[PATCH v2 14/18] d3d11/tests: Add test for depth/stencil shader resource views.

[Henri Verbeet]

On 9 December 2016 at 11:30, Józef Kucia <jkucia at codeweavers.com> wrote:
> Signed-off-by: Józef Kucia <jkucia at codeweavers.com>
> ---
>
> The test was slightly changed. If it still fails the failures should be more
> clear now after "d3d11/tests: Print expected result in case of failures in
> check_texture_sub_resource_*() functions."
>
Still fails:

    d3d11.c:6532: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6537: Test failed: Got 0.00000000e+000, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6542: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6563: Test failed: Got 6.00000024e-001, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6568: Test failed: Got 1.00000000e+000, expected
2.00000003e-001 at (0, 0), sub-resource 0.
    d3d11.c:6594: Test failed: Got 0.00000000e+000, expected
1.00000000e+002 at (0, 0), sub-resource 0.
    d3d11.c:6599: Test failed: Got 1.00000000e+002, expected
2.55000000e+002 at (0, 0), sub-resource 0.
    d3d11.c:6608: Test failed: Got 2.55199997e+002, expected
3.29999995e+000 at (0, 0), sub-resource 0.
    d3d11.c:6613: Test failed: Got 3.29999995e+000, expected
4.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6618: Test failed: Got 4.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6532: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6537: Test failed: Got 0.00000000e+000, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6542: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6563: Test failed: Got 6.00000024e-001, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6568: Test failed: Got 1.00000000e+000, expected
2.00000003e-001 at (0, 0), sub-resource 0.
    d3d11.c:6532: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6537: Test failed: Got 0.00000000e+000, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6542: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6563: Test failed: Got 6.00000024e-001, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6568: Test failed: Got 1.00000000e+000, expected
2.00000003e-001 at (0, 0), sub-resource 0.
    d3d11.c:6594: Test failed: Got 0.00000000e+000, expected
1.00000000e+002 at (0, 0), sub-resource 0.
    d3d11.c:6599: Test failed: Got 1.00000000e+002, expected
2.55000000e+002 at (0, 0), sub-resource 0.
    d3d11.c:6608: Test failed: Got 2.55199997e+002, expected
3.29999995e+000 at (0, 0), sub-resource 0.
    d3d11.c:6613: Test failed: Got 3.29999995e+000, expected
4.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6618: Test failed: Got 4.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6532: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6537: Test failed: Got 0.00000000e+000, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6542: Test failed: Got 1.00000000e+000, expected
0.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6563: Test failed: Got 6.00000024e-001, expected
1.00000000e+000 at (0, 0), sub-resource 0.
    d3d11.c:6568: Test failed: Got 1.00000000e+000, expected
2.00000003e-001 at (0, 0), sub-resource 0.

So I investigated this a little, and the issue seems to be that as
soon as D3D11_BIND_SHADER_RESOURCE is in the bind flags, reads from
the depth texture return stale values. E.g., the failure at line 6563
is from the 0.6f written by the clear at line 6544. This can also be
reproduced by calling check_texture_float() directly on the depth
texture, and is not specific to typeless formats. I'd guess at a
driver bug related to e.g. compressed depth or fast clear handling.
Regardless, I'd expect that using separate resources for the clear and
the shader reads would fix the failures, although it would perhaps
weaken the test a little.