Hal and Xorg output devices

Jon Smirl jonsmirl at gmail.com
Sat Feb 26 15:16:46 PST 2005


On Sun, 27 Feb 2005 00:45:19 +0200 (IST), Ely Levy
<elylevy-xserver at cs.huji.ac.il> wrote:
> Yea agreed.
> 
> > Parallel arguements exist for why X should not enumerate the PCI bus
> > from user space. The kernel instead provides a hotplug safe way to
> > find devices via sysfs that doesn't need root priv.
> 
> We have only one problem then, how to make that portable for system which
> doesn't have sysfs (and please no I don't care about anything by linux:)

X is not an operating system, and it needs to stop pretending to be
one. PCI bus support in X may have been a necessary feature ten years
ago but it is not something that should be in X today. I am working to
remove this in the Linux version of X.

There is nothing to prevent other OS's from continuing to run the XAA
current code or implementing their own versions of PCI support.
However, I am not interested in trying to build an OS independent PCI
probing library. There is wide variation is how Windows, Linux,
Solaris and the Mac kernel do PCI device support. People are free to
continue using the current code or undertake the effort to build
platform specific support. Writing a univeral library for this is
complex and would take a lot of time.

In the XGL model X is completely out of the driver business. At no
point should X be directly messing with the hardware. With XGL it is
each platform's responsibility to come up with a standalone OpenGL
stack that hides the hardware from X.

OpenGL does not provide all of the API calls needed to implement XGL.
We are picking up the additional APIs needed from OpenGL-ES. 
http://www.khronos.org/opengles/ Look for EGL. There are still a
couple more APIs that are needed, we are working on those. Discussions
on this are happening on dir-devel

-- 
Jon Smirl
jonsmirl at gmail.com



More information about the xorg mailing list