Re: [PATCH 0/2] Adding support for Wayland for DRM backends

[Karthik Ramanan <a0393906@ti.com>]

On 23-Aug-14 9:31 AM, Denys Dmytriyenko wrote:
> On Sat, Aug 23, 2014 at 09:18:00AM +0530, Karthik Ramanan wrote:
>> On 23-Aug-14 9:09 AM, Denys Dmytriyenko wrote:
>>> On Fri, Aug 22, 2014 at 02:24:20PM +0000, Cooper Jr., Franklin wrote:
>>>>
>>>>
>>>>> -----Original Message-----
>>>>> From: R, Karthik
>>>>> Sent: Friday, August 22, 2014 9:08 AM
>>>>> To: Cooper Jr., Franklin
>>>>> Cc: R, Karthik; meta-arago@arago-project.org; Devshatwar, Nikhil
>>>>> Subject: Re: [meta-arago] [PATCH 0/2] Adding support for Wayland for DRM
>>>>> backends
>>>>>
>>>>> On 22-Aug-14 7:34 PM, Cooper Jr., Franklin wrote:
>>>>>>
>>>>>>
>>>>>>> -----Original Message-----
>>>>>>> From: R, Karthik
>>>>>>> Sent: Friday, August 22, 2014 8:44 AM
>>>>>>> To: Cooper Jr., Franklin; meta-arago@arago-project.org
>>>>>>> Cc: Devshatwar, Nikhil
>>>>>>> Subject: RE: [meta-arago] [PATCH 0/2] Adding support for Wayland for
>>>>>>> DRM backends
>>>>>>>
>>>>>>> * This is good to go into the master branch
>>>>>>> * Would be nice if this can be back ported onto dora.
>>>>>> [Franklin] I'm assuming you really need this in dora correct? The reason I
>>>>> say this is currently your patches as is won't do anything for you since newer
>>>>> versions of the wayland recipes exist which means those newer versions will
>>>>> be used not this older version. If you really need this in master then I would
>>>>> suggest updating these patches against Wayland 1.4/1.5. I have pending
>>>>> patches for Wayland 1.5 so if it is accepted I would rebase those patches off
>>>>> of that version.
>>>>>
>>>>> I actually meant daisy and not dora.
>>>>> Yes, I agree but when its intended to be used I was thinking about setting the
>>>>> PREFERRED_PROVIDER.
>>>
>>>> [Franklin] Due to issues with Qt 5 we need Wayland 1.5 so going back to 1.3
>>>> won't be an option in master atleast for am57x. So unless you absolutely
>>>> must use Wayland 1.3 I would recommend at least in Arago master updating to
>>>> the newer version of wayland so we can stay on the same version.
>>>
>>> Franklin,
>>>
>>> My understanding is that GLSDK team has done extensive testing and validation
>>> of the patches against 1.3 and up-porting them now to either 1.4 or 1.5 is out
>>> of the question due to time constraints.
>>>
>>> But I would agree that from maintenance point of view it would be much more
>>> desirable to have a single version used by all our products.
>>>
>>
>> To add, migrating from one version of Wayland to another also has a
>> lot of underlying dependencies especially when you consider hardware
>> accelerated graphics via DRM. So there are lot of other components
>> that we have to add support for and you cannot consider
>> Wayland/Weston as a lone component. GLSDK was earlier on Wayland
>> 1.0.6 and then moved to 1.3.0, there are dependencies such as SGX
>> DDK and DRM (both userspace and kernel) that have to move together.
>>
>> If you are only considering minimal wayland support - that is fbdev,
>> and not using multiple display/desktop sharing and support for EGL
>> etc. then migration is kind of easier.
>>
>> This will be useful for those customers who are interested in these
>> other features and using DRM.
>
> Thanks, Karthik,
>
> Even though fbdev is still the main engine on other platforms, like am335 and
> am43x, we are slowly moving towards DRM support. And EGL/SGX is a bigger gap
> for sure.
>
> Anyway, what are your plans on migrating to Wayland 1.4 or 1.5 with all the
> corresponding dependencies?
>

Its mostly requirements driven. The move to Wayland 1.3 was more of a 
requirement from Genivi and the extended desktop support from several 
customers. We do have plans to move to Wayland 1.4 or later but that 
will happen near future. Can't comment on the timelines now.

While Kevin and Chase were here this week, we discussed the gaps that 
exist in the current LCPD kernel/core SDK that prevents us from getting 
to these newer versions. These points have been taken note of and we 
will start working towards addressing those gaps. It should happen in a 
matter of time. Like you said, this is a big gap and effort and we will 
work towards it.

And also with regard to the EGL/SGX, there are things that exist in the 
meta-glsdk layer (we discussed about this a month ago) and as the 
drivers and kernel divergence reduces these will be made available in 
the meta-ti/meta-arago.

And we will also keep pushing things from meta-glsdk into meta-ti/arago 
after the GLSDK releases. These will be good points to sync.