[igt-dev] [PATCH i-g-t 1/9] tests/prime_vgem: Skip fence and gtt specific tests

[Dixit, Ashutosh]

On Fri, 31 Jan 2020 09:31:01 -0800, Ramalingam C wrote:
>
> In the absence of the mappable aperture, skip GTT and
> fence specific tests.
>
> v2:
>   requirement check is moved to subtest start [Zbigniew]
>
> Signed-off-by: Ramalingam C <ramalingam.c at intel.com>
> ---
>  tests/prime_vgem.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/tests/prime_vgem.c b/tests/prime_vgem.c
> index 3bdb23007d56..c483cbe722ea 100644
> --- a/tests/prime_vgem.c
> +++ b/tests/prime_vgem.c
> @@ -846,14 +846,18 @@ igt_main
>	igt_subtest("basic-write")
>		test_write(vgem, i915);
>
> -	igt_subtest("basic-gtt")
> +	igt_subtest("basic-gtt") {
> +		gem_require_mappable_ggtt(i915);
>		test_gtt(vgem, i915);
> +	}

To me seems no reason to skip this. We could either unconditionally use
device_coherent() or use mmap_gtt() if gtt is available else
device_coherent()? Though probably the test would need to be renamed as
test_device_coherent()?

>
>	igt_subtest("shrink")
>		test_shrink(vgem, i915);
>
> -	igt_subtest("coherency-gtt")
> +	igt_subtest("coherency-gtt") {
> +		gem_require_mappable_ggtt(i915);
>		test_gtt_interleaved(vgem, i915);
> +	}

Probably similar situation here, though with WC instead of GTT the test
should always pass?

>
>	for (e = intel_execution_engines; e->name; e++) {
>		igt_subtest_f("%ssync-%s",
> @@ -902,8 +906,10 @@ igt_main
>
>		igt_subtest("basic-fence-read")
>			test_fence_read(i915, vgem);
> -		igt_subtest("basic-fence-mmap")
> +		igt_subtest("basic-fence-mmap") {
> +			gem_require_mappable_ggtt(i915);
>			test_fence_mmap(i915, vgem);
> +		}

Once again, just use device_coherent()?