[Intel-gfx] [PATCH 8/9] drm/i915/bxt: Enable BXT DSI dual link

[Hans de Goede]

Hi,

On 21-02-17 07:21, Srinivas, Vidya wrote:
>
>
>> -----Original Message-----
>> From: Nikula, Jani
>> Sent: Tuesday, February 21, 2017 12:49 AM
>> To: Hans de Goede <hdegoede at redhat.com>; Srinivas, Vidya
>> <vidya.srinivas at intel.com>; Paauwe, Bob J <bob.j.paauwe at intel.com>
>> Cc: intel-gfx at lists.freedesktop.org; Chauhan, Madhav
>> <madhav.chauhan at intel.com>
>> Subject: Re: [Intel-gfx] [PATCH 8/9] drm/i915/bxt: Enable BXT DSI dual link
>>
>> On Mon, 20 Feb 2017, Hans de Goede <hdegoede at redhat.com> wrote:
>>> Hi,
>>>
>>> On 20-02-17 12:00, Jani Nikula wrote:
>>>> On Mon, 20 Feb 2017, "Srinivas, Vidya" <vidya.srinivas at intel.com>
>> wrote:
>>>>> Thanks Jani. I will rebase and re-submit and also  to remove
>>>>> drm_panel interface dependency, I am planning to create panel
>>>>> sequence callbacks in intel_dsi structure itself. Is this approach okay?
>>>>
>>>> I think that's unnecessary overhead. I've come to think we should
>>>> just do what Hans suggested in his patch [1]. It's easiest, and we
>>>> don't really benefit anything from the drm_panel interface or
>>>> function pointer chasing.
>>>>
>>>> Hans, do you think you'll have the time or motivation to refresh your
>>>> series, or should we just let Vidya do this?
>>>>
>>>> As an added difficulty, there's Madhav doing the Geminilake enabling
>>>> at the same time, and these two/three series [2][3][4] are bound to
>>>> conflict to some extent.
>>>
>>> I've my series rebased in a personal repo. As I was planning on
>>> resubmitting it at some point in the future. I can send out a new
>>> version right now if you want ...
>>
>> Vidya, see [1] for Hans' series. I'm inclined to pick (most of) that, and
>> continue from there. Thoughts?
>>
>> BR,
>> Jani.
>
> We went through Hans' series.
> http://www.mail-archive.com/intel-gfx@lists.freedesktop.org/msg113101.html
> We feel its okay to call the panel sequences directly instead of introducing
> new call backs. We can use this approach. But, Hans' series has other changes
> as well.

Right, changes to actually make the driver follow the spec, instead of calling
a number of sequences at the wrong time, these seem like worthwhile changes
to have and it would be good to land them sooner rather then later so that
most bxt qa will be done with the sequences called at the proper times.

Regards,

Hans


  For now, can we have only the generic sequence changes merged and
> we will rebase our changes on top of that.
>
> Regards
> Vidya
>>
>>
>> [1] http://mid.mail-archive.com/20170220140845.1714-1-
>> hdegoede at redhat.com
>>
>>
>> --
>> Jani Nikula, Intel Open Source Technology Center