Re: [igt-dev] [PATCH i-g-t v3] tools/intel_dp_compliance: Use ARGB8888 format fbs for Gen 10

[Manasi Navare <manasi.d.navare@intel.com>]

On Fri, Jul 13, 2018 at 02:11:14PM -0700, Radhakrishna Sripada wrote:
> According to Display WA #1172, to truly bypass the color data on Gen 10
> use ARGB8888 instead of XRGB8888 to pass compliance.
> 
> v2: Use ARGB8888 format only for video pattern fb, set per pixel alpha
>     value to 0xff in fill_framebuffer.(Imre)
> v3: Set the aplha value for each pixel(Imre)
> 
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Manasi Navare <manasi.d.navare@intel.com>
> Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
> ---
>  tools/intel_dp_compliance.c | 20 ++++++++++++++------
>  1 file changed, 14 insertions(+), 6 deletions(-)
> 
> diff --git a/tools/intel_dp_compliance.c b/tools/intel_dp_compliance.c
> index c40548e79227..14631a46a9c1 100644
> --- a/tools/intel_dp_compliance.c
> +++ b/tools/intel_dp_compliance.c
> @@ -175,7 +175,7 @@ static int tio_fd;
>  struct termios saved_tio;
>  
>  drmModeRes *resources;
> -int drm_fd, modes;
> +int drm_fd, modes, gen;
>  uint64_t tiling = LOCAL_DRM_FORMAT_MOD_NONE;
>  uint32_t depth = 24, stride, bpp;
>  int specified_mode_num = -1, specified_disp_id = -1;
> @@ -506,9 +506,13 @@ static int setup_video_pattern_framebuffer(struct connector *dp_conn)
>  
>  	video_width = dp_conn->test_pattern.hdisplay;
>  	video_height = dp_conn->test_pattern.vdisplay;
> +	/*
> +	 * Display WA1172: Gen10 To pass the color data unaffected set either
> +	 * per-pixel alpha or Plane alpha to 0xff. Use ARGB8888 and set alpha to 0xff.
> +	 */
>  	dp_conn->test_pattern.fb = igt_create_fb(drm_fd,
>  						 video_width, video_height,
> -						 DRM_FORMAT_XRGB8888,
> +						 gen == 10 ? DRM_FORMAT_ARGB8888 : DRM_FORMAT_XRGB8888,
>  						 LOCAL_DRM_FORMAT_MOD_NONE,
>  						 &dp_conn->test_pattern.fb_pattern);
>  	igt_assert(dp_conn->test_pattern.fb);
> @@ -537,6 +541,7 @@ static int fill_framebuffer(struct connector *dp_conn)
>  	uint32_t *red_ptr, *green_ptr, *blue_ptr, *white_ptr, *src_ptr, *dst_ptr;
>  	int x, y;
>  	int32_t pixel_val;
> +	uint8_t alpha;
>  
>  	video_width = dp_conn->test_pattern.hdisplay;
>  	video_height = dp_conn->test_pattern.vdisplay;
> @@ -554,10 +559,12 @@ static int fill_framebuffer(struct connector *dp_conn)
>  	while (x < video_width) {
>  		for (pixel_val = 0; pixel_val < 256;
>  		     pixel_val = pixel_val + (256 / tile_width)) {
> -			red_ptr[x] = pixel_val << 16;
> -			green_ptr[x] = pixel_val << 8;
> -			blue_ptr[x] = pixel_val << 0;
> -			white_ptr[x] = red_ptr[x] | green_ptr[x] | blue_ptr[x];
> +			alpha = gen == 10 ? 0xff : 0;

Or you could just define alpha = 0xff << 24 here and or with alpha instead of shifting everytime.
But thats just an added optimization.

> +			red_ptr[x] = alpha << 24 | pixel_val << 16;
> +			green_ptr[x] = alpha << 24 | pixel_val << 8;
> +			blue_ptr[x] = alpha << 24 | pixel_val << 0;
> +			white_ptr[x] = alpha << 24 | red_ptr[x] | green_ptr[x] |

This looks good to me, it will keep the same color ramp as asked by the compliance test pattern
but with alpha value set to 0xff for all the pixels.

Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

Manasi

> +				       blue_ptr[x];
>  			if (++x >= video_width)
>  				break;
>  		}
> @@ -1036,6 +1043,7 @@ int main(int argc, char **argv)
>  	set_termio_mode();
>  
>  	drm_fd = drm_open_driver(DRIVER_ANY);
> +	gen = intel_gen(intel_get_drm_devid(drm_fd));
>  
>  	kmstest_set_vt_graphics_mode();
>  	setup_debugfs_files();
> -- 
> 2.9.3
> 
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev