[Intel-gfx] colorkey support for intel i915 gpu driver

[Ville Syrjälä]

On Tue, Apr 09, 2019 at 01:29:41PM +0000, Jim Zhang wrote:
> Villie:
> 
> What is Intel's plan for the colorkey patch?   Does Intel have any plan to review and release?

There is no real plan at this time. But if you have a use case
for it I can try to harass people until someone reviews it :)

> If I go with custom ioctl, and my custom ioctl will only used in Baytrail product, could it be atomic for Baytrail only?

No. We would need to define a new api for it.

> 
> Thanks,
> 
> Jim
> 
> 
> Caterpillar: Confidential Green
> 
> -----Original Message-----
> From: Ville Syrjälä <ville.syrjala at linux.intel.com> 
> Sent: Tuesday, April 2, 2019 7:32 AM
> To: Jim Zhang <Zhang_Jim at cat.com>
> Cc: dri-devel at lists.freedesktop.org; intel-gfx at lists.freedesktop.org
> Subject: Re: colorkey support for intel i915 gpu driver
> 
> On Mon, Apr 01, 2019 at 08:18:13PM +0000, Jim Zhang wrote:
> > Hi Sir/Madam:
> > 
> > I am using the open source Baytrail gpu drm driver.
> > 
> > Linux kernel version 3.10.61:
> > Libdrm package:  2.4.97
> > 
> > When calling function
> > properties = drmModeObjectGetProperties(drmfd, plane_id, 
> > DRM_MODE_OBJECT_PLANE);  it only returns only one property:
> > 
> > This property is:
> >   property->name = "rotation", property->prop_id =4 It looks like that 
> > the Baytrail gpu drm driver does not support colorkey.
> 
> There are two problems currently:
> - Destination colorkey is not implemented on BYT/CHV. I have
>   patches for it but they have not been reviewed by anyone:
>   https://urldefense.proofpoint.com/v2/url?u=https-3A__patchwork.freedesktop.org_series_43902_&d=DwIDAw&c=p0oa49nxxGtbbM2qgM-GB4r4m9OlGg-sEp8sXylY2aQ&r=kszNssoSc2KF2GeTYwo7za6kdvLoemctuEIYtXbA4PI&m=ztfgUCy9ePzHa0zagoDF75AfJJVbElfjXUWmbpBRM58&s=aXK10RLMgC4i_nBRGS6Jzzbv3pXo50PX79myDO5gEYA&e=
> - Colorkey can only be set via a custom i915 specific ioctl
>   (DRM_I915_SET_SPRITE_COLORKEY). There have been a few attempts at
>   a generic property based API that never really went anywhere. It's
>   a rather difficult problem making this generic as each hardware has
>   its own peculiar way of specifying colorkeying. The main problem
>   with the custom ioctl is that it's not atomic with other screen
>   updates.
> 
> So, what kind of use case do you have in mind?
> 
> --
> Ville Syrjälä
> Intel

-- 
Ville Syrjälä
Intel