Re: [PATCH v1 24/35] drm/vc4: vec: Add support for more analog TV standards

[Mateusz Kwiatkowski]

Hi Maxime,

W dniu 15.08.2022 o 10:37, Maxime Ripard pisze:
> Hi,
>
> On Fri, Jul 29, 2022 at 07:55:30PM +0200, Mateusz Kwiatkowski wrote:
>> Hi Maxime,
>>
>> I think that declaring PAL-B and SECAM-B as the only supported 576i
>> norms is a bit random.
>
> Starting with this patch, PAL-N should be supported as well, right?

Oh, sure. I forgot about it. My brain was too focused on the "standard PAL"
modes, which excludes PAL-N.

>
>> Norms B, D, G, H, I, K, K1 and L (for both PAL and SECAM) are
>> essentially identical if we're talking about baseband signals, AFAIK
>> they only differ when those are modulated as RF signals. I'm not sure
>> if there's a point to differentiating those (that's more about patch
>> 05/35) unless we need to deal with some device that actually features
>> an RF modulator.
>
> What I was aiming for is to have all the cases we have in all the
> drivers covered so that we can make that property generic. i915
> declares and uses all those variants:
>
> https://elixir.bootlin.com/linux/latest/source/drivers/gpu/drm/i915/display/intel_sdvo.c#L68
>
> Especially since it's i915 and it's pretty much the standard as far as
> the uAPI goes, I'd rather avoid any regression there.

OK, if there are already drivers that differentiate those, then it doesn't make
sense to introduce regressions. And yes, there is plenty of software already out
there that differentiate between those modes in the context of composite video.
It still doesn't make much sense from the engineering point of view, though.

>
>> But if we do want to have all those norms separate, then I'd say that
>> VC4 should declare support for all of those, and all should map to the
>> same VEC settings. Some users from e.g. the UK might think that they
>> won't get proper picture if PAL-I is not on the list of supported
>> norms. Same goes for e.g. SECAM-D/K in the former Soviet territories,
>> and so on.
>
> I'd be open to it, but we can always extend vc4 to support those modes
> later on. The work you did to make that easier should make it trivial.

Doing that in the future is OK as well. I just wanted to point out that
PAL-B/D/G/H/I/K/K1/L (and same for SECAM) is the same exact thing as far as
baseband composite video is concerned, so declaring only one of those as
supported is potentially misleading for anybody who is not aware of that fact.

>
> Maxime

Best regards,
Mateusz Kwiatkowski