[Intel-gfx] [PATCH i-g-t 2/2] tests/i915_query: Engine discovery tests

[Chris Wilson]

Quoting Tvrtko Ursulin (2019-05-01 12:42:59)
> +               switch (engine->engine.engine_class) {
> +               case I915_ENGINE_CLASS_RENDER:
> +                       /* Will be tested later. */
> +                       break;
> +               case I915_ENGINE_CLASS_COPY:
> +                       igt_assert(gem_has_blt(fd));
> +                       break;
> +               case I915_ENGINE_CLASS_VIDEO:
> +                       switch (engine->engine.engine_instance) {
> +                       case 0:
> +                               igt_assert(gem_has_bsd(fd));
> +                               break;
> +                       case 1:
> +                               igt_assert(gem_has_bsd2(fd));
> +                               break;


Is that relationship a given?

One could argue that gem_has_blt() means that I915_EXEC_BLT works, but
without !gem_has_blt() we could still access CLASS_COPY:0 via
ctx->engines[].

> +                       }
> +                       break;
> +               case I915_ENGINE_CLASS_VIDEO_ENHANCE:
> +                       igt_assert(gem_has_vebox(fd));
> +                       break;
> +               default:
> +                       igt_assert(0);
> +               }
> +       }
> +
> +       /* Reverse check to the above - all GET_PARAM engines are present. */
> +       igt_assert(has_engine(engines, I915_ENGINE_CLASS_RENDER, 0));
> +       if (gem_has_blt(fd))
> +               igt_assert(has_engine(engines, I915_ENGINE_CLASS_COPY, 0));
> +       if (gem_has_bsd(fd))
> +               igt_assert(has_engine(engines, I915_ENGINE_CLASS_VIDEO, 0));
> +       if (gem_has_bsd2(fd))
> +               igt_assert(has_engine(engines, I915_ENGINE_CLASS_VIDEO, 1));
> +       if (gem_has_vebox(fd))
> +               igt_assert(has_engine(engines, I915_ENGINE_CLASS_VIDEO_ENHANCE,
> +                                      0));

Whereas this should always make sense, given the legacy interface and
the modern interface, the modern interface should be a superset of the
legacy.

Just thinking aloud.
-Chris