory::createCompositionEngine() {
      return compositionengine::impl::createCompositionEngine();
}

  同样是通过make_unique创建了 CompositionEngine对象:

std::unique_ptr<compositionengine::CompositionEngine> createCompositionEngine() {
     return std::make_unique<CompositionEngine>();
}

  再回到flinger->init()中 processDisplayHotplugEventsLocked(); 处理任何初始热插拔和显示更改的结果,在此方法中主要有调用 initScheduler(displayId):

void SurfaceFlinger::initScheduler(DisplayId primaryDisplayId) {
    if (mScheduler) {
        // In practice it's not allowed to hotplug in/out the primary display once it's been
        // connected during startup, but some tests do it, so just warn and return.
        ALOGW("Can't re-init scheduler");
        return;
    }

    auto currentConfig = HwcConfigIndexType(getHwComposer().getActiveConfigIndex(primaryDisplayId));
    mRefreshRateConfigs =
            std::make_unique<scheduler::RefreshRateConfigs>(getHwComposer().getConfigs(
                                                                    primaryDisplayId),
                                                            currentConfig);
    mRefreshRateStats =
            std::make_unique<scheduler::RefreshRateStats>(*mRefreshRateConfigs, *mTimeStats,
                                                          currentConfig, hal::PowerMode::OFF);
    mRefreshRateStats->setConfigMode(currentConfig);

    mPhaseConfiguration = getFactory().createPhaseConfiguration(*mRefreshRateConfigs);

   // 处创建Scheduler对象
// start the EventThread mScheduler = getFactory().createScheduler([this](bool enabled) { setPrimaryVsyncEnabled(enabled); }, *mRefreshRateConfigs, *this); //创建app链接 和 sf链接
mAppConnectionHandle
= mScheduler->createConnection("app", mPhaseConfiguration->getCurrentOffsets().late.app, impl::EventThread::InterceptVSyncsCallback()); mSfConnectionHandle = mScheduler->createConnection("sf", mPhaseConfiguration->getCurrentOffsets().late.sf, [this](nsecs_t timestamp) { mInterceptor->saveVSyncEvent(timestamp); }); mEventQueue->setEventConnection(mScheduler->getEventConnection(mSfConnectionHandle)); mVSyncModulator.emplace(*mScheduler, mAppConnectionHandle, mSfConnectionHandle, mPhaseConfiguration->getCurrentOffsets()); mRegionSamplingThread = new RegionSamplingThread(*this, *mScheduler, RegionSamplingThread::EnvironmentTimingTunables()); // Dispatch a config change request for the primary display on scheduler // initialization, so that the EventThreads always contain a reference to a // prior configuration. // // This is a bit hacky, but this avoids a back-pointer into the main SF // classes from EventThread, and there should be no run-time binder cost // anyway since there are no connected apps at this point. const nsecs_t vsyncPeriod = mRefreshRateConfigs->getRefreshRateFromConfigId(currentConfig).getVsyncPeriod(); mScheduler->onPrimaryDisplayConfigChanged(mAppConnectionHandle, primaryDisplayId.value, currentConfig, vsyncPeriod); }

  详细看下app、sf的链接:

Scheduler::ConnectionHandle Scheduler::createConnection(
         const char* connectionName, nsecs_t phaseOffsetNs,
         impl::EventThread::InterceptVSyncsCallback interceptCallback) {
     auto vsyncSource = makePrimaryDispSyncSource(connectionName, phaseOffsetNs);   
     auto eventThread = std::make_unique<impl::EventThread>(std::move(vsyncSource), std::move(interceptCallback));  
     return createConnection(std::move(eventThread));
}

  可以看到创建了DispSyncSource对象,且构造方法传入了四个值,dispSync对象,phaseOffset偏移量,traceVsync为true,name就是 app或 sf

  DispSyncSource::DispSyncSource(DispSync* dispSync, nsecs_t phaseOffset, bool traceVsync,
                                 const char* name)
        : mName(name),
          mValue(base::StringPrintf("VSYNC-%s", name), 0), //对mValue进行了赋值,systrace上我们看到的 VSYNC-app VSYNC-sf 标签就是它
          mTraceVsync(traceVsync),  //mTraceVsync为true,在onDispSyncEvent方法中
          mVsyncOnLabel(base::StringPrintf("VsyncOn-%s", name)),
          mDispSync(dispSync),
          mPhaseOffset(base::StringPrintf("VsyncOffset-%s", name), phaseOffset)   //对mPhaseOffset进行初始化 vsync信号到来时候,sf、app的偏移量

  所以我们在systrace上面看到的 VSYNC-app/VSYNC-sf 驼峰 0 1变化,来源于这个。

 

 

   创建EventThread对象,传入sf 或 app 相关联的vsyncSource对象:

auto eventThread = std::make_unique<impl::EventThread>(std::move(vsyncSource),                                                           std::move(interceptCallback));

说明:

1)每个 ConnectionHandle 对象里有个 id,作为 Scheduler 对象中 mConnections 属性(map<id, Connection>)的键值,Connection 对象中又包含 ConnectionHandle、EventThreadConnection、EventThread 3个属性。

2)mScheduler->getEventConnection(mSfConnectionHandle) 中,以 mSfConnectionHandle 的 id 为键值,在 Scheduler 的 mConnections(unordered_map<int64_t, Connection>)中找到对应的Connection,并返回其 EventThreadConnection 成员属性。

3)getHwComposer().registerCallback() 中,依次调用 HwComposer、Device 的 registerCallback() 方法,并在 Device 中 将 SurfaceFlinger 对象封装到 ComposerCallbackBridge 中;对于封装后的对象,依次调用 Composer、IComposerClient 的 registerCallback() 方法,注入到 IComposerClient 的实现类中。

 

相关问题:

  ① 屏幕刷新速率比系统帧速率快:
    此时,在前缓冲区内容全部映射到屏幕上之后,后缓冲区尚未准备好下一帧,屏幕将无法读取下一帧,所以只能继续显示当前一帧的图形,造成一帧显示多次,也就是卡顿。 
  ② 系统帧速率比屏幕刷新率快 
    此时,屏幕未完全把前缓冲区的一帧映射到屏幕,而系统已经在后缓冲区准备好了下一帧,并要求读取下一帧到屏幕,将会导致屏幕上半部分是上一帧的图形,而下半部分是下一帧的图形,造成屏幕上显示多帧,也就是屏幕撕裂。

  为了解决上述问题,Android显示系统一般会有多级缓冲,即在屏幕刷新的同时在另外一个buffer准备下一帧数据,以此提高性能:

  


  前缓冲区:用来显示内容到屏幕的帧缓冲区 
  后缓冲区:用于后台合成下一帧图形的帧缓冲区

  垂直同步(VSync):当屏幕从缓冲区扫描完一帧到屏幕上之后,开始扫描下一帧之前,发出的一个同步信号,该信号用来切换前缓冲区和后缓冲区。

  屏幕刷新率(HZ):代表屏幕在一秒内刷新屏幕的次数,Android手机一般为60HZ(也就是1秒刷新60帧,大约16.67毫秒刷新1帧) 
  系统帧速率(FPS):代表了系统在一秒内合成的帧数,该值的大小由系统算法和硬件决定。

 

3. 服务启动配置文件:/frameworks/native/services/surfaceflinger/surfaceflinger.rc
 上面发现服务配置文件也在Android.mk中被加载:LOCAL_INIT_RC := surfaceflinger.rc

service surfaceflinger /system/bin/surfaceflinger
    class core animation
    user system
    group graphics drmrpc readproc
    onrestart restart zygote
    writepid /dev/stune/foreground/tasks
    socket pdx/system/vr/display/client     stream 0666 system graphics u:object_r:pdx_display_client_endpoint_socket:s0
    socket pdx/system/vr/display/manager    stream 0666 system graphics u:object_r:pdx_display_manager_endpoint_socket:s0
    socket pdx/system/vr/display/vsync      stream 0666 system graphics u:object_r:pdx_display_vsync_endpoint_socket:s0


4. Surface 创建过程

 

 Surface 创建的过程就是 Activity 显示的过程,在 ActivityThread.handleResumeActivity() 中调用了 Activity.makeVisible()具体实现:

void makeVisible() {
    if (!mWindowAdded) {
        ViewManager wm = getWindowManager();//此处 getWindowManager 获取的是 WindowManagerImpl 对象
        wm.addView(mDecor, getWindow().getAttributes());
        mWindowAdded = true;
    }
    mDecor.setVisibility(View.VISIBLE);
}

 WindowManagerImpl.java:

public void addView(@NonNull View view, @NonNull ViewGroup.LayoutParams params) {
    applyDefaultToken(params);
    mGlobal.addView(view, params, mDisplay, mParentWindow);
}

 WindowManagerGlobal.java:

public void addView(View view, ViewGroup.LayoutParams params, Display display, Window parentWindow) {
    ...
    final WindowManager.LayoutParams wparams = (WindowManager.LayoutParams) params;
    //创建 ViewRootImpl
    ViewRootImpl root = new ViewRootImpl(view.getContext(), display);
    view.setLayoutParams(wparams);
    mViews.add(view);
    mRoots.add(root);
    mParams.add(wparams);
    //设置 View
    root.setView(view, wparams, panelParentView);
    ...
}

 创建 ViewRootImpl:

    public ViewRootImpl(Context context, Display display) {
       //获取 IWindowSession的代理类
        this(context, display, WindowManagerGlobal.getWindowSession(),
                false /* useSfChoreographer */);
    }

 WindowManagerGlobal.java:

    @UnsupportedAppUsage
    public static IWindowSession getWindowSession() {
        synchronized (WindowManagerGlobal.class) {
            if (sWindowSession == null) {
                try {
                    // Emulate the legacy behavior.  The global instance of InputMethodManager
                    // was instantiated here.
                    // TODO(b/116157766): Remove this hack after cleaning up @UnsupportedAppUsage
                    //获取 IMS 的代理类
                    InputMethodManager.ensureDefaultInstanceForDefaultDisplayIfNecessary();
                    IWindowManager windowManager = getWindowManagerService();
                    //经过 Binder 调用,最终调用 WMS
                    sWindowSession = windowManager.openSession(
                            new IWindowSessionCallback.Stub() {
                                @Override
                                public void onAnimatorScaleChanged(float scale) {
                                    ValueAnimator.setDurationScale(scale);
                                }
                            });
                } catch (RemoteException e) {
                    throw e.rethrowFromSystemServer();
                }
            }
            return sWindowSession;
        }
    }

 WindowManagerService.openSession:

    // -------------------------------------------------------------
    // IWindowManager API
    // -------------------------------------------------------------

    @Override
    public IWindowSession openSession(IWindowSessionCallback callback) {
        //创建session对象
        return new Session(this, callback);
    }

 再次经过 Binder 将数据写回 app 进程,则获取的便是 Session 的代理对象 IWindowSession。
 创建完 ViewRootImpl 对象后,接下来调用该对象的 setView() 方法:
 ViewRootImpl:

public void setView(View view, WindowManager.LayoutParams attrs, View panelParentView) {
  synchronized (this) {
 
    requestLayout(); //详见下面分析
    ...
    //通过 Binder调用,进入 system 进程的 Session

    res = mWindowSession.addToDisplayAsUser(mWindow, mSeq, mWindowAttributes,
        getHostVisibility(), mDisplay.getDisplayId(), userId, mTmpFrame,
        mAttachInfo.mContentInsets, mAttachInfo.mStableInsets,
        mAttachInfo.mDisplayCutout, inputChannel,
        mTempInsets, mTempControls);

    ...
  }
}

Session.java

    @Override
    public int addToDisplayAsUser(IWindow window, int seq, WindowManager.LayoutParams attrs,
            int viewVisibility, int displayId, int userId, Rect outFrame,
            Rect outContentInsets, Rect outStableInsets,
            DisplayCutout.ParcelableWrapper outDisplayCutout, InputChannel outInputChannel,
            InsetsState outInsetsState, InsetsSourceControl[] outActiveControls) {
        //调用WMS的addWindow方法    
        return mService.addWindow(this, window, seq, attrs, viewVisibility, displayId, outFrame,
                outContentInsets, outStableInsets, outDisplayCutout, outInputChannel,
                outInsetsState, outActiveControls, userId);
    }

WindowManagerService.java:

    public int addWindow(Session session, IWindow client, int seq,
            LayoutParams attrs, int viewVisibility, int displayId, Rect outFrame,
            Rect outContentInsets, Rect outStableInsets,
            DisplayCutout.ParcelableWrapper outDisplayCutout, InputChannel outInputChannel,
            InsetsState outInsetsState, InsetsSourceControl[] outActiveControls,
            int requestUserId) {
        Arrays.fill(outActiveControls, null);
        int[] appOp = new int[1];
        final boolean isRoundedCornerOverlay = (attrs.privateFlags
                & PRIVATE_FLAG_IS_ROUNDED_CORNERS_OVERLAY) != 0;
        int res = mPolicy.checkAddPermission(attrs.type, isRoundedCornerOverlay, attrs.packageName,
                appOp);
        if (res != WindowManagerGlobal.ADD_OKAY) {
            return res;
        }

        WindowState parentWindow = null;
        final int callingUid = Binder.getCallingUid();
        final int callingPid = Binder.getCallingPid();
        final long origId = Binder.clearCallingIdentity();
        final int type = attrs.type;

        synchronized (mGlobalLock) {
            if (!mDisplayReady) {
                throw new IllegalStateException("Display has not been initialialized");
            }
...
            //创建 WindowState
            final WindowState win = new WindowState(this, session, client, token, parentWindow,
                    appOp[0], seq, attrs, viewVisibility, session.mUid, userId,
                    session.mCanAddInternalSystemWindow);
            if (win.mDeathRecipient == null) {
                // Client has apparently died, so there is no reason to
                // continue.
                ProtoLog.w(WM_ERROR, "Adding window client %s"
                        + " that is dead, aborting.", client.asBinder());
                return WindowManagerGlobal.ADD_APP_EXITING;
            }

            if (win.getDisplayContent() == null) {
                ProtoLog.w(WM_ERROR, "Adding window to Display that has been removed.");
                return WindowManagerGlobal.ADD_INVALID_DISPLAY;
            }

            // 调整 WindowManager的LayoutParams 参数
            final DisplayPolicy displayPolicy = displayContent.getDisplayPolicy();
            displayPolicy.adjustWindowParamsLw(win, win.mAttrs, callingPid, callingUid);

            res = displayPolicy.validateAddingWindowLw(attrs, callingPid, callingUid);
            if (res != WindowManagerGlobal.ADD_OKAY) {
                return res;
            }

            // 打开输入通道
            final boolean openInputChannels = (outInputChannel != null
                    && (attrs.inputFeatures & INPUT_FEATURE_NO_INPUT_CHANNEL) == 0);
            if  (openInputChannels) {
                win.openInputChannel(outInputChannel);
            }

...
            displayContent.getInputMonitor().setUpdateInputWindowsNeededLw();

            boolean focusChanged = false;
            //当该窗口能接收按键事件,则更新聚焦窗口
            if (win.canReceiveKeys()) {
                focusChanged = updateFocusedWindowLocked(UPDATE_FOCUS_WILL_ASSIGN_LAYERS,
                        false /*updateInputWindows*/);
                if (focusChanged) {
                    imMayMove = false;
                }
            }

            if (imMayMove) {
                displayContent.computeImeTarget(true /* updateImeTarget */);
            }
...
        }

        Binder.restoreCallingIdentity(origId);

        return res;
    }

 创建 SurfaceSession 对象,并将当前 Session 添加到 WMS.mSessions 成员变量。
 Session.java:

    void windowAddedLocked(String packageName) {
        mPackageName = packageName;
        mRelayoutTag = "relayoutWindow: " + mPackageName;
        if (mSurfaceSession == null) {
            if (DEBUG) {
                Slog.v(TAG_WM, "First window added to " + this + ", creating SurfaceSession");
            }
            mSurfaceSession = new SurfaceSession();
            ProtoLog.i(WM_SHOW_TRANSACTIONS, "  NEW SURFACE SESSION %s", mSurfaceSession);
            mService.mSessions.add(this);
            if (mLastReportedAnimatorScale != mService.getCurrentAnimatorScale()) {
                mService.dispatchNewAnimatorScaleLocked(this);
            }
        }
        mNumWindow++;
    }

 SurfaceSession 的创建会调用 JNI,在 JNI 调用 nativeCreate()。
 android_view_SurfaceSession.cpp:

static jlong nativeCreate(JNIEnv* env, jclass clazz) {
    SurfaceComposerClient* client = new SurfaceComposerClient();
    client->incStrong((void*)nativeCreate);
    return reinterpret_cast<jlong>(client);
}
static jlong nativeCreate(JNIEnv* env, jclass clazz, jobject sessionObj,
        jstring nameStr, jint w, jint h, jint format, jint flags, jlong parentObject,
        jobject metadataParcel) {
    ScopedUtfChars name(env, nameStr);
    sp<SurfaceComposerClient> client;
    if (sessionObj != NULL) {
        client = android_view_SurfaceSession_getClient(env, sessionObj);
    } else {
        client = SurfaceComposerClient::getDefault();
    }
    SurfaceControl *parent = reinterpret_cast<SurfaceControl*>(parentObject);
    sp<SurfaceControl> surface;
    LayerMetadata metadata;
    Parcel* parcel = parcelForJavaObject(env, metadataParcel);
    if (parcel && !parcel->objectsCount()) {
        status_t err = metadata.readFromParcel(parcel);
        if (err != NO_ERROR) {
          jniThrowException(env, "java/lang/IllegalArgumentException",
                            "Metadata parcel has wrong format");
        }
    }

    status_t err = client->createSurfaceChecked(
            String8(name.c_str()), w, h, format, &surface, flags, parent, std::move(metadata));
    if (err == NAME_NOT_FOUND) {
        jniThrowException(env, "java/lang/IllegalArgumentException", NULL);
        return 0;
    } else if (err != NO_ERROR) {
        jniThrowException(env, OutOfResourcesException, NULL);
        return 0;
    }

    surface->incStrong((void *)nativeCreate);
    return reinterpret_cast<jlong>(surface.get());
}

   通过以上JNI接口获取SurfaceComposerClient 对象,作为跟 SurfaceFlinger 通信的代理对象。

void SurfaceComposerClient::onFirstRef() {
    //getComposerService() 将返回 SF 的 Binder 代理端的 BpSurfaceFlinger 对象
    sp<ISurfaceComposer> sf(ComposerService::getComposerService());
    if (sf != nullptr && mStatus == NO_INIT) {
        sp<ISurfaceComposerClient> conn;
        //调用 SF 的 createConnection()
        conn = sf->createConnection();
        if (conn != nullptr) {
            mClient = conn;
            mStatus = NO_ERROR;
        }
    }
}

  比如截屏接口就会通过SurfaceControl调用到其中的 capture 接口:

status_t ScreenshotClient::capture(const sp<IBinder>& display, ui::Dataspace reqDataSpace,
                                   ui::PixelFormat reqPixelFormat, const Rect& sourceCrop,
                                   uint32_t reqWidth, uint32_t reqHeight, bool useIdentityTransform,
                                   ui::Rotation rotation, bool captureSecureLayers,
                                   sp<GraphicBuffer>* outBuffer, bool& outCapturedSecureLayers) {
    sp<ISurfaceComposer> s(ComposerService::getComposerService());
    if (s == nullptr) return NO_INIT;
    status_t ret = s->captureScreen(display, outBuffer, outCapturedSecureLayers, reqDataSpace,
                                    reqPixelFormat, sourceCrop, reqWidth, reqHeight,
                                    useIdentityTransform, rotation, captureSecureLayers);
    if (ret != NO_ERROR) {
        return ret;
    }
    return ret;
}

然后具体看下核心的SurfaceFlinger实现:

SurfaceFlinger.cpp:

sp<ISurfaceComposerClient> SurfaceFlinger::createConnection() {
    //创建一个Client
    const sp<Client> client = new Client(this);
    return client->initCheck() == NO_ERROR ? client : nullptr;
}

 回到之前,创建完 ViewRootImpl 对象后,接下来调用该对象的 setView() 方法。在 setView() 中调用了 requestLayout() 方法,现在具体来看下这个方法调用流程:

    @Override
    public void requestLayout() {
        if (!mHandlingLayoutInLayoutRequest) {
            checkThread();
            mLayoutRequested = true;
            scheduleTraversals();
        }
    }
    @UnsupportedAppUsage
    void scheduleTraversals() {
        if (!mTraversalScheduled) {
            mTraversalScheduled = true;
            mTraversalBarrier = mHandler.getLooper().getQueue().postSyncBarrier();
       //启动TraversalRunnable mChoreographer.postCallback( Choreographer.CALLBACK_TRAVERSAL, mTraversalRunnable,
null); notifyRendererOfFramePending(); pokeDrawLockIfNeeded(); } }
    final class TraversalRunnable implements Runnable {
        @Override
        public void run() {
            doTraversal();
        }
    }
    void doTraversal() {
        if (mTraversalScheduled) {
            mTraversalScheduled = false;
            mHandler.getLooper().getQueue().removeSyncBarrier(mTraversalBarrier);

            if (mProfile) {
                Debug.startMethodTracing("ViewAncestor");
            }
        //调用performTraversals 
performTraversals();
if (mProfile) {
                Debug.stopMethodTracing();
                mProfile = false;
            }
        }
    }
    private void performTraversals() {
        // cache mView since it is used so much below...
        final View host = mView; //它就是 DecorView
...
        if (mFirst || windowShouldResize || viewVisibilityChanged || cutoutChanged || params != null
                || mForceNextWindowRelayout) {
            mForceNextWindowRelayout = false;

            if (isViewVisible) {
                // If this window is giving internal insets to the window
                // manager, and it is being added or changing its visibility,
                // then we want to first give the window manager "fake"
                // insets to cause it to effectively ignore the content of
                // the window during layout.  This avoids it briefly causing
                // other windows to resize/move based on the raw frame of the
                // window, waiting until we can finish laying out this window
                // and get back to the window manager with the ultimately
                // computed insets.
                insetsPending = computesInternalInsets && (mFirst || viewVisibilityChanged);
            }
...
            try {
                if (DEBUG_LAYOUT) {
                    Log.i(mTag, "host=w:" + host.getMeasuredWidth() + ", h:" +
                            host.getMeasuredHeight() + ", params=" + params);
                }

                if (mAttachInfo.mThreadedRenderer != null) {
                    // relayoutWindow may decide to destroy mSurface. As that decision
                    // happens in WindowManager service, we need to be defensive here
                    // and stop using the surface in case it gets destroyed.
                    if (mAttachInfo.mThreadedRenderer.pause()) {
                        // Animations were running so we need to push a frame
                        // to resume them
                        mDirty.set(0, 0, mWidth, mHeight);
                    }
                    mChoreographer.mFrameInfo.addFlags(FrameInfo.FLAG_WINDOW_LAYOUT_CHANGED);
                }
                // 关键函数relayoutWindow
                relayoutResult = relayoutWindow(params, viewVisibility, insetsPending);

                if (DEBUG_LAYOUT) Log.v(mTag, "relayout: frame=" + frame.toShortString()
                        + " cutout=" + mPendingDisplayCutout.get().toString()
                        + " surface=" + mSurface);

                // If the pending {@link MergedConfiguration} handed back from
                // {@link #relayoutWindow} does not match the one last reported,
                // WindowManagerService has reported back a frame from a configuration not yet
                // handled by the client. In this case, we need to accept the configuration so we
                // do not lay out and draw with the wrong configuration.
                if (!mPendingMergedConfiguration.equals(mLastReportedMergedConfiguration)) {
                    if (DEBUG_CONFIGURATION) Log.v(mTag, "Visible with new config: "
                            + mPendingMergedConfiguration.getMergedConfiguration());
                    performConfigurationChange(mPendingMergedConfiguration, !mFirst,
                            INVALID_DISPLAY /* same display */);
                    updatedConfiguration = true;
                }

            } catch (RemoteException e) {
            }
...
        }

        boolean cancelDraw = mAttachInfo.mTreeObserver.dispatchOnPreDraw() || !isViewVisible;

        if (!cancelDraw) {
            if (mPendingTransitions != null && mPendingTransitions.size() > 0) {
                for (int i = 0; i < mPendingTransitions.size(); ++i) {
                    mPendingTransitions.get(i).startChangingAnimations();
                }
                mPendingTransitions.clear();
            }

            //开始绘制,其中调用了draw(fullRedrawNeeded);
            performDraw();
        } else {
            if (isViewVisible) {
                // Try again
                scheduleTraversals();
            } else if (mPendingTransitions != null && mPendingTransitions.size() > 0) {
                for (int i = 0; i < mPendingTransitions.size(); ++i) {
                    mPendingTransitions.get(i).endChangingAnimations();
                }
                mPendingTransitions.clear();
            }
        }

        if (mAttachInfo.mContentCaptureEvents != null) {
            notifyContentCatpureEvents();
        }

        mIsInTraversal = false;
    }

    再看下relayoutWindow的实现:

...
if (mSurfaceControl.isValid()) { if (!useBLAST()) { //先创建一个本地Surface,然后调用copyFrom 将SurfaceControl信息拷贝到Surface中 mSurface.copyFrom(mSurfaceControl); } else { final Surface blastSurface = getOrCreateBLASTSurface(mSurfaceSize.x, mSurfaceSize.y); // If blastSurface == null that means it hasn't changed since the last time we // called. In this situation, avoid calling transferFrom as we would then // inc the generation ID and cause EGL resources to be recreated. if (blastSurface != null) { mSurface.transferFrom(blastSurface); } } } else { destroySurface(); }
....

 SurfaceControl 类可以看作是一个 wrapper 类,最后会执行 copyFrom() 将其返回给 App 客户端:

    @UnsupportedAppUsage
    public void copyFrom(SurfaceControl other) {
        if (other == null) {
            throw new IllegalArgumentException("other must not be null");
        }

        long surfaceControlPtr = other.mNativeObject;
        if (surfaceControlPtr == 0) {
            throw new NullPointerException(
                    "null SurfaceControl native object. Are you using a released SurfaceControl?");
        }

        //通过JNI获取源SurfaceControl
        long newNativeObject = nativeGetFromSurfaceControl(mNativeObject, surfaceControlPtr);

        synchronized (mLock) {
            if (newNativeObject == mNativeObject) {
                return;
            }
            if (mNativeObject != 0) {
                nativeRelease(mNativeObject);
            }
            //保存到全局mNativeObject用于外部调用
            setNativeObjectLocked(newNativeObject);
        }
    }

Surface 显示过程总结:

  在 App 进程中创建 PhoneWindow 后会创建 ViewRoot。ViewRoot 的创建会创建一个 Surface,这个 Surface 其实是空的,通过与 WindowManagerService 通信 copyFrom() 一个NativeSurface 与 SurfaceFlinger 通信时。

 

关于Native Window:

          Native Window是OpenGL与本地窗口系统之间搭建了桥梁。整个GUI系统至少需要两种本地窗口:

  (1)面向管理者(SurfaceFlinger)
    SurfaceFlinger是系统中所有UI界面的管理者,需要直接或间接的持有“本地窗口”,此本地窗口是FramebufferNativeWindow。
  (2)面向应用程序
    这类本地窗口是Surface。

  正常情况按照SDK向导生成APK应用程序,是采用Skia等第三方图形库,而对于希望使用OpenGL ES来完成复杂界面渲染的应用开发者来说,Android也提供封装的GLSurfaceView(或其他方式)来实现图形显示。

  ①FramebufferNativeWindow
    EGL需要通过本地窗口来为OpenGL/OpenGL ES创建环境。由于OpenGL/ES对多平台支持,考虑到兼容性和移植性。不同平台的本地窗口EGLNativeWindowType数据类型不同。
    Android平台的数据类型是ANativeWindow,像是一份“协议”,规定了一个本地窗口的形态和功能。ANativeWindow是FramebufferNativeWindow的父类。
    Android中,由于多缓冲技术,EGLNativeWindowType所管理的缓冲区最少2个,最大3个。
    FramebufferNativeWindow初始化需要Golloc支持,步骤如下:

    • 加载GRALLOC_HARDWARE_MODULE_ID模块,参见上节。
    • 分别打开fb和gralloc设备,打开后的设备由全局变量fbDev和grDev管理。
    • 根据设备的属性来给FramebufferNativeWindow赋初值。
    • 根据FramebufferNativeWindow的实现来填充ANativeWindow中的“协议”
    • 其他一些必要的初始化

  ②应用程序的本地窗口 - Surface
    Surface也继承了ANativeWindow

 class Surface : public ANativeObjectBase<ANativeWindow, Surface, RefBase>

     Surface是面向Android系统中所有UI应用程序的,即它承担着应用进程中的UI显示需求。
    需要面向上层实现(主要是Java层)提供绘制图像的画板。SurfaceFlinger需要收集系统中所有应用程序绘制的图像数据,然后集中显示到物理屏幕上。Surface需要扮演相应角色,本质上还是由SurfaceFlinger服务统一管理的,涉及到很多跨进程的通信细节。

  ③Surface的创建
    Surface将通过mGraphicBufferProducer来获取buffer,这些缓冲区会被记录在mSlots中数据中。mGraphicBufferProducer这一核心成员的初始化流程如下:
      ViewRootImpl持有一个Java层的Surface对象(mSurface)。
      ViewRootImpl向WindowManagerService发起relayout请求,此时mSurface被赋予真正的有效值,将辗转生成的SurfaceControl通过Surface.copyFrom()函数复制到mSurface中。
    由此,Surface由SurfaceControl管理,SurfaceControl由SurfaceComposerClient创建。SurfaceComposerClient获得的匿名Binder是ISurfaceComposer,其服务端实现是SurfaceFlinger。而Surface依赖的IGraphicBufferProducer对象在Service端的实现是BufferQueue。

class SurfaceFlinger : 
  public BinderService<SurfaceFlinger>, //在ServiceManager中注册为SurfaceFlinger
  public BnSurfaceComposer,//实现的接口却叫ISurfaceComposer

  ④SurfaceFlinger服务框架:
    Buffer,Consumer,Producer是“生产者-消费者”模型中的3个参与对象,如何协调好它们的工作是应用程序能否正常显示UI的关键。
    Buffer是BufferQueue,Producer是应用程序,Consumer是SurfaceFlinger。

 Surface内部提供一个BufferQueue,与上层和SurfaceFlinger形成一个生产者消费者模型,上层对应Producer,SurfaceFlinger对应Consumer。三者通过Buffer产生联系,每个Buffer都有四种状态:

  • Free:可被上层使用;
  • Dequeued:出列,正在被上层使用;
  • Queued:入列,已完成上层绘制,等待SurfaceFlinger合成;
  • Acquired:被获取,SurfaceFlinger正持有该Buffer进行合成;

  如此循环,形成一个Buffer被循环使用的过程(FREE-> DEQUEUED->QUEUED->ACQUIRED->FREE)。

  BufferQueue中的mSlots数组用于管理期内的缓冲区,最大容器是32。数据缓冲区的空间是动态分配的,应用程序与SurfaceFlinger都是使用OpenGL ES来完成UI显示。Layer类在SurfaceFlinger中表示“层”,通俗地讲就是代表了一个“画面”,最终物理屏幕上的显示结果就是通过对系统中同时存在的所有“画面”进行处理叠加而成。

 

posted on 2019-07-30 21:00  sheldon_blogs  阅读(51833)  评论(5编辑  收藏  举报

导航