[randr-1.3] ramdac awareness

[Maarten Maathuis]

Ofcource that name should have been Dave Airlie, sorry for the typo ;-)

Maarten.

On 11/23/07, Maarten Maathuis <madman2003 at gmail.com> wrote:
> A talk with dave arlie convinced that the bulk of so called "ramdac"
> registers actually belonged to the crtc and were oddly placed.
>
> He seems to be right (for the time being). I withdraw my request and
> will re-request it should it be needed.
>
> Maarten.
>
> On 11/23/07, Maarten Maathuis <madman2003 at gmail.com> wrote:
> > Just for the record, when i talk of ramdac i mean output stage, guess
> > that's a sideeffect of working on only one type of hardware :-)
> >
> > Maarten.
> >
> > On 11/23/07, Dave Airlie <airlied at gmail.com> wrote:
> > > On Nov 23, 2007 7:41 PM, Maarten Maathuis <madman2003 at gmail.com> wrote:
> > > > But there are outputs which can run on more than one ramdac and those
> > > > that can't, so static approaches are out of the question.
> > > >
> > > > You actually need to occupy ramdac 0, even if the "normal" output on
> > > > there isn't used. The usual situation is like this:
> > > >
> > > > 2 ramdac's, 2 outputs, one can run on the primary ramdac's, the second
> > > > can run on both ramdac's
> > >
> > > But you should be able to deal with all of that inside the driver, by
> > > restricting the combos..
> > >
> > > I can't see how exposing the internals of a specific card arch in the
> > > protocol is going to help..
> > >
> > > A ramdac is just a DAC (analog output), nvidia have registers
> > > co-located with the RAMDAC control that aren't ramdac registers from
> > > what I can see, they can be extended crtc timing regs..
> > >
> > > So you need to distinguish on the hw between what is actually a DAC
> > > output and what is digitial timings etc..
> > >
> > > Dave.
> > >
> >
>