copybit.mini6410.so？这个有用么？我在2.3的framework/base/opengl/libagl里面没用发现copybit的上层实现，友善弄个下层的hareware层干嘛？求解答

这个太具有迷惑性了！难道友善在android2.3里面添加了copybit支持？

通过查看友善光盘2,3里面libagl里面的代码发现，与2.1相比，这里已经完全去除了copybit接口，贴个简单的例子，注意看两个的最后不同
2.1的ogles_context_t初始化
ogles_context_t *ogles_init(size_t extra)
{
    void* const base = malloc(extra + sizeof(ogles_context_t) + 32);
    if (!base) return 0;

    ogles_context_t *c =
            (ogles_context_t *)((ptrdiff_t(base) + extra + 31) & ~0x1FL);
    memset(c, 0, sizeof(ogles_context_t));
    ggl_init_context(&(c->rasterizer));

    // XXX: this should be passed as an argument
    sp<EGLSurfaceManager> smgr(new EGLSurfaceManager());
    c->surfaceManager = smgr.get();
    c->surfaceManager->incStrong(c);

    sp<EGLBufferObjectManager> bomgr(new EGLBufferObjectManager());
    c->bufferObjectManager = bomgr.get();
    c->bufferObjectManager->incStrong(c);

    ogles_init_array(c);
    ogles_init_matrix(c);
    ogles_init_vertex(c);
    ogles_init_light(c);
    ogles_init_texture(c);

    c->rasterizer.base = base;
    c->point.size = TRI_ONE;
    c->line.width = TRI_ONE;

    // in OpenGL, writing to the depth buffer is enabled by default.
    c->rasterizer.procs.depthMask(c, 1);

    // OpenGL enables dithering by default
    c->rasterizer.procs.enable(c, GL_DITHER);

    c->copybits.blitEngine = NULL;
    c->copybits.minScale = 0;
    c->copybits.maxScale = 0;
    c->copybits.drawSurfaceBuffer = 0;

#ifdef LIBAGL_USE_GRALLOC_COPYBITS
    hw_module_t const* module;
    if (hw_get_module(COPYBIT_HARDWARE_MODULE_ID, &module) == 0) {
        struct copybit_device_t* copyBits;
        if (copybit_open(module, &copyBits) == 0) {
            c->copybits.blitEngine = copyBits;
            {
                int minLim = copyBits->get(copyBits,
                        COPYBIT_MINIFICATION_LIMIT);
                if (minLim != -EINVAL && minLim > 0) {
                    c->copybits.minScale = (1 << 16) / minLim;
                }
            }
            {
                int magLim = copyBits->get(copyBits,
                        COPYBIT_MAGNIFICATION_LIMIT);
                if (magLim != -EINVAL && magLim > 0) {
                    c->copybits.maxScale = min(32*1024-1, magLim) << 16;
                }
            }
        }
    }
#endif // LIBAGL_USE_GRALLOC_COPYBITS

    return c;
}
2.3的ogles_context_t初始化
ogles_context_t *ogles_init(size_t extra)
{
    void* const base = malloc(extra + sizeof(ogles_context_t) + 32);
    if (!base) return 0;

    ogles_context_t *c =
            (ogles_context_t *)((ptrdiff_t(base) + extra + 31) & ~0x1FL);
    memset(c, 0, sizeof(ogles_context_t));
    ggl_init_context(&(c->rasterizer));

    // XXX: this should be passed as an argument
    sp<EGLSurfaceManager> smgr(new EGLSurfaceManager());
    c->surfaceManager = smgr.get();
    c->surfaceManager->incStrong(c);

    sp<EGLBufferObjectManager> bomgr(new EGLBufferObjectManager());
    c->bufferObjectManager = bomgr.get();
    c->bufferObjectManager->incStrong(c);

    ogles_init_array(c);
    ogles_init_matrix(c);
    ogles_init_vertex(c);
    ogles_init_light(c);
    ogles_init_texture(c);

    c->rasterizer.base = base;
    c->point.size = TRI_ONE;
    c->line.width = TRI_ONE;

    // in OpenGL, writing to the depth buffer is enabled by default.
    c->rasterizer.procs.depthMask(c, 1);

    // OpenGL enables dithering by default
    c->rasterizer.procs.enable(c, GL_DITHER);

    return c;
}

而且/frameworks/base/opengl/libagl下面都没有copybit.cpp这个copybit的上层实现的文件了。
请问prebuild里面的copybit.mini6410.so有什么用？

Android 2.2起调用copybit的地方就只有几处, 这也是为何copybit效果不太明显的一个原因了

但是Android2.3已经去掉copybit的上层处理了，为啥下层还能做？

那copybit.mini6410.so里面封装了那些函数？这些函数在2.3里面有可能被调用么

虽然少了, 但还是有地方会调用的

额，找到了。在egl里面

自己按照2.1的方式 在 egl 与 skia 中加上 copybit 会有效果么