Re: [Piglit] [Bug] Fragment shader interlock sample shading test assumes wrong guarantees (plain text copy)

[Triang3l]

Follow-up: In further testing, I did get some races with my test 
condition, and it turns out that in OpenGL and Vulkan, unlike in 
Direct3D, gl_SampleMaskIn contains only 1 sample when sample shading is 
used. So there's even less possible meaningful usage of pixel interlock 
with sample shading, maybe only with very tiny primitives covering only 
1 sample, or with all samples but one masked out via the fixed-function 
sample mask (which is exposed in Vulkan directly, but in OpenGL, only as 
a fraction via glSampleCoverage).