spirv-val: Re-enable OpControlBarrier VU

[spencer-lunarg]

this todo seemed to be added 6 years ago in 12e6860

I am not sure what the problem is, but if there is still CTS issues, that should be something fixed in CTS (or the spec to remove the VU)

[alan-baker]

Have you run validation on CTS with this? It would be good to know if it is now clean or at least which tests fail so we can determine where the problem really is.

[spencer-lunarg]

The issue is CTS is already not 100% clean from validation, I did run locally all the potential tests and it didn't display the VUID-StandaloneSpirv-OpControlBarrier-04650 error

[alan-baker]

The issue is CTS is already not 100% clean from validation, I did run locally all the potential tests and it didn't display the VUID-StandaloneSpirv-OpControlBarrier-04650 error

When I run build-programs in CTS with validation it is clean for me. Do you mean it's not clean of runtime validation? It should be clean of static validation.

[spencer-lunarg]

Do you mean it's not clean of runtime validation? It should be clean of static validation.

sorry, not for spirv-val, I ment with Vulkan Validation Layers... I didn't know if there was a way to run with just spirv-val and not calling it through VVL

[alan-baker]

Do you mean it's not clean of runtime validation? It should be clean of static validation.

sorry, not for spirv-val, I ment with Vulkan Validation Layers... I didn't know if there was a way to run with just spirv-val and not calling it through VVL

vk-build-programs -v in the CTS build. Run it in the data directory.

[spencer-lunarg]

ok, I was unaware about the vk-build-programs executable (and the -v option to validate)

I also see it clean for this spirv-val change ran locally

[alan-baker]

The one odd case for barriers in Vulkan is in tessellation shaders. That should get caught above.