Linux GPU系列-04-ARM MALI GPU OpenGL端到端流程

前面已经讲过了整个软件堆栈，MALI GPU工作流程，这次用一个简单的OpenGL的例子讲讲端到端的整体流程。

OpenGL简介

OpenGL是非常接近GPU模型的API，整体类似一个状态机，对应到GPU的pipeline:

OpenGL编程模型如下图

实际例子

例子的完整源码
下面是关键片段的代码：

float vertices[] = {
    -0.5f, -0.5f, 0.0f, // left
    0.5f, -0.5f, 0.0f, // right
    0.0f,  0.5f, 0.0f  // top
};

unsigned int VBO, VAO;
glGenVertexArrays(1, &VAO);
glGenBuffers(1, &VBO);
// bind the Vertex Array Object first, then bind and set vertex buffer(s), and then configure vertex attributes(s).
glBindVertexArray(VAO);

glBindBuffer(GL_ARRAY_BUFFER, VBO);
glBufferData(GL_ARRAY_BUFFER, sizeof(vertices), vertices, GL_STATIC_DRAW);

glVertexAttribPointer(1, 3, GL_FLOAT, GL_FALSE, 3 * sizeof(float), (void*)0);
glEnableVertexAttribArray(1);

glBindBuffer(GL_ARRAY_BUFFER, 0);
glBindVertexArray(0);

// render loop
// -----------
while (!glfwWindowShouldClose(window))
{
    // render
    // ------
    glClearColor(0.2f, 0.3f, 0.3f, 1.0f);
    glClear(GL_COLOR_BUFFER_BIT);

    // draw our first triangle
    glUseProgram(shaderProgram);
    glBindVertexArray(VAO); // seeing as we only have a single VAO there's no need to bind it every time, but we'll do so to keep things a bit more organized
    glDrawArrays(GL_TRIANGLES, 0, 3);

    // glfw: swap buffers and poll IO events (keys pressed/released, mouse moved etc.)
    // -------------------------------------------------------------------------------
    glfwSwapBuffers(window);
    glfwPollEvents();
}

准备Vertext数据

顶点数据通过glBufferData拷贝给VBO，VBO之前已经通过glBindBuffer/glBindVertextArray和VAO绑定，这些数据通过
以下流程拷贝到host内存，准备给GPU使用，这部分代码都在MESA中：

glBufferData
  main::_mesa_BufferData
    gl_context::Driver.BufferData
        state_trackder::st_bufferobj_data
      state_tracker::bufferobj_data
        util::pipe_buffer_write
          pipe_context::buffer_subdata
            panfrost:u_default_buffer_subdata //各GPU厂商实现，panfrost使用MESA公共函数

准备job

顶点数据放到buffer之后，就通过glDrawArrays绘制一个三角形，最终组织成一个job，流程如下，这部分代码仍然在MESA中：

glDrawArrays
  main::_mesa_DrawArrays
    gl_context::Driver.DrawGallium
      state_tracker::st_draw_gallium
        cso_context:cso_multi_draw
      u_vbuf::u_vbuf_draw_vbo / pipe_context:draw_vbo
        panfrost::panfrost_draw_emit_vertext
          panforst::panfrost_emit_draw_descs
        panfrost::panfrost_emit_vertext_tiler_jobs
          panfrost__panfrost_add_job

提交job

再通过glfwSwapBuffers(glFlush)把之前准备好的job提交给GPU硬件，这里涉及到和DRM、Kernel驱动的交互，具体流程如下：

glfwSwapBuffers
  main::_mesa_Flush
    gl_context::Driver.Flush
      state_tracker::st_glFlush
        draw_do_flush
      panfrost::panfrost_flush
        panfrost::panfrost_batch_submit_jobs
          panfrost::panfrost_batch_submit_ioctl
            drm::drmIoctl(DRM_IOCTL_PANFROST_SUBMIT)

陷入到Kernel之后，由drivers/gpu/drm/panfrost驱动接管，把之前的准备好的job提交给GPU：

panfrost_ioctl_submit
  panfrost_job_push
    drm_sched_entity_push_job
      drm_sched_rq_add_entity
        panfrost_job_run
      panfrost_job_hw_submit
        panfrost_job_write_affinity （设置job和core的亲和性）
          job_write （写job manager寄存器）

参考

Best practices for effective OpenGL programming
OpenGL44PipelineMap
Computer Graphics
The OpenGL Programming Guide 9th Edition
ANATOMIE D’UN PROGRAMME OPENGL