Announcing Fedora 11 Alpha (blink)

[Patrick O'Callaghan]

On Tue, 2009-02-10 at 13:10 -0500, Adam Jackson wrote:
> On Tue, 2009-02-10 at 12:17 -0430, Patrick O'Callaghan wrote:
> > On Tue, 2009-02-10 at 11:43 -0500, Adam Jackson wrote:
> > > But nooooo.  Gotta keep having VTs so we can recover when X screws up.
> > > I mean, it's the unix way.  Which apparently means designing failure in
> > > from the start, and calling it a feature.
> > > 
> > > This is a bad design.  The panic button doesn't make it better, it just
> > > makes it okay to be even worse than the design requires.  Take the
> > > training wheels off already.
> > 
> > Do you have an alternative suggestion?
> 
> KMS.  Move the setup and command submission bits of the driver into the
> kernel, so the kernel always has the wherewithal to kick the X server
> off the hardware.  Run a new minimal display server on that, that isn't
> X, but that can host a nested X server.  Audit the display server and
> kernel driver for correctness, which is now tractable since they're
> small and in the right places.  Emulate VTs under this display server,
> and eviscerate most of the rest from the kernel.

I'd forgotten about KMS. It certainly sounds promising, but when are we
likely to see something useful? I know it's kind of supported on Intel
video, but I think not to the extent you mean here.

poc