Re: [Dri-devel] Re: AGP writeback experiments

[Benjamin H. <be...@ke...>]

> I'd prefer to silently fallback to the INREG() solution when the fancy 
> writeback solution looks like it's failing.
> 
> I guess we could add a blacklist as well, but we shouldn't rely on it.

Do we have a reliable way to verify if writeback works ?

Do any of you have enough contacts at ATI to get at least some
confirmation of a reliability problem with writeback and approx.
how much bridges are affected ? Maybe they would be willing
to release some of the workarounds here...

Ben.