[igt-dev] [PATCH] tests/kms_panel_fitting: Move Intel GEN check under a is_i915_device

[Rodrigo Siqueira]

On 05/13, Mark Yacoub wrote:
> From: Mark Yacoub <markyacoub at google.com>
> 
> Encapsulate intel-only operations such as intel_get_drm_devid under a is_i915_device check.
> 
> This allows the subtests to run on non-i915 devices.
> ---
>  tests/kms_panel_fitting.c | 61 +++++++++++++++++++++++----------------
>  1 file changed, 36 insertions(+), 25 deletions(-)
> 
> diff --git a/tests/kms_panel_fitting.c b/tests/kms_panel_fitting.c
> index 4f4c5608..0c028c69 100644
> --- a/tests/kms_panel_fitting.c
> +++ b/tests/kms_panel_fitting.c
> @@ -89,7 +89,7 @@ static void test_panel_fitting(data_t *d)
>  
>  	for_each_pipe_with_valid_output(display, pipe, output) {
>  		drmModeModeInfo *mode, native_mode;
> -		uint32_t devid = intel_get_drm_devid(display->drm_fd);
> +		bool is_plane_scaling_active = true;
>  
>  		/* Check that the "scaling mode" property has been set. */
>  		if (!igt_output_has_prop(output, IGT_CONNECTOR_SCALING_MODE))
> @@ -132,25 +132,26 @@ static void test_panel_fitting(data_t *d)
>  		igt_fb_set_size(&d->fb2, d->plane2, d->fb2.width-200, d->fb2.height-200);
>  		igt_plane_set_position(d->plane2, 100, 100);
>  
> -		/*
> -		 * Most of gen7 and all of gen8 doesn't support plane scaling
> -		 * at all.
> -		 *
> -		 * gen9 pipe C has only 1 scaler shared with the crtc, which
> -		 * means pipe scaling can't work simultaneously with panel
> -		 * fitting.
> -		 *
> -		 * Since this is the legacy path, userspace has to know about
> -		 * the HW limitations, whereas atomic can ask.
> -		 */
> -		if (IS_GEN8(devid) ||
> -		    (IS_GEN7(devid) && !IS_IVYBRIDGE(devid)) ||
> -		    (IS_GEN9(devid) && pipe == PIPE_C))
> -			/* same as visible area of fb => no scaling */
> -			igt_plane_set_size(d->plane2,
> -					   d->fb2.width-200,
> -					   d->fb2.height-200);
> -		else
> +		if (is_i915_device(display->drm_fd)) {
> +			uint32_t devid = intel_get_drm_devid(display->drm_fd);
> +			/*
> +			 * Most of gen7 and all of gen8 doesn't support plane scaling
> +			 * at all.
> +			 *
> +			 * gen9 pipe C has only 1 scaler shared with the crtc, which
> +			 * means pipe scaling can't work simultaneously with panel
> +			 * fitting.
> +			 *
> +			 * Since this is the legacy path, userspace has to know about
> +			 * the HW limitations, whereas atomic can ask.
> +			 */
> +			if (IS_GEN8(devid) ||
> +				(IS_GEN7(devid) && !IS_IVYBRIDGE(devid)) ||
> +				(IS_GEN9(devid) && pipe == PIPE_C)) {
> +				is_plane_scaling_active = false;
> +			}
> +		}
> +		if (is_plane_scaling_active) {
>  			/*
>  			 * different than visible area of fb => plane scaling
>  			 * active
> @@ -158,6 +159,13 @@ static void test_panel_fitting(data_t *d)
>  			igt_plane_set_size(d->plane2,
>  					   mode->hdisplay-200,
>  					   mode->vdisplay-200);
> +		}
> +		else {
> +			/* same as visible area of fb => no scaling */
> +			igt_plane_set_size(d->plane2,
> +				d->fb2.width - 200,
> +				d->fb2.height - 200);
> +		}
>  
>  		/* Plane scaling active (if possible), pfit off */
>  		igt_display_commit2(display, COMMIT_UNIVERSAL);
> @@ -229,13 +237,16 @@ static void test_atomic_fastset(data_t *data)
>  	int valid_tests = 0;
>  	struct stat sb;
>  
> -	/* Until this is force enabled, force modeset evasion. */
> -	if (stat("/sys/module/i915/parameters/fastboot", &sb) == 0)
> -		igt_set_module_param_int(data->drm_fd, "fastboot", 1);
>  
> -	igt_require(display->is_atomic);
> -	igt_require(intel_gen(intel_get_drm_devid(display->drm_fd)) >= 5);
> +	if (is_i915_device(display->drm_fd)) {
> +		/* Until this is force enabled, force modeset evasion. */
> +		if (stat("/sys/module/i915/parameters/fastboot", &sb) == 0)
> +			igt_set_module_param_int(data->drm_fd, "fastboot", 1);
>  
> +		igt_require(intel_gen(intel_get_drm_devid(display->drm_fd)) >= 5);
> +	}
> +
> +	igt_require(display->is_atomic);
>  	for_each_pipe_with_valid_output(display, pipe, output) {
>  		if (!igt_output_has_prop(output, IGT_CONNECTOR_SCALING_MODE))
>  			continue;
> -- 
> 2.31.1.607.g51e8a6a459-goog
> 

Hi Mark,

I tested it in a Raven system, and now I can see it failing in a good
way in my system. Nice patch, thanks.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira at amd.com>

-- 
Rodrigo Siqueira
https://siqueira.tech
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/igt-dev/attachments/20210513/7e54594a/attachment.sig>