Re: [PATCH] drm: DON'T require each CRTC to have a unique primary plane
From: Paul Cercueil
Date: Mon Mar 29 2021 - 11:22:25 EST
Le lun. 29 mars 2021 à 17:35, Pekka Paalanen <ppaalanen@xxxxxxxxx> a
écrit :
On Mon, 29 Mar 2021 12:41:00 +0100
Paul Cercueil <paul@xxxxxxxxxxxxxxx> wrote:
Hi,
Le lun. 29 mars 2021 à 11:15, Pekka Paalanen <ppaalanen@xxxxxxxxx>
a
écrit :
> On Sat, 27 Mar 2021 11:26:26 +0000
> Paul Cercueil <paul@xxxxxxxxxxxxxxx> wrote:
>
>> It has two mutually exclusive background planes (same Z level)
+ one
>> overlay plane.
>
> What's the difference between the two background planes?
>
> How will generic userspace know to pick the "right" one?
First primary plane cannot scale, supports RGB and C8. Second
primary
plane goes through the IPU, and as such can scale and convert pixel
formats; it supports RGB, non-planar YUV, and multi-planar YUV.
Right now the userspace apps we have will simply pick the first one
that fits the bill.
What would be the downside of exposing just one "virtual" primary
plane, and then have the driver pick one of the two hardware planes as
appropriate per modeset?
The IPU plane is in a different driver, so all the callbacks are
different. That sounds like it would be a mess.
-Paul
Thanks,
pq
>> Le sam. 27 mars 2021 à 11:24, Simon Ser <contact@xxxxxxxxxxx> a
>> écrit
>> :
>> > On Saturday, March 27th, 2021 at 12:22 PM, Paul Cercueil
>> > <paul@xxxxxxxxxxxxxxx> wrote:
>> >
>> >> The ingenic-drm driver has two mutually exclusive primary
planes
>> >> already; so the fact that a CRTC must have one and only one
>> primary
>> >> plane is an invalid assumption.
>> >
>> > Why does this driver expose two primary planes, if it only
has a
>> > single
>> > CRTC?
>>
>>
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel@xxxxxxxxxxxxxxxxxxxxx
>> https://lists.freedesktop.org/mailman/listinfo/dri-devel
>