3 帖子 • 分页： 1 / 1
There is increasing interest in delivering a composited X desktop, and debate about what
driver model can best accelerate the compositing operations. Two candidate models are: 1)
the existing XFree86/X.Org DDX loadable driver framework, and 2) a replacement DDX
that uses OpenGL for its rendering and modesetting.
Candidate 1, the XFree86/X.Org DDX loadable driver framework, has been in use since the
release of XFree86 4.0 in March 2000, and is what both the XFree86 and X.Org X servers use
in shipping operating systems. In this framework, a hardware-specific driver is loaded and
performs the tasks of rendering and modesetting. The framework does not dictate how the
loaded driver performs these tasks.
Candidate 2, the X-on-OpenGL model, eliminates hardware-specific X drivers, and instead
uses OpenGL to perform all X rendering. The X server, as proposed in this model, also uses
the OpenGL driver to perform modesetting, through the EGL_MESA_screen_surface 
.............and make the case that the existing loadable driver model is
better suited for achieving our goals. Finally, we end with a roadmap for how to achieve a
production-quality composited X desktop.
no security measure is worth anything if an attacker has physical access to the machine