Status of VGA Arbiter and Legacy VGA I/O?

[Dave Airlie]

On Tue, 2009-11-24 at 00:37 +0200, Tiago Vignatti wrote:
> Dave Airlie wrote:
> > On Fri, 2009-11-20 at 13:44 -0800, Mike Travis wrote:
> >> Adam Jackson wrote:
> >>> My understanding is, by default the VGA arbiter support code in X will
> >>> assume the driver _does_ need VGA arbitration, unless the driver opts
> >>> out of it, and will wrap device access for you.  So whatever extra setup
> >>> your platform needs should Just Work if you hook it up to the
> >>> arbitration grant in the kernel.
> >>>
> 
> correct, but ...
> 
> 
> > 
> > If you are using a kms supporting driver, its quite easy to remove it
> > from VGA arbitration, radeon and intel in the kernel do it already,
> > if you can turn off VGA decodes on the GPU then you can ask the 
> > kernel to remove it from VGA arbitration.
> > 
> 
> ... there's an unfinished part of the server API concerning the way 
> non-kms driver could skip from VGA arbitration. The function 
> xf86VGAarbiterDeviceDecodes is kind dumb now and should be used for such 
> purpose. We should fix it :/
> 

The problem I had was making sure drivers on VT switch did the correct
dance, at least with kms we know the driver isn't going to just up and
disappear, without KMS this is just another let X hose the system
behind our backs but I'm willing to accept it could be needed.

Dave.