[igt-dev] [PATCH i-g-t v4] tests/kms_atomic: add test to validate immutable zpos

Sharma, Swati2 swati2.sharma at intel.com
Tue Mar 24 12:46:50 UTC 2020



On 24-Mar-20 2:33 PM, Petri Latvala wrote:
> On Tue, Mar 17, 2020 at 12:49:06PM +0530, Swati Sharma wrote:
>> i915 implements immutable zpos property whereas the existing test
>> case is written to validate mutable zpos.
>>
>> Added a new sub-test to validate immutable zpos. Test validates
>> the reported zpos property of plane by making sure only higher
>> zpos plane covers the lower zpos one (lower plane covers full screen,
>> upper plane covers half screen).
>>
>> Only two planes at a time are tested in increasing fashion to avoid
>> combinatorial explosion. By transitive property,
>> if (p1, z1) < (p2, z2) and (p2, z2) < (p3, z3) then (p1, z1) < (p3, z3)
>>
>> where p and z denotes planes and zpos
>>
>> v2: -Removed intel only checks [Martin]
>>      -Used XRGB8888 pixel format as used in other IGTs
>>      -Added documentation [Martin]
>>      -Removed skip, instead continue if plane doesn't support zpos [Martin]
>>      -Sorted planes in increasing order of zpos [Martin]
>> v3: -Fix description [Martin]
>>      -Avoid sorting [Ankit]
>> v4: -Change in commit message [Ankit]
>>      -Few minor changes [Ankit]
>>
>> Issue: https://gitlab.freedesktop.org/drm/intel/issues/404
>> Signed-off-by: Swati Sharma <swati2.sharma at intel.com>
>> Reviewed-by: Martin Peres <martin.peres at linux.intel.com>
>> Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal at intel.com>
>> ---
>>   tests/kms_atomic.c | 160 ++++++++++++++++++++++++++++++++++++++++++---
>>   1 file changed, 150 insertions(+), 10 deletions(-)
>>
>> diff --git a/tests/kms_atomic.c b/tests/kms_atomic.c
>> index 8462d128..4cb34d6d 100644
>> --- a/tests/kms_atomic.c
>> +++ b/tests/kms_atomic.c
>> @@ -121,7 +121,7 @@ static void plane_check_current_state(igt_plane_t *plane, const uint64_t *values
>>   }
>>   
>>   static void plane_commit(igt_plane_t *plane, enum igt_commit_style s,
>> -                                enum kms_atomic_check_relax relax)
>> +                         enum kms_atomic_check_relax relax)
>>   {
>>   	igt_display_commit2(plane->pipe->display, s);
>>   	plane_check_current_state(plane, plane->values, relax);
>> @@ -277,9 +277,9 @@ static uint32_t plane_get_igt_format(igt_plane_t *plane)
>>   }
>>   
>>   static void
>> -plane_primary_overlay_zpos(igt_pipe_t *pipe, igt_output_t *output,
>> -			   igt_plane_t *primary, igt_plane_t *overlay,
>> -			   uint32_t format_primary, uint32_t format_overlay)
>> +plane_primary_overlay_mutable_zpos(igt_pipe_t *pipe, igt_output_t *output,
>> +			           igt_plane_t *primary, igt_plane_t *overlay,
>> +			           uint32_t format_primary, uint32_t format_overlay)
>>   {
>>   	struct igt_fb fb_primary, fb_overlay;
>>   	drmModeModeInfo *mode = igt_output_get_mode(output);
>> @@ -320,7 +320,7 @@ plane_primary_overlay_zpos(igt_pipe_t *pipe, igt_output_t *output,
>>   	igt_plane_set_prop_value(overlay, IGT_PLANE_ZPOS, 1);
>>   
>>   	igt_info("Committing with overlay on top, it has a hole "\
>> -		  "through which the primary should be seen\n");
>> +		 "through which the primary should be seen\n");
>>   	plane_commit(primary, COMMIT_ATOMIC, ATOMIC_RELAX_NONE);
>>   
>>   	igt_assert_eq_u64(igt_plane_get_prop(primary, IGT_PLANE_ZPOS), 0);
>> @@ -346,7 +346,7 @@ plane_primary_overlay_zpos(igt_pipe_t *pipe, igt_output_t *output,
>>   	igt_put_cairo_ctx(pipe->display->drm_fd, &fb_primary, cr);
>>   
>>   	igt_info("Committing with a hole in the primary through "\
>> -		  "which the underlay should be seen\n");
>> +		 "which the underlay should be seen\n");
>>   	plane_commit(primary, COMMIT_ATOMIC, ATOMIC_RELAX_NONE);
>>   
>>   	/* reset it back to initial state */
>> @@ -358,6 +358,136 @@ plane_primary_overlay_zpos(igt_pipe_t *pipe, igt_output_t *output,
>>   	igt_assert_eq_u64(igt_plane_get_prop(overlay, IGT_PLANE_ZPOS), 1);
>>   }
>>   
>> +static void
>> +plane_immutable_zpos(igt_display_t *display, igt_pipe_t *pipe,
>> +		     igt_output_t *output)
>> +{
>> +	cairo_t *cr;
>> +	struct igt_fb fb_ref;
>> +	igt_plane_t *primary;
>> +	drmModeModeInfo *mode;
>> +	igt_pipe_crc_t *pipe_crc;
>> +	igt_crc_t ref_crc, new_crc;
>> +	int n_planes = pipe->n_planes;
>> +	igt_plane_t *plane_ptr[n_planes];
>> +	uint32_t w_lower, h_lower, w_upper, h_upper;
>> +
>> +	igt_require(n_planes >= 2);
>> +
>> +	mode = igt_output_get_mode(output);
>> +	primary = igt_pipe_get_plane_type(pipe, DRM_PLANE_TYPE_PRIMARY);
>> +
>> +	/* for lower plane */
>> +	w_lower = mode->hdisplay;
>> +	h_lower = mode->vdisplay;
>> +
>> +	/* for upper plane */
>> +	w_upper = mode->hdisplay / 2;
>> +	h_upper = mode->vdisplay / 2;
>> +
>> +	igt_create_color_fb(display->drm_fd,
>> +			    w_lower, h_lower,
>> +			    DRM_FORMAT_XRGB8888,
>> +			    I915_TILING_NONE,
>> +			    0.0, 0.0, 0.0, &fb_ref);
>> +
>> +	/* create reference image */
>> +	cr = igt_get_cairo_ctx(display->drm_fd, &fb_ref);
>> +	igt_assert(cairo_status(cr) == 0);
>> +	igt_paint_color(cr, 0, 0, w_lower, h_lower, 0.0, 0.0, 1.0);
>> +	igt_paint_color(cr, w_upper / 2, h_upper / 2, w_upper, h_upper, 1.0, 1.0, 0.0);
>> +	igt_put_cairo_ctx(display->drm_fd, &fb_ref, cr);
>> +	igt_plane_set_fb(primary, &fb_ref);
>> +	igt_display_commit2(display, COMMIT_ATOMIC);
>> +
>> +	/* create the pipe_crc object for this pipe */
>> +	pipe_crc = igt_pipe_crc_new(pipe->display->drm_fd, pipe->pipe,
>> +				    INTEL_PIPE_CRC_SOURCE_AUTO);
>> +
>> +	/* get reference crc */
>> +	igt_pipe_crc_start(pipe_crc);
>> +	igt_pipe_crc_get_current(display->drm_fd, pipe_crc, &ref_crc);
>> +
>> +	igt_plane_set_fb(primary, NULL);
>> +
>> +	for (int k = 0; k < n_planes; k++) {
>> +		int zpos;
>> +		igt_plane_t *temp;
>> +
>> +		temp = &display->pipes[pipe->pipe].planes[k];
>> +
>> +		if (!igt_plane_has_prop(temp, IGT_PLANE_ZPOS))
>> +			continue;
>> +
>> +		assert(zpos < n_planes);
>> +
>> +		zpos = igt_plane_get_prop(temp, IGT_PLANE_ZPOS);
> 
> Do not use plain assert in tests. Use igt_assert() and its
> brothers. igt_assert_lt(zpos, n_planes) prints the values compared
> when it fails.
> 
> You have a test failure (lucky that it happened) because you check
> zpos before you set it to a value...

Thanks Petri and sorry for this :(
> 
> 

-- 
~Swati Sharma


More information about the igt-dev mailing list