[igt-dev] [PATCH i-g-t 6/7] tests/gem_render_copy: Add subtests for different tiling modes

[Chris Wilson]

Quoting Ville Syrjala (2018-07-04 17:16:45)
> From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> 
> Make sure our rendercopy implementations do the right thing with
> tiled buffers.
> 
> For now we'll just do linear->linear, x-tiled->x-tiled, and
> y-tiled->y-tiled. Not sure there's much point in adding tests
> for different src vs. dst tiling modes?
> 
> Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> ---
>  static void scratch_buf_init(data_t *data, struct igt_buf *buf,
> -                            int width, int height, int stride)
> +                            int width, int height, int stride,
> +                            uint32_t tiling)
>  {
>         drm_intel_bo *bo;
>         int size = height * stride;
> @@ -210,9 +226,12 @@ static void scratch_buf_init(data_t *data, struct igt_buf *buf,
>  
>         buf->bo = bo;
>         buf->stride = stride;
> -       buf->tiling = I915_TILING_NONE;
> +       buf->tiling = tiling;
>         buf->size = size;
>  
> +       drm_intel_bo_set_tiling(buf->bo, &tiling, buf->stride);
> +       igt_assert_eq(tiling, buf->tiling);

I'd feel more confident with

        drm_intel_bo *bo;
-       int size = height * stride;
+       uint32_t tiling = req_tiling;
+       unsigned long pitch;

-       bo = drm_intel_bo_alloc(data->bufmgr, "", size, 4096);
+       bo = drm_intel_bo_alloc_tiled(data->bufmgr, "",
+                                     width, height, 4,
+                                     &tiling, &pitch, 0);
+	igt_assert(tiling === req_tiling);

        buf->bo = bo;
-       buf->stride = stride;
-       buf->tiling = I915_TILING_NONE;
-       buf->size = size;
+       buf->stride = pitch;
+       buf->tiling = tiling;
+       buf->size = pitch * height;

It just so happens that STRIDE is a tile width, but testing odd
drawables might be useful (considering recent bugs ;)
-Chris