文章出处:http://danielwood.cublog.cn
作者:Daniel Wood
SurfaceFlinger的启动过程还是从Zygote说起。Zygote起来后会调用SystemServer.java[frameworks/base/services/java/com/android/server]里面的main函数,然后调用本地函数init1(),然后调用的是JNI的com_android_server_SystemServer.cpp里面的android_server_SystemServer_init1函数。
static void android_server_SystemServer_init1(JNIEnv* env, jobject clazz) { system_init(); } |
然后调用
System_init.cpp[frameworks/base/cmds/system_server/library]的system_init函数,通过获取属性字段system_init.startsurfaceflinger,如果字段值为1,那么就在这里启动surfaceflinger。
char propBuf[PROPERTY_VALUE_MAX]; property_get("system_init.startsurfaceflinger", propBuf, "1"); if (strcmp(propBuf, "1") == 0) { // Start the SurfaceFlinger SurfaceFlinger::instantiate(); } |
然而,另一方面,有一个可执行文件surfaceflinger,由目录framework/base/cmds/surfaceflinger编译产生,目录下的主要文件main_surfaceflinger.cpp里面就一个main函数:
int main(int argc, char** argv) { sp<ProcessState> proc(ProcessState::self()); sp<IServiceManager> sm = defaultServiceManager(); LOGI("ServiceManager: %p", sm.get()); SurfaceFlinger::instantiate(); ProcessState::self()->startThreadPool(); IPCThreadState::self()->joinThreadPool(); } |
以上两者都会调用SurfaceFlinger.cpp文件的instantiate函数。
void SurfaceFlinger::instantiate() { defaultServiceManager()->addService( String16("SurfaceFlinger"), new SurfaceFlinger()); } |
如果你想在可执行文件中启动SurfaceFlinger,那么你可以在init.rc文件中增加类似如下语句:
service surfaceflinger /system/bin/surfaceflinger user root onrestart restart zygote disabled |
当然你也必须设置属性字段system_init.startsurfaceflinger为0,这个工作可以在init.rc中完成。
setprop system_init.startsurfaceflinger 0 |
surfaceflinger构造函数调用init()函数【surfaceflinger.cpp】,init函数主要打印"SurfaceFlinger is starting"的Log信息,并且对一些debug属性进行配置。
surfaceflinger构造函数调用readyToRun函数【surfaceflinger.cpp】,至于为什么会调用readyToRun函数(并没有显式的调用语句),主要是因为surfaceflinger是一个线程类,必须实现并会调用如下两个函数:一是readyToRun(),该函数定义了线程循环前需要初始化的内容;二是threadLoop(),每个线程都必须实现,该函数定义了线程执行的内容,如果该函数返回true,线程会继续调用threadLoop(),如果返回false,线程将退出。-->选自参考文献。
关于readyToRun将在下节分析
SurfaceFlinger启动过程分析(二)
上节说到SurfaceFlinger的readyToRun函数。先来看看它的代码:
(Google Android 2.2)
SurfaceFlinger.cpp
status_t SurfaceFlinger::readyToRun() { LOGI( "SurfaceFlinger's main thread ready to run. " "Initializing graphics H/W..."); // we only support one display currently int dpy = 0; { // initialize the main display GraphicPlane& plane(graphicPlane(dpy)); DisplayHardware* const hw = new DisplayHardware(this, dpy); plane.setDisplayHardware(hw); } // create the shared control-block mServerHeap = new MemoryHeapBase(4096, MemoryHeapBase::READ_ONLY, "SurfaceFlinger read-only heap"); LOGE_IF(mServerHeap==0, "can't create shared memory dealer"); mServerCblk = static_cast<surface_flinger_cblk_t*>(mServerHeap->getBase()); LOGE_IF(mServerCblk==0, "can't get to shared control block's address"); new(mServerCblk) surface_flinger_cblk_t; // initialize primary screen // (other display should be initialized in the same manner, but // asynchronously, as they could come and go. None of this is supported // yet). const GraphicPlane& plane(graphicPlane(dpy)); const DisplayHardware& hw = plane.displayHardware(); const uint32_t w = hw.getWidth(); const uint32_t h = hw.getHeight(); const uint32_t f = hw.getFormat(); hw.makeCurrent(); // initialize the shared control block mServerCblk->connected |= 1<<dpy; display_cblk_t* dcblk = mServerCblk->displays + dpy; memset(dcblk, 0, sizeof(display_cblk_t)); dcblk->w = plane.getWidth(); dcblk->h = plane.getHeight(); dcblk->format = f; dcblk->orientation = ISurfaceComposer::eOrientationDefault; dcblk->xdpi = hw.getDpiX(); dcblk->ydpi = hw.getDpiY(); dcblk->fps = hw.getRefreshRate(); dcblk->density = hw.getDensity(); asm volatile ("":::"memory"); // Initialize OpenGL|ES glActiveTexture(GL_TEXTURE0); glBindTexture(GL_TEXTURE_2D, 0); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); glTexEnvx(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); glPixelStorei(GL_UNPACK_ALIGNMENT, 4); glPixelStorei(GL_PACK_ALIGNMENT, 4); glEnableClientState(GL_VERTEX_ARRAY); glEnable(GL_SCISSOR_TEST); glShadeModel(GL_FLAT); glDisable(GL_DITHER); glDisable(GL_CULL_FACE); const uint16_t g0 = pack565(0x0F,0x1F,0x0F); const uint16_t g1 = pack565(0x17,0x2f,0x17); const uint16_t textureData[4] = { g0, g1, g1, g0 }; glGenTextures(1, &mWormholeTexName); glBindTexture(GL_TEXTURE_2D, mWormholeTexName); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT); glTexParameterx(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT); glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, 2, 2, 0, GL_RGB, GL_UNSIGNED_SHORT_5_6_5, textureData); glViewport(0, 0, w, h); glMatrixMode(GL_PROJECTION); glLoadIdentity(); glOrthof(0, w, h, 0, 0, 1); LayerDim::initDimmer(this, w, h); mReadyToRunBarrier.open(); /* * We're now ready to accept clients... */ // start boot animation property_set("ctl.start", "bootanim"); return NO_ERROR; } |
调用readyToRun函数用于初始化整个显示系统。
readyToRun()调用过程如下[这部分摘自网上资料]:
(1)执行new DisplayHardware(this,dpy),通过DisplayHardware初始化Framebuffer、EGL并获取OpenGL ES信息。
(2)创建共享的内存控制块。
(3)将EGL与当前屏幕绑定。
(4)初始化共享内存控制块。
(5)初始化OpenGL ES。
(6)显示开机动画。
上面的六点作为阅读代码的提纲及参考,下面对照代码进行分析:
(1)创建一个DisplayHardware,通过它的init函数去初始化Framebuffer、EGL并获取OpenGL ES信息。
DisplayHardware.cpp[frameworks/base/libs/surfaceflinger/displayhardware]
DisplayHardware::DisplayHardware( const sp<SurfaceFlinger>& flinger, uint32_t dpy) : DisplayHardwareBase(flinger, dpy) { init(dpy); } |
init函数的代码狠长,我们一块一块,一句一句地分析:
void DisplayHardware::init(uint32_t dpy) { mNativeWindow = new FramebufferNativeWindow(); ... |
首先亮相的是第一句(如上),new一个FramebufferNativeWindow。
FramebufferNativeWindow构造函数的代码也不少,我们去掉一些次要的代码,挑重要的关键的说:
FramebufferNativeWindow::FramebufferNativeWindow() : BASE(), fbDev(0), grDev(0), mUpdateOnDemand(false) { hw_module_t const* module; if (hw_get_module(GRALLOC_HARDWARE_MODULE_ID, &module) == 0) { int stride; int err; err = framebuffer_open(module, &fbDev); LOGE_IF(err, "couldn't open framebuffer HAL (%s)", strerror(-err)); err = gralloc_open(module, &grDev); LOGE_IF(err, "couldn't open gralloc HAL (%s)", strerror(-err)); // bail out if we can't initialize the modules if (!fbDev || !grDev) return; mUpdateOnDemand = (fbDev->setUpdateRect != 0); // initialize the buffer FIFO mNumBuffers = 2; mNumFreeBuffers = 2; mBufferHead = mNumBuffers-1; buffers[0] = new NativeBuffer( fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB); buffers[1] = new NativeBuffer( fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB); err = grDev->alloc(grDev, fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB, &buffers[0]->handle, &buffers[0]->stride); LOGE_IF(err, "fb buffer 0 allocation failed w=%d, h=%d, err=%s",fbDev->width,fbDev->height, strerror(-err)); err = grDev->alloc(grDev, fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB, &buffers[1]->handle, &buffers[1]->stride); LOGE_IF(err, "fb buffer 1 allocation failed w=%d, h=%d, err=%s",fbDev->width,fbDev->height, strerror(-err)); ... } else { LOGE("Couldn't get gralloc module"); } ... } |
关键的代码都被我高亮了,从最后一行的else的LOGE中可以看出这里主要是获得gralloc这个模块。模块ID定义在:gralloc.h[hardware/libhardware/include/hardware]
#define GRALLOC_HARDWARE_MODULE_ID "gralloc" |
ps:有时候代码中的log狠有用,可以帮助我们读懂代码,而且logcat也是我们调试代码的好东西。
首先打开framebuffer和gralloc这两个模块
framebuffer_open
和
gralloc_open
这两个接口在
gralloc.h
里面定义
static inline int framebuffer_open(const struct hw_module_t* module, struct framebuffer_device_t** device) { return module->methods->open(module, GRALLOC_HARDWARE_FB0, (struct hw_device_t**)device); } static inline int gralloc_open(const struct hw_module_t* module, struct alloc_device_t** device) { return module->methods->open(module, GRALLOC_HARDWARE_GPU0, (struct hw_device_t**)device); } |
两者指定的是gralloc.cpp中同一个函数gralloc_device_open,但是用的是不同的设备名,函数名和设备名分别在gralloc.cpp和gralloc.h中定义。
gralloc.h[hardware/libhardware/include/hardware] #define GRALLOC_HARDWARE_FB0 "fb0" #define GRALLOC_HARDWARE_GPU0 "gpu0" gralloc.cpp[hardware/libhardware/modules/gralloc] static struct hw_module_methods_t gralloc_module_methods = { open: gralloc_device_open }; |
gralloc.cpp[hardware/libhardware/modules/gralloc]
int gralloc_device_open(const hw_module_t* module, const char* name, hw_device_t** device) { int status = -EINVAL; if (!strcmp(name, GRALLOC_HARDWARE_GPU0)) { gralloc_context_t *dev; dev = (gralloc_context_t*)malloc(sizeof(*dev)); /* initialize our state here */ memset(dev, 0, sizeof(*dev)); /* initialize the procs */ dev->device.common.tag = HARDWARE_DEVICE_TAG; dev->device.common.version = 0; dev->device.common.module = const_cast<hw_module_t*>(module); dev->device.common.close = gralloc_close; dev->device.alloc = gralloc_alloc; dev->device.free = gralloc_free; *device = &dev->device.common; status = 0; } else { status = fb_device_open(module, name, device); } return status; } |
gralloc_device_open函数通过设备名字来进行相关的初始化工作。打开framebuffer则调用fb_device_open函数。fb_device_open函数定义在framebuffer.cpp中。
int fb_device_open(hw_module_t const* module, const char* name, hw_device_t** device) { int status = -EINVAL; if (!strcmp(name, GRALLOC_HARDWARE_FB0)) { alloc_device_t* gralloc_device; status = gralloc_open(module, &gralloc_device); if (status < 0) return status; /* initialize our state here */ fb_context_t *dev = (fb_context_t*)malloc(sizeof(*dev)); memset(dev, 0, sizeof(*dev)); /* initialize the procs */ dev->device.common.tag = HARDWARE_DEVICE_TAG; dev->device.common.version = 0; dev->device.common.module = const_cast<hw_module_t*>(module); dev->device.common.close = fb_close; dev->device.setSwapInterval = fb_setSwapInterval; dev->device.post = fb_post; dev->device.setUpdateRect = 0; private_module_t* m = (private_module_t*)module; status = mapFrameBuffer(m); if (status >= 0) { ... *device = &dev->device.common; } } return status; } |
fb_device_open函数是framebuffer.cpp里面的函数它会再次调用gralloc_open函数,调用gralloc_open并没有什么实际的用途,只是检测模块的正确性,感觉这句话没有必要,还是我哪里理解错了???因为gralloc_device这个变量在后面都没有用到啊。
哈哈,经过测试,把以下几句注释掉,然后make,烧到手机上,手机基本功能仍旧正常,看来这几句代码狠有可能是没有什么特别用处的。
alloc_device_t* gralloc_device; status = gralloc_open(module, &gralloc_device); if (status < 0) return status; |
然后调用mapFrameBuffer函数,就是将显示缓冲区映射到用户空间,这样在用户空间就可以直接对显示缓冲区进行读写操作。mapFrameBuffer函数的主体功能是在mapFrameBufferLocked函数里面完成的。
关于mapFrameBuffer函数,在下节讲解。
SurfaceFlinger启动过程分析(三)
内存映射对于framebuffer来说非常重要,因为通常用户是不能直接操作物理地址空间的(也就是物理内存?),然而通过mmap映射之后,将framebuffer的物理地址空间映射到用户空间的一段虚拟地址中,用户就可以通过操作这段虚拟内存而间接操作framebuffer了,你在那段虚拟内存中画了图,相应的图就会显示到屏幕上。
——这段是自己的理解,有错必究!
下面是framebuffer.cpp中的mapFrameBufferLocked函数。
int mapFrameBufferLocked(struct private_module_t* module) { // already initialized... if (module->framebuffer) { return 0; } char const * const device_template[] = { "/dev/graphics/fb%u", "/dev/fb%u", 0 }; int fd = -1; int i=0; char name[64]; while ((fd==-1) && device_template[i]) { snprintf(name, 64, device_template[i], 0); fd = open(name, O_RDWR, 0); i++; } if (fd < 0) return -errno; struct fb_fix_screeninfo finfo; if (ioctl(fd, FBIOGET_FSCREENINFO, &finfo) == -1) return -errno; struct fb_var_screeninfo info; if (ioctl(fd, FBIOGET_VSCREENINFO, &info) == -1) return -errno; info.reserved[0] = 0; info.reserved[1] = 0; info.reserved[2] = 0; info.xoffset = 0; info.yoffset = 0; info.activate = FB_ACTIVATE_NOW; /* * Explicitly request 5/6/5 */ info.bits_per_pixel = 16; info.red.offset = 11; info.red.length = 5; info.green.offset = 5; info.green.length = 6; info.blue.offset = 0; info.blue.length = 5; info.transp.offset = 0; info.transp.length = 0; /* * Request NUM_BUFFERS screens (at lest 2 for page flipping) */ info.yres_virtual = info.yres * NUM_BUFFERS; uint32_t flags = PAGE_FLIP; if (ioctl(fd, FBIOPUT_VSCREENINFO, &info) == -1) { info.yres_virtual = info.yres; flags &= ~PAGE_FLIP; LOGW("FBIOPUT_VSCREENINFO failed, page flipping not supported"); } if (info.yres_virtual < info.yres * 2) { // we need at least 2 for page-flipping info.yres_virtual = info.yres; flags &= ~PAGE_FLIP; LOGW("page flipping not supported (yres_virtual=%d, requested=%d)", info.yres_virtual, info.yres*2); } if (ioctl(fd, FBIOGET_VSCREENINFO, &info) == -1) return -errno; int refreshRate = 1000000000000000LLU / ( uint64_t( info.upper_margin + info.lower_margin + info.yres ) * ( info.left_margin + info.right_margin + info.xres ) * info.pixclock ); if (refreshRate == 0) { // bleagh, bad info from the driver refreshRate = 60*1000; // 60 Hz } if (int(info.width) <= 0 || int(info.height) <= 0) { // the driver doesn't return that information // default to 160 dpi info.width = ((info.xres * 25.4f)/160.0f + 0.5f); info.height = ((info.yres * 25.4f)/160.0f + 0.5f); } float xdpi = (info.xres * 25.4f) / info.width; float ydpi = (info.yres * 25.4f) / info.height; float fps = refreshRate / 1000.0f; LOGI( "using (fd=%d)/n" "id = %s/n" "xres = %d px/n" "yres = %d px/n" "xres_virtual = %d px/n" "yres_virtual = %d px/n" "bpp = %d/n" "r = %2u:%u/n" "g = %2u:%u/n" "b = %2u:%u/n", fd, finfo.id, info.xres, info.yres, info.xres_virtual, info.yres_virtual, info.bits_per_pixel, info.red.offset, info.red.length, info.green.offset, info.green.length, info.blue.offset, info.blue.length ); LOGI( "width = %d mm (%f dpi)/n" "height = %d mm (%f dpi)/n" "refresh rate = %.2f Hz/n", info.width, xdpi, info.height, ydpi, fps ); if (ioctl(fd, FBIOGET_FSCREENINFO, &finfo) == -1) return -errno; if (finfo.smem_len <= 0) return -errno; module->flags = flags; module->info = info; module->finfo = finfo; module->xdpi = xdpi; module->ydpi = ydpi; module->fps = fps; /* * map the framebuffer */ int err; size_t fbSize = roundUpToPageSize(finfo.line_length * info.yres_virtual);//对齐页 module->framebuffer = new private_handle_t(dup(fd), fbSize, 0); module->numBuffers = info.yres_virtual / info.yres; module->bufferMask = 0; void* vaddr = mmap(0, fbSize, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0); if (vaddr == MAP_FAILED) { LOGE("Error mapping the framebuffer (%s)", strerror(errno)); return -errno; } module->framebuffer->base = intptr_t(vaddr); memset(vaddr, 0, fbSize); return 0; } |
这个函数就是和驱动相关的调用,其实结合驱动去看代码是很有意思的,把一路都打通了。
该函数首先通过open函数打开设备结点。
"/dev/graphics/fb%u"和"/dev/fb%u",如果前一个顺利打开的话,那么就不打开第二个。我的Log显示打开的是第一个设备结点/dev/graphics/fb%u。
然后通过ioctl读取设备的固定参数(FBIOGET_FSCREENINFO)和可变参数(FBIOGET_VSCREENINFO)。
【kernel部分的代码在drivers/video/fbmem.c中。】
然后对可变参数进行修改,通过ioctl设置(FBIOPUT_VSCREENINFO)显示屏的可变参数。
设置好以后再ioctl-FBIOGET_VSCREENINFO获得可变参数,然后在log上打出显示屏的各个参数设置,也就是我们开机看到的一长串log。
I/gralloc ( 1620): using (fd=8) I/gralloc ( 1620): id = truly-ILI9327 I/gralloc ( 1620): xres = 240 px I/gralloc ( 1620): yres = 400 px I/gralloc ( 1620): xres_virtual = 240 px I/gralloc ( 1620): yres_virtual = 800 px I/gralloc ( 1620): bpp = 16 I/gralloc ( 1620): r = 11:5 I/gralloc ( 1620): g = 5:6 I/gralloc ( 1620): b = 0:5 I/gralloc ( 1620): width = 38 mm (160.421051 dpi) I/gralloc ( 1620): height = 64 mm (158.750000 dpi) I/gralloc ( 1620): refresh rate = 60.00 Hz |
然后通过mmap完成对显示缓存区的映射。这样mapFrameBufferLocked函数的任务算是完成了。
好了,以上所讲的只是(1)中的第一句话而已
Displayhardware.cpp中的init函数。
mNativeWindow = new FramebufferNativeWindow(); |
SurfaceFlinger启动过程分析(四)
在加载完framebuffer和gralloc模块之后,我们来看FramebufferNativeWindow构造函数中的代码:
buffers[0] = new NativeBuffer( fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB); buffers[1] = new NativeBuffer( fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB); err = grDev->alloc(grDev, fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB, &buffers[0]->handle, &buffers[0]->stride); LOGE_IF(err, "fb buffer 0 allocation failed w=%d, h=%d, err=%s", fbDev->width,fbDev->height, strerror(-err)); err = grDev->alloc(grDev, fbDev->width, fbDev->height, fbDev->format, GRALLOC_USAGE_HW_FB, &buffers[1]->handle, &buffers[1]->stride); LOGE_IF(err, "fb buffer 1 allocation failed w=%d, h=%d, err=%s",fbDev->width,fbDev->height, strerror(-err)); |
该构造函数中关键的就剩下这四句高亮代码了,这四句也是framebuffer双缓存机制的关键。
首先新建了两个NativeBuffer,然后通过grDev为它们分配内存空间。这个grDev就是上面gralloc_open的gralloc设备模块。
grDev
-
>alloc这个函数在gralloc_device_open函数里面指定了是gralloc.cpp中的gralloc_alloc函数。
dev->device.alloc = gralloc_alloc; |
为两个缓冲区分配完内存之后,FramebufferNativeWindow构造函数的事情就算完了。下面继续看DisplayHardware.cpp中init函数接下去的代码。
DisplayHardware.cpp
if (hw_get_module(OVERLAY_HARDWARE_MODULE_ID, &module) == 0) { overlay_control_open(module, &mOverlayEngine); } // initialize EGL ... |
接下去就获得overlay模块,前提是你的设备支持overlay。
然后就初始化EGL。
DisplayHardware.cpp
EGLDisplay display = eglGetDisplay(EGL_DEFAULT_DISPLAY); eglInitialize(display, NULL, NULL); eglGetConfigs(display, NULL, 0, &numConfigs); EGLConfig config; status_t err = EGLUtils::selectConfigForNativeWindow( display, attribs, mNativeWindow.get(), &config); |
eglGetDisplay是EGL用来获取物理屏幕句柄的函数。返回的是EGLDisplay,代表一个物理显示设备。调用这个函数进入的是egl.cpp[frameworks/base/opengl/libs/egl]
EGLDisplay eglGetDisplay(NativeDisplayType display) { uint32_t index = uint32_t(display); if (index >= NUM_DISPLAYS) { return setError(EGL_BAD_PARAMETER, EGL_NO_DISPLAY); } if (egl_init_drivers() == EGL_FALSE) { return setError(EGL_BAD_PARAMETER, EGL_NO_DISPLAY); } EGLDisplay dpy = EGLDisplay(uintptr_t(display) + 1LU); return dpy; } |
它会调用egl_init_drivers去初始化设备。
egl_init_drivers->egl_init_drivers_locked
下面简单贴一下egl_init_drivers_locked代码:
EGLBoolean egl_init_drivers_locked() { if (sEarlyInitState) { // initialized by static ctor. should be set here. return EGL_FALSE; } // get our driver loader Loader& loader(Loader::getInstance()); cnx = &gEGLImpl[IMPL_SOFTWARE]; if (cnx->dso == 0) { cnx->hooks[GLESv1_INDEX] = &gHooks[GLESv1_INDEX][IMPL_SOFTWARE]; cnx->hooks[GLESv2_INDEX] = &gHooks[GLESv2_INDEX][IMPL_SOFTWARE]; cnx->dso = loader.open(EGL_DEFAULT_DISPLAY, 0, cnx); if (cnx->dso) { EGLDisplay dpy = cnx->egl.eglGetDisplay(EGL_DEFAULT_DISPLAY); LOGE_IF(dpy==EGL_NO_DISPLAY, "No EGLDisplay for software EGL!"); d->disp[IMPL_SOFTWARE].dpy = dpy; if (dpy == EGL_NO_DISPLAY) { loader.close(cnx->dso); cnx->dso = NULL; } } } cnx = &gEGLImpl[IMPL_HARDWARE]; if (cnx->dso == 0) { ... } else { LOGD("3D hardware acceleration is disabled"); } } return EGL_TRUE; } |
egl_init_drivers_locked()
函数的作用就是填充
gEGLImpl[IMPL_SOFTWARE]
和
gEGLImpl[IMPL_ HARDWARE]
两个数组项。达到通过
gEGLImpl[IMPL_SOFTWARE]
和
gEGLImpl[IMPL_ HARDWARE]
两个数组项就可以调用
libGLES_android.so
库中所有函数的目的
。
cnx->hooks[GLESv1_INDEX] = &gHooks[GLESv1_INDEX][IMPL_SOFTWARE]; cnx->hooks[GLESv2_INDEX] = &gHooks[GLESv2_INDEX][IMPL_SOFTWARE];
|
上面这两句代码的作用是引用赋值,在loader
.
open完以后
,
cnx
-
>hooks[GLESv1_INDEX]会被赋值,而相对应的
gHooks
[GLESv1_INDEX
]
[IMPL_SOFTWARE
]也会被赋值。
Loader的构造函数先从/system/lib/egl/egl.cfg中读取配置,如果不存在,那就选用默认配置。
Loader::Loader() { char line[256]; char tag[256]; FILE* cfg = fopen("/system/lib/egl/egl.cfg", "r"); if (cfg == NULL) { // default config LOGD("egl.cfg not found, using default config"); gConfig.add( entry_t(0, 0, "android") ); } else { while (fgets(line, 256, cfg)) { int dpy; int impl; if (sscanf(line, "%u %u %s", &dpy, &impl, tag) == 3) { //LOGD(">>> %u %u %s", dpy, impl, tag); gConfig.add( entry_t(dpy, impl, tag) ); } } fclose(cfg); } } |
默认的配置为(0, 0, "android")并把它放在gConfig中,以备在调用Loader.open的时候使用。
void* Loader::open(EGLNativeDisplayType display, int impl, egl_connection_t* cnx) { /* * TODO: if we don't find display/0, then use 0/0 * (0/0 should always work) */ void* dso; char path[PATH_MAX]; int index = int(display); driver_t* hnd = 0; const char* const format = "/system/lib/egl/lib%s_%s.so"; char const* tag = getTag(index, impl); if (tag) { snprintf(path, PATH_MAX, format, "GLES", tag); dso = load_driver(path, cnx, EGL | GLESv1_CM | GLESv2); if (dso) { hnd = new driver_t(dso); } else { // Always load EGL first snprintf(path, PATH_MAX, format, "EGL", tag); dso = load_driver(path, cnx, EGL); if (dso) { hnd = new driver_t(dso); // TODO: make this more automated snprintf(path, PATH_MAX, format, "GLESv1_CM", tag); hnd->set( load_driver(path, cnx, GLESv1_CM), GLESv1_CM ); snprintf(path, PATH_MAX, format, "GLESv2", tag); hnd->set( load_driver(path, cnx, GLESv2), GLESv2 ); } } } LOG_FATAL_IF(!index && !impl && !hnd, "couldn't find the default OpenGL ES implementation " "for default display"); return (void*)hnd; } |
Loader::open
这个函数首先去加载
/system/lib/egl/libGLES_android.so
,如果加载成功,那么对
EGL | GLESv1_CM | GLESv2
三个函数库,进行初始化。如果加载不成功,那么就加载
libEGL_android.so
,
libGLESv1_CM_android.so
,
libGLESv2_android.so
这三个库
,事实上我们的
/system/lib/egl
目录下面只有
libGLES_android.so
这一个库,所以加载
libGLES_android.so
库。
Ps:libEGL.so ,libGLESv1_CM.so, libGLESv2.so三个库在/system/lib目录下面。
下面简单地分析下EGL的配置。首先在Loader的构造函数中获取了EGL的配置信息0, 0, "android",然后把它放在一个结构体中,这个结构体名为entry_t,定义如下
struct entry_t { entry_t() { } entry_t(int dpy, int impl, const char* tag); int dpy; int impl; String8 tag; }; |
随后在Loader::open中调用getTag(index, impl),其实为getTag(0, 0)。所以getTag返回的是字符串android。
const char* Loader::getTag(int dpy, int impl) { const Vector<entry_t>& cfgs(gConfig); const size_t c = cfgs.size(); for (size_t i=0 ; i<c ; i++) { if (dpy == cfgs[i].dpy) if (impl == cfgs[i].impl) return cfgs[i].tag.string(); } return 0; } |
现在有了库的路径path = /system/lib/egl/libGLES_android.so,通过load_driver函数来加载函数库。
Loader::load_driver
void *Loader::load_driver(const char* driver_absolute_path, egl_connection_t* cnx, uint32_t mask) { if (access(driver_absolute_path, R_OK)) { // this happens often, we don't want to log an error return 0; }//加载libGLES_android.so void* dso = dlopen(driver_absolute_path, RTLD_NOW | RTLD_LOCAL); if (dso == 0) { const char* err = dlerror(); LOGE("load_driver(%s): %s", driver_absolute_path, err?err:"unknown"); return 0; } LOGD("loaded %s", driver_absolute_path); if (mask & EGL) {//加载EGL函数库 getProcAddress = (getProcAddressType)dlsym(dso, "eglGetProcAddress"); LOGE_IF(!getProcAddress, "can't find eglGetProcAddress() in %s", driver_absolute_path); egl_t* egl = &cnx->egl;//把函数赋值到cnx->egl中 __eglMustCastToProperFunctionPointerType* curr = (__eglMustCastToProperFunctionPointerType*)egl; char const * const * api = egl_names; while (*api) { char const * name = *api; __eglMustCastToProperFunctionPointerType f = (__eglMustCastToProperFunctionPointerType)dlsym(dso, name); if (f == NULL) { // couldn't find the entry-point, use eglGetProcAddress() f = getProcAddress(name); if (f == NULL) { f = (__eglMustCastToProperFunctionPointerType)0; } } *curr++ = f; api++; } } if (mask & GLESv1_CM) {//加载GLESv1_CM函数库 init_api(dso, gl_names, (__eglMustCastToProperFunctionPointerType*) &cnx->hooks[GLESv1_INDEX]->gl, getProcAddress); } if (mask & GLESv2) {//加载GLESv2函数库 init_api(dso, gl_names, (__eglMustCastToProperFunctionPointerType*) &cnx->hooks[GLESv2_INDEX]->gl, getProcAddress); } return dso; } |
通过系统调用
dlopen
打开一个动态链接库。
以下是百度百科对
dlopen
的解释:
dlopen() 功能:打开一个动态链接库 包含头文件: #include <dlfcn.h> 函数定义: void * dlopen( const char * pathname, int mode ); 函数描述: 在dlopen的()函数以指定模式打开指定的动态连接库文件,并返回一个句柄给调用进程。使用dlclose()来卸载打开的库。
|
然后通过dlsym函数获得指向函数地址指针。
以下是百度百科对
dlsym
的解释:
dlsym()的函数原型是 void* dlsym(void* handle,const char* symbol) 该函数在<dlfcn.h>文件中。 handle是由dlopen打开动态链接库后返回的指针,symbol就是要求获取的函数的名称,函数返回值是void*,指向函数的地址,供调用使用。 |
dlsym首先去得到eglGetProcAddress的函数指针,这个函数的原型:void (*eglGetProcAddress(const char *procname)) ();
该函数的作用是返回由procname指定的扩展函数地址。
下面综述一下load_driver函数所做的工作:首先通过dlopen加载libGLES_android.so库,库所在路径为/system/lib/egl/libGLES_android.so,然后从libGLES_android.so库中提取EGL的各个API函数的地址放到cnx->egl中,从libGLES_android.so获取GLESv1_CM的API保存到cnx->hooks[GLESv1_INDEX]->gl中,从libGLES_android.so获取GLESv1_CM的API保存到cnx->hooks[GLESv2_INDEX]->gl。
提取EGLAPI地址的方法是首先通过dlsym函数获得一个获取函数地址的函数eglGetProcAddress的地址,然后遍历EGL的API所在文件frameworks/base/opengl/libs/EGL/egl_entries.in。先通过dlsym获取各个API地址,如果返回NULL再利用eglGetProcAddress去获得,如果依旧为空就把函数地址赋值为0;提取GLESv1_CM和GLESv1_CM库中函数地址方法和提取EGL差不多,只是他们的函数文件保存在frameworks/base/opengl/libs/entries.in中。还有它们把函数地址复制给了cnx->hooks[GLESv1_INDEX]->gl和cnx->hooks[GLESv2_INDEX]->gl。
等加载完库以后在libs/egl/egl.cpp里面的egl_init_drivers_locked就通过cnx->egl.eglGetDisplay(EGL_DEFAULT_DISPLAY);调用eglGetDisplay函数,其实就是调用libGLES_android.so里面的eglGetDisplay函数,libGLES_android.so库是由目录frameworks/base/opengl/libagl生成的,所以libGLES_android.so里面的eglGetDisplay函数是文件libagl/egl.cpp里面的。
其实libs/egl/egl.cpp中的函数,大多是调用libGLES_android.so库里面的,是对其的一种封装,也就是说调用libagl/egl.cpp文件里面的同名函数,如eglGetDisplay,eglCreateWindowSurface,eglCreateContext等。因为libGLES_android.so库是由rameworks/base/opengl/libagl目录生成