On 01/19/2012 03:37 PM, Somebody in the thread at some point said:

Thanks for all your responses. We will try it as soon as possible :)

So this tells me that the build determines which resolution it has, this can not be changed through any setting. Is this right?

You can try messing with dss kernel commandlines but I dunno you get anywhere good.

Fracture in video status reflects fracture in TI approach to how the work is done.

There's an upstream effort to get definitive mainline support in, and separately there's customer support based approach from TI that is mainly targeted at Android / Omapzoom.

The code quality is best in mainline but support is incomplete, especially about multihead. In OZ and OZ-ish kernels, filter for what goes in is not as aggressive as mainline, enablement is better but basis is always an old release. That means, the old, working 3.0 code diverges from what got into mainline for later releases and can't be brought on easily unless treated as alien code like SGX.

Linaro TI LT kernel is trying to be a third approach, while worshipping mainline HEAD above all, so we have recent development basis, we glue on whatever missing pieces we can scavenge from wherever, keeping them up to date with Linus HEAD with to goal of being a highly enabled "fake upstream" with the missing pieces integrated.

In your case that boils down to: Jassi added code to come out on HDMI connector at 1080p via EDID on our kernel, AOSP omap 3.0-based kernel doesn't have that support.

-Andy

On 01/23/2012 04:40 PM, Somebody in the thread at some point said:

Hi all,

We have both tried all the builds on the pandaboard 4430 but we have encountered some problems. It seems like those builds are depending on the screen on how they will behave.

The problem at my side is that the display doesn't want to show something. I have tried all the builds on my Asus computer screen (HDMI 1920x1080) and my Sony Bravia http://store.sony.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=10551&storeId=10151&langId=-1&partNumber=KDL40NX700#features (HDMI 1920x1080). But every time the screen stayed black. This was all tested with the HDMI port to HDMI in the screen.

At Mark, the builds worked on his tv at 720p (also Sony Bravia but older) through HDMI. But when he connected the pandaboard from HDMI to DVI (with an converter) it started in phone mode with a low resolution.

We followed the instructions of the links you send to us carefully and we can confirm that Android was running.

Could there be a problem with the EDID protocol in combination with our screens? And is there a way to override it when the screen doesn't support the EDID protocol?

Yeah there seems to be a race between connector presence detect and edid fetch, vs when your TV is ready to give edid. As it stands it seems to work for most TVs OK. The problem's exacerbated by hotplug not working as it should so getting the EDID later is not the same as getting it at the start.

In 3.3-rc1, Rob Clark has gotten some improved DRM-based stuff building on Tomi's previous upstream patches so all this is getting better.

For Android, the video stuff is stuck in a 3.0 kernel timewarp. We have forwardported the special version of dss and other pieces from 3.0 aosp kernel to make accelerated graphics work, but that approach means all the dss pieces are from 3.0 status and are not benefiting from gradual upstream improvements.

-Andy