[Openchrome-devel] Fix modeline list generation for tv outputs

Xavier Bachelot xavier at bachelot.org
Fri Dec 21 01:41:10 PST 2012


On 12/09/2012 09:53 PM, Harry de Boer wrote:
> whoops, I accidentally attached the old patches last time, these are the
> correct ones (possible_crtcs not yet updated). Time to sleep now.
> 
> Harry
> 
> On 09-12-12 19:58, Harry de Boer wrote:
>> Hi Xavier,
>>
>> no hurries, I am still looking at getting all the modes to work so
>> there is at least one other patch to come. I can already get one more
>> mode to work (a bit) by hardcoding some stuff. The problem seems to be
>> that the initial selected mode is lost (desired_mode not set?) and
>> then does some guessing to recover that messes up such that a modeline
>> is chosen that does not match the tv register table. Any ideas about
>> why that mode might be lost? The attached log has some extra output
>> that shows this.
>>
>> Also Willem is still getting a crash when using the tv, we are mailing
>> about that (in Dutch) outside the mailing list.
>>
>> I'll update the possible_crtcs value, it should be possible to set the
>> tv on both crtcs indeed. I did not realise earlier that it was a bit
>> mask so the value of 0x03 did not make sense to me then, but it does now.
>>
>> Cheers,
>> Harry
>>
>> On 09-12-12 17:57, Xavier Bachelot wrote:
>>> I'm still hunting my pile of hardware for a long enough Svideo cable to
>>> test your patches. If I can't find it I'll move the Epia EX closer to
>>> the tv so I can test your patches.
>>>

So I stopped hunting for the 10 meters svideo cable and instead moved
the box near the TV. And as I was moving the box and pulling cables all
over the living room, I thought testing for just Svideo was good but
testing for composite and component too was better ;-)

All of the outputs are properly detected. I tested Composite, SVideo and
YCbCr.
I had to force the TVMode to PAL (the bios is probably set to NTSC) and
to force the standard 720x576 PAL mode rather than the default 1024x768
to actually get something on the screen. I haven't tried any other mode.

When started with all of the 3 outputs connected (composite, svideo,
component, which is probably not a supported setup, the driver is
detecting an RGB output.

I'd say your patches are good to go, unless you have some revised
version that needs to be tested. 2 minor things to fix, there is a white
space at the end of the line 32 in 0001-Fix-VT1625-output-sensing.patch,
and output->possible_crtcs should be set to 0x03 in
0003-There-is-a-possible-crtc-for-TV-1.patch.

I also noticed a weird behaviour : if I start X with say Svideo
connected, it's properly detected and I get a nice login screen. Now, if
I disconnect svideo and instead plug in composite or component, I still
get a picture. I guess the DACs that are not in use are never turned off.

Regards,
Xavier





More information about the Openchrome-devel mailing list