Mailinglist Archive: radeonhd (265 mails)

[Alex Deucher <alexdeucher@xxxxxxxxx>]

On Sun, Jan 25, 2009 at 4:59 AM, Kevin Cody Jr. <kcodyjr@xxxxxxxxx> wrote:
> Attached is a patch against airlied's tree which does 3 things to DRM/KMS:
>
>
> 1. Increases DDC read retry from 1 to 10. This seems to have solved, or
> actually just masked, my problem with intermittent read failures. I
> believe this is a sign of a timing issue.
>
> 2. Checks the expected number of extensions before returning the EDID
> blob, reissuing the read to obtain the full block if needed.
>
> 3. Adds my RV610 to the drm_pciids.h; feel free to ignore this part ;)
>
> I am actively working on a standalone EDID parser in userspace. Now, I
> have test data for dealing with extensions. When reasonably complete, I
> suggest that my parser be ported into the kernel as a "library" module,
> and I will adapt the DRM/KMS code to make use of it. This will allow for
> the use of all mode timings, not just those in the first block; as well
> as other extended data such as chromaticity and HDMI and other features
> that I haven't yet read through the specs for.
>
> However it goes down, I do strongly recommend separating the EDID
> parsing from the code which fills the DRM/KMS structs. EDID is
> surprisingly flexible when you account for all possible versions,
> despite its rigid binary format, and there is no real telling just
> exactly where a given datum might be until you're well into the parsing.

I would recommend sending these patches to the dri-devel mailing list
as that is where most drm discussion takes place.

Alex
-- 
To unsubscribe, e-mail: radeonhd+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: radeonhd+help@xxxxxxxxxxxx