[PATCH v2 2/2] d3d11/tests: Test sampling a texture attached as render target.

[Paul Gofman]

I also verified now that not just conflicting OMSetRenderTargets resets
SRV, but also PSGetShaderResources also returns NULL SRV right after
PSSetShaderResources if the resource is already bound as RTV, which
makes fixing it very straightforward. So I will probably send the fix
along with the test.

On 10/22/19 17:45, Paul Gofman wrote:
> Hello Philip,
>
>     yes, thanks, the comment in the test is actually wrong, while the
> test itself still seem to test intended thing to me (while it tests
> setting SRV after RTV only and not doing that in the reverse order). The
> color in the 'backbuffer' (which is checked in the first loop) would be
> '0x80808080' if the texture was actually sampled or contain some other
> mix of values in case of race condition. Consistently keeping the
> constant value added in PS shader suggests the input from texture
> sampling is 0 (while the texture has 0x40404040 as checked in the next
> loop). I will change the comment and maybe tweak a test a bit to make it
> more explicit.
>
> Regards,
>     Paul.
>
> On 10/22/19 17:04, Philip Rebohle wrote:
>> Hi,
>>
>> while this test passes on Windows, I believe it doesn't tests the
>> intended thing, especially given the following comment in the code:
>>
>>> /* The texture is sampled from SRV but is not updated through RTV. */
>