Re: [igt-dev] [PATCH i-g-t] tests/kms_rotation_crc: Fix multiplane-rotation-cropping* tests

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Tue, Jan 22, 2019 at 04:38:57PM +0200, Juha-Pekka Heikkila wrote:
> Make certain viewports are divisible by four due to intel
> hw workarounds for NV12.
> 
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109381
> Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
> ---
>  tests/kms_rotation_crc.c | 24 ++++++++++++++++++++----
>  1 file changed, 20 insertions(+), 4 deletions(-)
> 
> diff --git a/tests/kms_rotation_crc.c b/tests/kms_rotation_crc.c
> index 8f343e0..d66b489 100644
> --- a/tests/kms_rotation_crc.c
> +++ b/tests/kms_rotation_crc.c
> @@ -467,6 +467,7 @@ static bool get_multiplane_crc(data_t *data, igt_output_t *output,
>  			igt_plane_set_size(planes[c].plane, h, w);
>  
>  		igt_plane_set_position(planes[c].plane, planeinfo[c].x1, planeinfo[c].y1);
> +
>  		igt_plane_set_rotation(planes[c].plane, planeinfo[c].rotation_hw);
>  	}
>  
> @@ -486,10 +487,25 @@ static bool get_multiplane_crc(data_t *data, igt_output_t *output,
>  static void pointlocation(data_t *data, planeinfos *p, drmModeModeInfo *mode,
>  			  int c)
>  {
> -	p[c].x1 = (int32_t)(data->planepos[c].x * mode->hdisplay)
> -			+ ((data->planepos[c].origo & p_right) ? mode->hdisplay : 0);
> -	p[c].y1 = (int32_t)(data->planepos[c].y * mode->vdisplay)
> -			+ ((data->planepos[c].origo & p_bottom) ? mode->vdisplay : 0);
> +	if (data->planepos[c].origo & p_right) {
> +		p[c].x1 = (int32_t)(data->planepos[c].x * mode->hdisplay
> +				+ mode->hdisplay);
> +		p[c].x1 &= ~3;
> +		p[c].x1 -= (p[c].x1+p[c].width-mode->hdisplay)&3;

So assuming the plane is extending past the right edge of the screen (is
that always true here?) the part we want a multiple of four is
'hdisplay-x1', so not quite sure why we're looking at the chopped off
part instead. But since the plane size is already aligned to four
(maybe?) I guess this works.

Hmm, aren't most display resolutions multiple of four anyway? So I wonder
why a simple 'x1 &= ~3' etc. is not sufficient?

> +	} else {
> +		p[c].x1 = (int32_t)(data->planepos[c].x * mode->hdisplay);
> +		p[c].x1 &= ~3;
> +	}
> +
> +	if (data->planepos[c].origo & p_bottom) {
> +		p[c].y1 = (int32_t)(data->planepos[c].y * mode->vdisplay
> +				+ mode->vdisplay);
> +		p[c].y1 &= ~3;
> +		p[c].y1 -= (p[c].y1+p[c].height-mode->vdisplay)&3;
> +	} else {
> +		p[c].y1 = (int32_t)(data->planepos[c].y * mode->vdisplay);
> +		p[c].y1 &= ~3;
> +	}
>  }
>  
>  /*
> -- 
> 2.7.4
> 
> _______________________________________________
> igt-dev mailing list
> igt-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/igt-dev

-- 
Ville Syrjälä
Intel
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev