[patch] XVideo on my Radeon

[Alex Deucher]

On 1/25/06, Benjamin Herrenschmidt <benh at kernel.crashing.org> wrote:
>
> > I'm not sure what this specific change exactly should do, it does not
> > really seem to match the cvs log ("Fix bug in Radeon connector table
> > parsing -- parse all entries, not just the first." - at least by just
> > looking at the code it seems feasible that the old code would parse the
> > 2nd entry when the first one would be empty where the new code wouldn't,
> > though since the new code uses a 16bit value instead of an 8bit one I'm
> > not sure this interpretation is correct, as I've no idea how those bios
> > tables really are supposed to look like).
>
> I did that change... The spec says the table is termined with 0x0000 and
> testing only 8 bits makes the code stop parsing before the end as some
> valid entries can start with 0x00 (I had actual examples of that).
>
> However, I don't know about the i > 1 thing. I assumed it was a hack to
> workaround the former bug, it didn't make much sense based on the spec.
> However the spec also provides a connector count so maybe I should test
> that too.
>
> In any way, I don't know for sure what's up with Xv, unless we end up
> flipping CRTC1 and 2 but I'm not sure how/why that would have happened.

Perhaps something gets detected now that wasn't before so the port in
use gets driven by crtc2 as a clone crtc1 (as per the default
behavior).

Alex

>
> Ben.
>
>