AuroraMiddleware/SPIRV-Cross
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #2161 from KhronosGroup/fix-2151

Browse Source 
Fix more edge cases in bool <-> short remapping
This commit is contained in:
Hans-Kristian Arntzen 2023-06-07 15:30:10 +02:00 committed by GitHub
commit 74702f5077
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
35 changed files with 373 additions and 126 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
reference/opt/shaders-msl/comp/composite-array-initialization.comp
View File

@ -71,8 +71,7 @@ kernel void main0(device SSBO& _53 [[buffer(0)]], uint3 gl_WorkGroupID [[threadg
spvUnsafeArray<Data, 2> _25 = spvUnsafeArray<Data, 2>({ Data{ 1.0, 2.0 }, Data{ 3.0, 4.0 } });
spvUnsafeArray<Data, 2> _31 = spvUnsafeArray<Data, 2>({ Data{ X, 2.0 }, Data{ 3.0, 5.0 } });
spvUnsafeArray<Data, 2> data2;
data2 = _31;
spvUnsafeArray<Data, 2> data2 = _31;
_53.outdata[gl_WorkGroupID.x].a = _25[gl_LocalInvocationID.x].a + data2[gl_LocalInvocationID.x].a;
_53.outdata[gl_WorkGroupID.x].b = _25[gl_LocalInvocationID.x].b + data2[gl_LocalInvocationID.x].b;
}

3
reference/opt/shaders-msl/comp/composite-construct.comp
View File

@ -59,8 +59,7 @@ constant uint3 gl_WorkGroupSize [[maybe_unused]] = uint3(1u);
kernel void main0(device SSBO0& _16 [[buffer(0)]], device SSBO1& _32 [[buffer(1)]], uint3 gl_GlobalInvocationID [[thread_position_in_grid]], uint gl_LocalInvocationIndex [[thread_index_in_threadgroup]])
{
spvUnsafeArray<float4, 2> _37 = spvUnsafeArray<float4, 2>({ _16.as[gl_GlobalInvocationID.x], _32.bs[gl_GlobalInvocationID.x] });
spvUnsafeArray<float4, 2> values;
values = _37;
spvUnsafeArray<float4, 2> values = _37;
_16.as[0] = values[gl_LocalInvocationIndex];
_32.bs[1] = float4(40.0);
}

3
reference/opt/shaders-ue4/asm/tesc/hs-incorrect-base-type.invalid.asm.tesc
View File

@ -317,8 +317,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_View& View [[buffer(0)
spvUnsafeArray<uint, 12> _150 = spvUnsafeArray<uint, 12>({ gl_in[0].in_var_LIGHTMAP_ID, gl_in[1].in_var_LIGHTMAP_ID, gl_in[2].in_var_LIGHTMAP_ID, gl_in[3].in_var_LIGHTMAP_ID, gl_in[4].in_var_LIGHTMAP_ID, gl_in[5].in_var_LIGHTMAP_ID, gl_in[6].in_var_LIGHTMAP_ID, gl_in[7].in_var_LIGHTMAP_ID, gl_in[8].in_var_LIGHTMAP_ID, gl_in[9].in_var_LIGHTMAP_ID, gl_in[10].in_var_LIGHTMAP_ID, gl_in[11].in_var_LIGHTMAP_ID });
spvUnsafeArray<float4, 12> _259 = spvUnsafeArray<float4, 12>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position, gl_in[3].in_var_VS_To_DS_Position, gl_in[4].in_var_VS_To_DS_Position, gl_in[5].in_var_VS_To_DS_Position, gl_in[6].in_var_VS_To_DS_Position, gl_in[7].in_var_VS_To_DS_Position, gl_in[8].in_var_VS_To_DS_Position, gl_in[9].in_var_VS_To_DS_Position, gl_in[10].in_var_VS_To_DS_Position, gl_in[11].in_var_VS_To_DS_Position });
spvUnsafeArray<FBasePassVSToDS, 12> _284 = spvUnsafeArray<FBasePassVSToDS, 12>({ FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[0], _145[0], _146[0], _147[0], _148[0], _149[0], _150[0] } }, FBasePassInterpolantsVSToDS{ { } }, _259[0] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[1], _145[1], _146[1], _147[1], _148[1], _149[1], _150[1] } }, FBasePassInterpolantsVSToDS{ { } }, _259[1] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[2], _145[2], _146[2], _147[2], _148[2], _149[2], _150[2] } }, FBasePassInterpolantsVSToDS{ { } }, _259[2] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[3], _145[3], _146[3], _147[3], _148[3], _149[3], _150[3] } }, FBasePassInterpolantsVSToDS{ { } }, _259[3] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[4], _145[4], _146[4], _147[4], _148[4], _149[4], _150[4] } }, FBasePassInterpolantsVSToDS{ { } }, _259[4] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[5], _145[5], _146[5], _147[5], _148[5], _149[5], _150[5] } }, FBasePassInterpolantsVSToDS{ { } }, _259[5] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[6], _145[6], _146[6], _147[6], _148[6], _149[6], _150[6] } }, FBasePassInterpolantsVSToDS{ { } }, _259[6] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[7], _145[7], _146[7], _147[7], _148[7], _149[7], _150[7] } }, FBasePassInterpolantsVSToDS{ { } }, _259[7] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[8], _145[8], _146[8], _147[8], _148[8], _149[8], _150[8] } }, FBasePassInterpolantsVSToDS{ { } }, _259[8] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[9], _145[9], _146[9], _147[9], _148[9], _149[9], _150[9] } }, FBasePassInterpolantsVSToDS{ { } }, _259[9] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[10], _145[10], _146[10], _147[10], _148[10], _149[10], _150[10] } }, FBasePassInterpolantsVSToDS{ { } }, _259[10] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[11], _145[11], _146[11], _147[11], _148[11], _149[11], _150[11] } }, FBasePassInterpolantsVSToDS{ { } }, _259[11] } });
spvUnsafeArray<FBasePassVSToDS, 12> param_var_I;
param_var_I = _284;
spvUnsafeArray<FBasePassVSToDS, 12> param_var_I = _284;
float4 _301 = float4(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, 0.0);
float3 _310 = View_PrimitiveSceneData._m0[(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.PrimitiveId * 26u) + 22u].xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
uint _313 = (gl_InvocationID < 2u) ? (gl_InvocationID + 1u) : 0u;

3
reference/opt/shaders-ue4/asm/tesc/hs-input-array-access.invalid.asm.tesc
View File

@ -355,8 +355,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_View& View [[buffer(0)
spvUnsafeArray<float4, 12> _192 = spvUnsafeArray<float4, 12>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position, gl_in[3].in_var_VS_To_DS_Position, gl_in[4].in_var_VS_To_DS_Position, gl_in[5].in_var_VS_To_DS_Position, gl_in[6].in_var_VS_To_DS_Position, gl_in[7].in_var_VS_To_DS_Position, gl_in[8].in_var_VS_To_DS_Position, gl_in[9].in_var_VS_To_DS_Position, gl_in[10].in_var_VS_To_DS_Position, gl_in[11].in_var_VS_To_DS_Position });
spvUnsafeArray<uint, 12> _193 = spvUnsafeArray<uint, 12>({ gl_in[0].in_var_VS_To_DS_VertexID, gl_in[1].in_var_VS_To_DS_VertexID, gl_in[2].in_var_VS_To_DS_VertexID, gl_in[3].in_var_VS_To_DS_VertexID, gl_in[4].in_var_VS_To_DS_VertexID, gl_in[5].in_var_VS_To_DS_VertexID, gl_in[6].in_var_VS_To_DS_VertexID, gl_in[7].in_var_VS_To_DS_VertexID, gl_in[8].in_var_VS_To_DS_VertexID, gl_in[9].in_var_VS_To_DS_VertexID, gl_in[10].in_var_VS_To_DS_VertexID, gl_in[11].in_var_VS_To_DS_VertexID });
spvUnsafeArray<FHitProxyVSToDS, 12> _230 = spvUnsafeArray<FHitProxyVSToDS, 12>({ FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[0], _143[0] } }, _192[0], _193[0] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[1], _143[1] } }, _192[1], _193[1] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[2], _143[2] } }, _192[2], _193[2] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[3], _143[3] } }, _192[3], _193[3] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[4], _143[4] } }, _192[4], _193[4] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[5], _143[5] } }, _192[5], _193[5] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[6], _143[6] } }, _192[6], _193[6] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[7], _143[7] } }, _192[7], _193[7] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[8], _143[8] } }, _192[8], _193[8] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[9], _143[9] } }, _192[9], _193[9] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[10], _143[10] } }, _192[10], _193[10] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[11], _143[11] } }, _192[11], _193[11] } });
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I;
param_var_I = _230;
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I = _230;
float4 _247 = float4(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, 0.0);
float3 _251 = Primitive.Primitive_NonUniformScale.xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
uint _254 = (gl_InvocationID < 2u) ? (gl_InvocationID + 1u) : 0u;

3
reference/opt/shaders-ue4/asm/tesc/hs-texcoord-array.invalid.asm.tesc
View File

@ -332,8 +332,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_View& View [[buffer(0)
spvUnsafeArray<spvUnsafeArray<float2, 2>, 12> _132 = spvUnsafeArray<spvUnsafeArray<float2, 2>, 12>({ spvUnsafeArray<float2, 2>({ gl_in[0].in_var_TEXCOORD0_0, gl_in[0].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[1].in_var_TEXCOORD0_0, gl_in[1].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[2].in_var_TEXCOORD0_0, gl_in[2].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[3].in_var_TEXCOORD0_0, gl_in[3].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[4].in_var_TEXCOORD0_0, gl_in[4].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[5].in_var_TEXCOORD0_0, gl_in[5].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[6].in_var_TEXCOORD0_0, gl_in[6].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[7].in_var_TEXCOORD0_0, gl_in[7].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[8].in_var_TEXCOORD0_0, gl_in[8].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[9].in_var_TEXCOORD0_0, gl_in[9].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[10].in_var_TEXCOORD0_0, gl_in[10].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[11].in_var_TEXCOORD0_0, gl_in[11].in_var_TEXCOORD0_1 }) });
spvUnsafeArray<float4, 12> _205 = spvUnsafeArray<float4, 12>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position, gl_in[3].in_var_VS_To_DS_Position, gl_in[4].in_var_VS_To_DS_Position, gl_in[5].in_var_VS_To_DS_Position, gl_in[6].in_var_VS_To_DS_Position, gl_in[7].in_var_VS_To_DS_Position, gl_in[8].in_var_VS_To_DS_Position, gl_in[9].in_var_VS_To_DS_Position, gl_in[10].in_var_VS_To_DS_Position, gl_in[11].in_var_VS_To_DS_Position });
spvUnsafeArray<FHitProxyVSToDS, 12> _230 = spvUnsafeArray<FHitProxyVSToDS, 12>({ FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[0], _130[0], _131[0], _132[0] } }, _205[0] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[1], _130[1], _131[1], _132[1] } }, _205[1] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[2], _130[2], _131[2], _132[2] } }, _205[2] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[3], _130[3], _131[3], _132[3] } }, _205[3] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[4], _130[4], _131[4], _132[4] } }, _205[4] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[5], _130[5], _131[5], _132[5] } }, _205[5] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[6], _130[6], _131[6], _132[6] } }, _205[6] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[7], _130[7], _131[7], _132[7] } }, _205[7] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[8], _130[8], _131[8], _132[8] } }, _205[8] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[9], _130[9], _131[9], _132[9] } }, _205[9] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[10], _130[10], _131[10], _132[10] } }, _205[10] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[11], _130[11], _131[11], _132[11] } }, _205[11] } });
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I;
param_var_I = _230;
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I = _230;
float4 _247 = float4(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, 0.0);
float3 _251 = Primitive.Primitive_NonUniformScale.xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
uint _254 = (gl_InvocationID < 2u) ? (gl_InvocationID + 1u) : 0u;

3
reference/opt/shaders-ue4/asm/tesc/tess-factor-must-be-threadgroup.invalid.asm.tesc
View File

@ -147,8 +147,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_Primitive& Primitive [
spvUnsafeArray<float4, 3> _91 = spvUnsafeArray<float4, 3>({ gl_in[0].in_var_TEXCOORD11_centroid, gl_in[1].in_var_TEXCOORD11_centroid, gl_in[2].in_var_TEXCOORD11_centroid });
spvUnsafeArray<float4, 3> _104 = spvUnsafeArray<float4, 3>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position });
spvUnsafeArray<FBasePassVSToDS, 3> _111 = spvUnsafeArray<FBasePassVSToDS, 3>({ FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _90[0], _91[0] } }, FBasePassInterpolantsVSToDS{ { } }, _104[0] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _90[1], _91[1] } }, FBasePassInterpolantsVSToDS{ { } }, _104[1] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _90[2], _91[2] } }, FBasePassInterpolantsVSToDS{ { } }, _104[2] } });
spvUnsafeArray<FBasePassVSToDS, 3> param_var_I;
param_var_I = _111;
spvUnsafeArray<FBasePassVSToDS, 3> param_var_I = _111;
float3 _128 = Primitive.Primitive_NonUniformScale.xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
gl_out[gl_InvocationID].out_var_TEXCOORD10_centroid = param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0;
gl_out[gl_InvocationID].out_var_TEXCOORD11_centroid = param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2;

3
reference/shaders-msl-no-opt/asm/comp/block-like-array-type-construct.asm.comp
View File

@ -70,8 +70,7 @@ kernel void main0()
foo[1] = 2.0;
foo[2] = 3.0;
foo[3] = 4.0;
spvUnsafeArray<float, 4> foo2;
foo2 = foo;
spvUnsafeArray<float, 4> foo2 = foo;
_10 _37 = _10{ { foo[0], foo[1], foo[2], foo[3] }, { foo2[0], foo2[1], foo2[2], foo2[3] } };
}

3
reference/shaders-msl-no-opt/asm/frag/scalar-select.spv14.asm.frag
View File

@ -65,8 +65,7 @@ fragment main0_out main0()
out.FragColor = select(float4(0.0, 0.0, 0.0, 1.0), float4(1.0, 1.0, 0.0, 1.0), bool4(false, true, false, true));
out.FragColor = float4(bool4(false, true, false, true));
_15 _32 = false ? (_15{ 0.0 }) : (_15{ 1.0 });
spvUnsafeArray<float, 2> _33;
_33 = true ? _29 : _30;
spvUnsafeArray<float, 2> _33 = true ? _29 : _30;
return out;
}

6
reference/shaders-msl-no-opt/asm/tesc/copy-tess-level-tri.asm.tesc
View File

@ -59,10 +59,8 @@ kernel void main0(uint gl_InvocationID [[thread_index_in_threadgroup]], uint gl_
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0] = half(_25[0]);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1] = half(_25[1]);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[2] = half(_25[2]);
spvUnsafeArray<float, 2> inner;
inner = spvUnsafeArray<float, 2>({ float(spvTessLevel[gl_PrimitiveID].insideTessellationFactor), 0.0 });
spvUnsafeArray<float, 4> outer;
outer = spvUnsafeArray<float, 4>({ float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[2]), 0.0 });
spvUnsafeArray<float, 2> inner = spvUnsafeArray<float, 2>({ float(spvTessLevel[gl_PrimitiveID].insideTessellationFactor), 0.0 });
spvUnsafeArray<float, 4> outer = spvUnsafeArray<float, 4>({ float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[2]), 0.0 });
gl_out[gl_InvocationID].gl_Position = float4(1.0);
}

6
reference/shaders-msl-no-opt/asm/tesc/tess-fixed-input-array-builtin-array.invalid.asm.tesc
View File

@ -120,8 +120,7 @@ kernel void main0(main0_in in [[stage_in]], uint gl_InvocationID [[thread_index_
p[2].pos = gl_in[2].gl_Position;
p[2].uv = gl_in[2].p_uv;
uint i = gl_InvocationID;
spvUnsafeArray<VertexOutput, 3> param;
param = p;
spvUnsafeArray<VertexOutput, 3> param = p;
uint param_1 = i;
HSOut flattenTemp = _hs_main(param, param_1);
gl_out[gl_InvocationID].gl_Position = flattenTemp.pos;
@ -129,8 +128,7 @@ kernel void main0(main0_in in [[stage_in]], uint gl_InvocationID [[thread_index_
threadgroup_barrier(mem_flags::mem_device | mem_flags::mem_threadgroup);
if (int(gl_InvocationID) == 0)
{
spvUnsafeArray<VertexOutput, 3> param_2;
param_2 = p;
spvUnsafeArray<VertexOutput, 3> param_2 = p;
HSConstantOut _patchConstantResult = PatchHS(param_2);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0] = half(_patchConstantResult.EdgeTess[0]);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1] = half(_patchConstantResult.EdgeTess[1]);

6
reference/shaders-msl-no-opt/asm/tesc/tess-fixed-input-array-builtin-array.invalid.multi-patch.asm.tesc
View File

@ -119,8 +119,7 @@ kernel void main0(uint3 gl_GlobalInvocationID [[thread_position_in_grid]], devic
p[2].pos = gl_in[2].gl_Position;
p[2].uv = gl_in[2].p.uv;
uint i = gl_InvocationID;
spvUnsafeArray<VertexOutput, 3> param;
param = p;
spvUnsafeArray<VertexOutput, 3> param = p;
uint param_1 = i;
HSOut flattenTemp = _hs_main(param, param_1);
gl_out[gl_InvocationID].gl_Position = flattenTemp.pos;
@ -128,8 +127,7 @@ kernel void main0(uint3 gl_GlobalInvocationID [[thread_position_in_grid]], devic
threadgroup_barrier(mem_flags::mem_device | mem_flags::mem_threadgroup);
if (int(gl_InvocationID) == 0)
{
spvUnsafeArray<VertexOutput, 3> param_2;
param_2 = p;
spvUnsafeArray<VertexOutput, 3> param_2 = p;
HSConstantOut _patchConstantResult = PatchHS(param_2);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0] = half(_patchConstantResult.EdgeTess[0]);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1] = half(_patchConstantResult.EdgeTess[1]);

98
reference/shaders-msl-no-opt/frag/struct-bool-edge-cases.frag Normal file
View File

@ -0,0 +1,98 @@
#pragma clang diagnostic ignored "-Wmissing-prototypes"
#pragma clang diagnostic ignored "-Wmissing-braces"
#include <metal_stdlib>
#include <simd/simd.h>
using namespace metal;
template<typename T, size_t Num>
struct spvUnsafeArray
{
T elements[Num ? Num : 1];
thread T& operator [] (size_t pos) thread
{
return elements[pos];
}
constexpr const thread T& operator [] (size_t pos) const thread
{
return elements[pos];
}
device T& operator [] (size_t pos) device
{
return elements[pos];
}
constexpr const device T& operator [] (size_t pos) const device
{
return elements[pos];
}
constexpr const constant T& operator [] (size_t pos) const constant
{
return elements[pos];
}
threadgroup T& operator [] (size_t pos) threadgroup
{
return elements[pos];
}
constexpr const threadgroup T& operator [] (size_t pos) const threadgroup
{
return elements[pos];
}
};
struct Test
{
short a;
short2 b;
short3 c;
short4 d;
};
struct Test2
{
spvUnsafeArray<short, 2> a;
spvUnsafeArray<short2, 2> b;
spvUnsafeArray<short3, 2> c;
spvUnsafeArray<short4, 2> d;
Test e;
};
constant spvUnsafeArray<bool, 2> _40 = spvUnsafeArray<bool, 2>({ true, false });
constant spvUnsafeArray<bool2, 2> _44 = spvUnsafeArray<bool2, 2>({ bool2(true, false), bool2(false, true) });
constant spvUnsafeArray<bool3, 2> _48 = spvUnsafeArray<bool3, 2>({ bool3(true, false, true), bool3(false, true, false) });
constant spvUnsafeArray<bool4, 2> _52 = spvUnsafeArray<bool4, 2>({ bool4(true, false, true, false), bool4(false, true, false, true) });
constant spvUnsafeArray<bool, 2> _82 = spvUnsafeArray<bool, 2>({ true, true });
constant spvUnsafeArray<bool2, 2> _85 = spvUnsafeArray<bool2, 2>({ bool2(true), bool2(false) });
constant spvUnsafeArray<bool3, 2> _87 = spvUnsafeArray<bool3, 2>({ bool3(true), bool3(false) });
constant spvUnsafeArray<bool4, 2> _89 = spvUnsafeArray<bool4, 2>({ bool4(true), bool4(false) });
fragment void main0()
{
spvUnsafeArray<Test, 2> _95 = spvUnsafeArray<Test, 2>({ Test{ short(true), short2(bool2(true, false)), short3(bool3(true)), short4(bool4(false)) }, Test{ short(false), short2(bool2(false, true)), short3(bool3(false)), short4(bool4(true)) } });
Test t;
t.a = short(true);
t.b = short2(bool2(true, false));
t.c = short3(bool3(true, false, true));
t.d = short4(bool4(true, false, true, false));
Test2 t2;
t2.a = { short(_40[0]), short(_40[1]) };
t2.b = { short2(_44[0]), short2(_44[1]) };
t2.c = { short3(_48[0]), short3(_48[1]) };
t2.d = { short4(_52[0]), short4(_52[1]) };
bool a = bool(t.a);
bool2 b = bool2(t.b);
bool3 c = bool3(t.c);
bool4 d = bool4(t.d);
spvUnsafeArray<bool, 2> a2 = { bool(t2.a[0]), bool(t2.a[1]) };
spvUnsafeArray<bool2, 2> b2 = { bool2(t2.b[0]), bool2(t2.b[1]) };
spvUnsafeArray<bool3, 2> c2 = { bool3(t2.c[0]), bool3(t2.c[1]) };
spvUnsafeArray<bool4, 2> d2 = { bool4(t2.d[0]), bool4(t2.d[1]) };
t = Test{ short(true), short2(bool2(true, false)), short3(bool3(true)), short4(bool4(false)) };
t2 = Test2{ spvUnsafeArray<short, 2>({ short(true), short(true) }), spvUnsafeArray<short2, 2>({ short2(bool2(true)), short2(bool2(false)) }), spvUnsafeArray<short3, 2>({ short3(bool3(true)), short3(bool3(false)) }), spvUnsafeArray<short4, 2>({ short4(bool4(true)), short4(bool4(false)) }), Test{ short(true), short2(bool2(true, false)), short3(bool3(true)), short4(bool4(false)) } };
}

6
reference/shaders-msl-no-opt/tesc/copy-tess-level.tesc
View File

@ -61,10 +61,8 @@ kernel void main0(uint gl_InvocationID [[thread_index_in_threadgroup]], uint gl_
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1] = half(_21[1]);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[2] = half(_21[2]);
spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[3] = half(_21[3]);
spvUnsafeArray<float, 2> inner;
inner = spvUnsafeArray<float, 2>({ float(spvTessLevel[gl_PrimitiveID].insideTessellationFactor[0]), float(spvTessLevel[gl_PrimitiveID].insideTessellationFactor[1]) });
spvUnsafeArray<float, 4> outer;
outer = spvUnsafeArray<float, 4>({ float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[2]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[3]) });
spvUnsafeArray<float, 2> inner = spvUnsafeArray<float, 2>({ float(spvTessLevel[gl_PrimitiveID].insideTessellationFactor[0]), float(spvTessLevel[gl_PrimitiveID].insideTessellationFactor[1]) });
spvUnsafeArray<float, 4> outer = spvUnsafeArray<float, 4>({ float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[0]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[1]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[2]), float(spvTessLevel[gl_PrimitiveID].edgeTessellationFactor[3]) });
gl_out[gl_InvocationID].gl_Position = float4(1.0);
}

6
reference/shaders-msl-no-opt/vert/pass-array-by-value.vert
View File

@ -60,10 +60,8 @@ struct main0_in
static inline __attribute__((always_inline))
float4 consume_constant_arrays2(spvUnsafeArray<float4, 4> positions, spvUnsafeArray<float4, 4> positions2, thread int& Index1, thread int& Index2)
{
spvUnsafeArray<float4, 4> indexable;
indexable = positions;
spvUnsafeArray<float4, 4> indexable_1;
indexable_1 = positions2;
spvUnsafeArray<float4, 4> indexable = positions;
spvUnsafeArray<float4, 4> indexable_1 = positions2;
return indexable[Index1] + indexable_1[Index2];
}

3
reference/shaders-msl/comp/composite-array-initialization.comp
View File

@ -78,8 +78,7 @@ kernel void main0(device SSBO& _53 [[buffer(0)]], uint3 gl_WorkGroupID [[threadg
{
spvUnsafeArray<Data, 2> data = spvUnsafeArray<Data, 2>({ Data{ 1.0, 2.0 }, Data{ 3.0, 4.0 } });
spvUnsafeArray<Data, 2> _31 = spvUnsafeArray<Data, 2>({ Data{ X, 2.0 }, Data{ 3.0, 5.0 } });
spvUnsafeArray<Data, 2> data2;
data2 = _31;
spvUnsafeArray<Data, 2> data2 = _31;
Data param = data[gl_LocalInvocationID.x];
Data param_1 = data2[gl_LocalInvocationID.x];
Data _73 = combine(param, param_1);

3
reference/shaders-msl/comp/composite-construct.comp
View File

@ -67,8 +67,7 @@ constant spvUnsafeArray<float4, 2> _43 = spvUnsafeArray<float4, 2>({ float4(20.0
kernel void main0(device SSBO0& _16 [[buffer(0)]], device SSBO1& _32 [[buffer(1)]], uint3 gl_GlobalInvocationID [[thread_position_in_grid]], uint gl_LocalInvocationIndex [[thread_index_in_threadgroup]])
{
spvUnsafeArray<float4, 2> _37 = spvUnsafeArray<float4, 2>({ _16.as[gl_GlobalInvocationID.x], _32.bs[gl_GlobalInvocationID.x] });
spvUnsafeArray<float4, 2> values;
values = _37;
spvUnsafeArray<float4, 2> values = _37;
Composite c = Composite{ values[0], _43[1] };
_16.as[0] = values[gl_LocalInvocationIndex];
_32.bs[1] = c.b;

9
reference/shaders-msl/comp/copy-array-of-arrays.comp
View File

@ -60,8 +60,7 @@ constant spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> _21 = sp
kernel void main0(device BUF& o [[buffer(0)]])
{
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> c;
c = _21;
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> c = _21;
o.a = int(c[1][1][1]);
spvUnsafeArray<float, 2> _43 = spvUnsafeArray<float, 2>({ o.b, o.c });
spvUnsafeArray<float, 2> _48 = spvUnsafeArray<float, 2>({ o.b, o.b });
@ -70,10 +69,8 @@ kernel void main0(device BUF& o [[buffer(0)]])
spvUnsafeArray<float, 2> _59 = spvUnsafeArray<float, 2>({ o.c, o.b });
spvUnsafeArray<spvUnsafeArray<float, 2>, 2> _60 = spvUnsafeArray<spvUnsafeArray<float, 2>, 2>({ _54, _59 });
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> _61 = spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2>({ _49, _60 });
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> d;
d = _61;
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> e;
e = d;
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> d = _61;
spvUnsafeArray<spvUnsafeArray<spvUnsafeArray<float, 2>, 2>, 2> e = d;
o.b = e[1][0][1];
}

3
reference/shaders-msl/tesc/load-control-point-array-of-matrix.multi-patch.tesc
View File

@ -61,8 +61,7 @@ kernel void main0(uint3 gl_GlobalInvocationID [[thread_position_in_grid]], devic
uint gl_InvocationID = gl_GlobalInvocationID.x % 4;
uint gl_PrimitiveID = min(gl_GlobalInvocationID.x / 4, spvIndirectParams[1] - 1);
spvUnsafeArray<float4x4, 32> _16 = spvUnsafeArray<float4x4, 32>({ gl_in[0].vInputs, gl_in[1].vInputs, gl_in[2].vInputs, gl_in[3].vInputs, gl_in[4].vInputs, gl_in[5].vInputs, gl_in[6].vInputs, gl_in[7].vInputs, gl_in[8].vInputs, gl_in[9].vInputs, gl_in[10].vInputs, gl_in[11].vInputs, gl_in[12].vInputs, gl_in[13].vInputs, gl_in[14].vInputs, gl_in[15].vInputs, gl_in[16].vInputs, gl_in[17].vInputs, gl_in[18].vInputs, gl_in[19].vInputs, gl_in[20].vInputs, gl_in[21].vInputs, gl_in[22].vInputs, gl_in[23].vInputs, gl_in[24].vInputs, gl_in[25].vInputs, gl_in[26].vInputs, gl_in[27].vInputs, gl_in[28].vInputs, gl_in[29].vInputs, gl_in[30].vInputs, gl_in[31].vInputs });
spvUnsafeArray<float4x4, 32> tmp;
tmp = _16;
spvUnsafeArray<float4x4, 32> tmp = _16;
gl_out[gl_InvocationID].vOutputs = tmp[gl_InvocationID];
}

3
reference/shaders-msl/tesc/load-control-point-array-of-matrix.tesc
View File

@ -66,8 +66,7 @@ kernel void main0(main0_in in [[stage_in]], uint gl_InvocationID [[thread_index_
if (gl_InvocationID >= 4)
return;
spvUnsafeArray<float4x4, 32> _16 = spvUnsafeArray<float4x4, 32>({ float4x4(gl_in[0].vInputs_0, gl_in[0].vInputs_1, gl_in[0].vInputs_2, gl_in[0].vInputs_3), float4x4(gl_in[1].vInputs_0, gl_in[1].vInputs_1, gl_in[1].vInputs_2, gl_in[1].vInputs_3), float4x4(gl_in[2].vInputs_0, gl_in[2].vInputs_1, gl_in[2].vInputs_2, gl_in[2].vInputs_3), float4x4(gl_in[3].vInputs_0, gl_in[3].vInputs_1, gl_in[3].vInputs_2, gl_in[3].vInputs_3), float4x4(gl_in[4].vInputs_0, gl_in[4].vInputs_1, gl_in[4].vInputs_2, gl_in[4].vInputs_3), float4x4(gl_in[5].vInputs_0, gl_in[5].vInputs_1, gl_in[5].vInputs_2, gl_in[5].vInputs_3), float4x4(gl_in[6].vInputs_0, gl_in[6].vInputs_1, gl_in[6].vInputs_2, gl_in[6].vInputs_3), float4x4(gl_in[7].vInputs_0, gl_in[7].vInputs_1, gl_in[7].vInputs_2, gl_in[7].vInputs_3), float4x4(gl_in[8].vInputs_0, gl_in[8].vInputs_1, gl_in[8].vInputs_2, gl_in[8].vInputs_3), float4x4(gl_in[9].vInputs_0, gl_in[9].vInputs_1, gl_in[9].vInputs_2, gl_in[9].vInputs_3), float4x4(gl_in[10].vInputs_0, gl_in[10].vInputs_1, gl_in[10].vInputs_2, gl_in[10].vInputs_3), float4x4(gl_in[11].vInputs_0, gl_in[11].vInputs_1, gl_in[11].vInputs_2, gl_in[11].vInputs_3), float4x4(gl_in[12].vInputs_0, gl_in[12].vInputs_1, gl_in[12].vInputs_2, gl_in[12].vInputs_3), float4x4(gl_in[13].vInputs_0, gl_in[13].vInputs_1, gl_in[13].vInputs_2, gl_in[13].vInputs_3), float4x4(gl_in[14].vInputs_0, gl_in[14].vInputs_1, gl_in[14].vInputs_2, gl_in[14].vInputs_3), float4x4(gl_in[15].vInputs_0, gl_in[15].vInputs_1, gl_in[15].vInputs_2, gl_in[15].vInputs_3), float4x4(gl_in[16].vInputs_0, gl_in[16].vInputs_1, gl_in[16].vInputs_2, gl_in[16].vInputs_3), float4x4(gl_in[17].vInputs_0, gl_in[17].vInputs_1, gl_in[17].vInputs_2, gl_in[17].vInputs_3), float4x4(gl_in[18].vInputs_0, gl_in[18].vInputs_1, gl_in[18].vInputs_2, gl_in[18].vInputs_3), float4x4(gl_in[19].vInputs_0, gl_in[19].vInputs_1, gl_in[19].vInputs_2, gl_in[19].vInputs_3), float4x4(gl_in[20].vInputs_0, gl_in[20].vInputs_1, gl_in[20].vInputs_2, gl_in[20].vInputs_3), float4x4(gl_in[21].vInputs_0, gl_in[21].vInputs_1, gl_in[21].vInputs_2, gl_in[21].vInputs_3), float4x4(gl_in[22].vInputs_0, gl_in[22].vInputs_1, gl_in[22].vInputs_2, gl_in[22].vInputs_3), float4x4(gl_in[23].vInputs_0, gl_in[23].vInputs_1, gl_in[23].vInputs_2, gl_in[23].vInputs_3), float4x4(gl_in[24].vInputs_0, gl_in[24].vInputs_1, gl_in[24].vInputs_2, gl_in[24].vInputs_3), float4x4(gl_in[25].vInputs_0, gl_in[25].vInputs_1, gl_in[25].vInputs_2, gl_in[25].vInputs_3), float4x4(gl_in[26].vInputs_0, gl_in[26].vInputs_1, gl_in[26].vInputs_2, gl_in[26].vInputs_3), float4x4(gl_in[27].vInputs_0, gl_in[27].vInputs_1, gl_in[27].vInputs_2, gl_in[27].vInputs_3), float4x4(gl_in[28].vInputs_0, gl_in[28].vInputs_1, gl_in[28].vInputs_2, gl_in[28].vInputs_3), float4x4(gl_in[29].vInputs_0, gl_in[29].vInputs_1, gl_in[29].vInputs_2, gl_in[29].vInputs_3), float4x4(gl_in[30].vInputs_0, gl_in[30].vInputs_1, gl_in[30].vInputs_2, gl_in[30].vInputs_3), float4x4(gl_in[31].vInputs_0, gl_in[31].vInputs_1, gl_in[31].vInputs_2, gl_in[31].vInputs_3) });
spvUnsafeArray<float4x4, 32> tmp;
tmp = _16;
spvUnsafeArray<float4x4, 32> tmp = _16;
gl_out[gl_InvocationID].vOutputs = tmp[gl_InvocationID];
}

3
reference/shaders-msl/tesc/load-control-point-array-of-struct.multi-patch.tesc
View File

@ -68,8 +68,7 @@ kernel void main0(uint3 gl_GlobalInvocationID [[thread_position_in_grid]], devic
uint gl_InvocationID = gl_GlobalInvocationID.x % 4;
uint gl_PrimitiveID = min(gl_GlobalInvocationID.x / 4, spvIndirectParams[1] - 1);
spvUnsafeArray<VertexData, 32> _19 = spvUnsafeArray<VertexData, 32>({ gl_in[0].vInputs, gl_in[1].vInputs, gl_in[2].vInputs, gl_in[3].vInputs, gl_in[4].vInputs, gl_in[5].vInputs, gl_in[6].vInputs, gl_in[7].vInputs, gl_in[8].vInputs, gl_in[9].vInputs, gl_in[10].vInputs, gl_in[11].vInputs, gl_in[12].vInputs, gl_in[13].vInputs, gl_in[14].vInputs, gl_in[15].vInputs, gl_in[16].vInputs, gl_in[17].vInputs, gl_in[18].vInputs, gl_in[19].vInputs, gl_in[20].vInputs, gl_in[21].vInputs, gl_in[22].vInputs, gl_in[23].vInputs, gl_in[24].vInputs, gl_in[25].vInputs, gl_in[26].vInputs, gl_in[27].vInputs, gl_in[28].vInputs, gl_in[29].vInputs, gl_in[30].vInputs, gl_in[31].vInputs });
spvUnsafeArray<VertexData, 32> tmp;
tmp = _19;
spvUnsafeArray<VertexData, 32> tmp = _19;
VertexData tmp_single = gl_in[gl_InvocationID ^ 1].vInputs;
gl_out[gl_InvocationID].vOutputs = ((tmp[gl_InvocationID].a[1] + tmp[gl_InvocationID].b[1]) + tmp[gl_InvocationID].c) + tmp_single.c;
}

3
reference/shaders-msl/tesc/load-control-point-array-of-struct.tesc
View File

File diff suppressed because one or more lines are too long

3
reference/shaders-msl/tesc/load-control-point-array.multi-patch.tesc
View File

@ -62,8 +62,7 @@ kernel void main0(uint3 gl_GlobalInvocationID [[thread_position_in_grid]], devic
uint gl_InvocationID = gl_GlobalInvocationID.x % 4;
uint gl_PrimitiveID = min(gl_GlobalInvocationID.x / 4, spvIndirectParams[1] - 1);
spvUnsafeArray<float4, 32> _15 = spvUnsafeArray<float4, 32>({ gl_in[0].vInputs, gl_in[1].vInputs, gl_in[2].vInputs, gl_in[3].vInputs, gl_in[4].vInputs, gl_in[5].vInputs, gl_in[6].vInputs, gl_in[7].vInputs, gl_in[8].vInputs, gl_in[9].vInputs, gl_in[10].vInputs, gl_in[11].vInputs, gl_in[12].vInputs, gl_in[13].vInputs, gl_in[14].vInputs, gl_in[15].vInputs, gl_in[16].vInputs, gl_in[17].vInputs, gl_in[18].vInputs, gl_in[19].vInputs, gl_in[20].vInputs, gl_in[21].vInputs, gl_in[22].vInputs, gl_in[23].vInputs, gl_in[24].vInputs, gl_in[25].vInputs, gl_in[26].vInputs, gl_in[27].vInputs, gl_in[28].vInputs, gl_in[29].vInputs, gl_in[30].vInputs, gl_in[31].vInputs });
spvUnsafeArray<float4, 32> tmp;
tmp = _15;
spvUnsafeArray<float4, 32> tmp = _15;
gl_out[gl_InvocationID].vOutputs = tmp[gl_InvocationID];
}

3
reference/shaders-msl/tesc/load-control-point-array.tesc
View File

@ -63,8 +63,7 @@ kernel void main0(main0_in in [[stage_in]], uint gl_InvocationID [[thread_index_
if (gl_InvocationID >= 4)
return;
spvUnsafeArray<float4, 32> _15 = spvUnsafeArray<float4, 32>({ gl_in[0].vInputs, gl_in[1].vInputs, gl_in[2].vInputs, gl_in[3].vInputs, gl_in[4].vInputs, gl_in[5].vInputs, gl_in[6].vInputs, gl_in[7].vInputs, gl_in[8].vInputs, gl_in[9].vInputs, gl_in[10].vInputs, gl_in[11].vInputs, gl_in[12].vInputs, gl_in[13].vInputs, gl_in[14].vInputs, gl_in[15].vInputs, gl_in[16].vInputs, gl_in[17].vInputs, gl_in[18].vInputs, gl_in[19].vInputs, gl_in[20].vInputs, gl_in[21].vInputs, gl_in[22].vInputs, gl_in[23].vInputs, gl_in[24].vInputs, gl_in[25].vInputs, gl_in[26].vInputs, gl_in[27].vInputs, gl_in[28].vInputs, gl_in[29].vInputs, gl_in[30].vInputs, gl_in[31].vInputs });
spvUnsafeArray<float4, 32> tmp;
tmp = _15;
spvUnsafeArray<float4, 32> tmp = _15;
gl_out[gl_InvocationID].vOutputs = tmp[gl_InvocationID];
}

3
reference/shaders-msl/tese/load-control-point-array-of-matrix.tese
View File

@ -76,8 +76,7 @@ struct main0_patchIn
vBoo[2] = patchIn.vBoo_2;
vBoo[3] = patchIn.vBoo_3;
spvUnsafeArray<float4x4, 32> _16 = spvUnsafeArray<float4x4, 32>({ float4x4(patchIn.gl_in[0].vInputs_0, patchIn.gl_in[0].vInputs_1, patchIn.gl_in[0].vInputs_2, patchIn.gl_in[0].vInputs_3), float4x4(patchIn.gl_in[1].vInputs_0, patchIn.gl_in[1].vInputs_1, patchIn.gl_in[1].vInputs_2, patchIn.gl_in[1].vInputs_3), float4x4(patchIn.gl_in[2].vInputs_0, patchIn.gl_in[2].vInputs_1, patchIn.gl_in[2].vInputs_2, patchIn.gl_in[2].vInputs_3), float4x4(patchIn.gl_in[3].vInputs_0, patchIn.gl_in[3].vInputs_1, patchIn.gl_in[3].vInputs_2, patchIn.gl_in[3].vInputs_3), float4x4(patchIn.gl_in[4].vInputs_0, patchIn.gl_in[4].vInputs_1, patchIn.gl_in[4].vInputs_2, patchIn.gl_in[4].vInputs_3), float4x4(patchIn.gl_in[5].vInputs_0, patchIn.gl_in[5].vInputs_1, patchIn.gl_in[5].vInputs_2, patchIn.gl_in[5].vInputs_3), float4x4(patchIn.gl_in[6].vInputs_0, patchIn.gl_in[6].vInputs_1, patchIn.gl_in[6].vInputs_2, patchIn.gl_in[6].vInputs_3), float4x4(patchIn.gl_in[7].vInputs_0, patchIn.gl_in[7].vInputs_1, patchIn.gl_in[7].vInputs_2, patchIn.gl_in[7].vInputs_3), float4x4(patchIn.gl_in[8].vInputs_0, patchIn.gl_in[8].vInputs_1, patchIn.gl_in[8].vInputs_2, patchIn.gl_in[8].vInputs_3), float4x4(patchIn.gl_in[9].vInputs_0, patchIn.gl_in[9].vInputs_1, patchIn.gl_in[9].vInputs_2, patchIn.gl_in[9].vInputs_3), float4x4(patchIn.gl_in[10].vInputs_0, patchIn.gl_in[10].vInputs_1, patchIn.gl_in[10].vInputs_2, patchIn.gl_in[10].vInputs_3), float4x4(patchIn.gl_in[11].vInputs_0, patchIn.gl_in[11].vInputs_1, patchIn.gl_in[11].vInputs_2, patchIn.gl_in[11].vInputs_3), float4x4(patchIn.gl_in[12].vInputs_0, patchIn.gl_in[12].vInputs_1, patchIn.gl_in[12].vInputs_2, patchIn.gl_in[12].vInputs_3), float4x4(patchIn.gl_in[13].vInputs_0, patchIn.gl_in[13].vInputs_1, patchIn.gl_in[13].vInputs_2, patchIn.gl_in[13].vInputs_3), float4x4(patchIn.gl_in[14].vInputs_0, patchIn.gl_in[14].vInputs_1, patchIn.gl_in[14].vInputs_2, patchIn.gl_in[14].vInputs_3), float4x4(patchIn.gl_in[15].vInputs_0, patchIn.gl_in[15].vInputs_1, patchIn.gl_in[15].vInputs_2, patchIn.gl_in[15].vInputs_3), float4x4(patchIn.gl_in[16].vInputs_0, patchIn.gl_in[16].vInputs_1, patchIn.gl_in[16].vInputs_2, patchIn.gl_in[16].vInputs_3), float4x4(patchIn.gl_in[17].vInputs_0, patchIn.gl_in[17].vInputs_1, patchIn.gl_in[17].vInputs_2, patchIn.gl_in[17].vInputs_3), float4x4(patchIn.gl_in[18].vInputs_0, patchIn.gl_in[18].vInputs_1, patchIn.gl_in[18].vInputs_2, patchIn.gl_in[18].vInputs_3), float4x4(patchIn.gl_in[19].vInputs_0, patchIn.gl_in[19].vInputs_1, patchIn.gl_in[19].vInputs_2, patchIn.gl_in[19].vInputs_3), float4x4(patchIn.gl_in[20].vInputs_0, patchIn.gl_in[20].vInputs_1, patchIn.gl_in[20].vInputs_2, patchIn.gl_in[20].vInputs_3), float4x4(patchIn.gl_in[21].vInputs_0, patchIn.gl_in[21].vInputs_1, patchIn.gl_in[21].vInputs_2, patchIn.gl_in[21].vInputs_3), float4x4(patchIn.gl_in[22].vInputs_0, patchIn.gl_in[22].vInputs_1, patchIn.gl_in[22].vInputs_2, patchIn.gl_in[22].vInputs_3), float4x4(patchIn.gl_in[23].vInputs_0, patchIn.gl_in[23].vInputs_1, patchIn.gl_in[23].vInputs_2, patchIn.gl_in[23].vInputs_3), float4x4(patchIn.gl_in[24].vInputs_0, patchIn.gl_in[24].vInputs_1, patchIn.gl_in[24].vInputs_2, patchIn.gl_in[24].vInputs_3), float4x4(patchIn.gl_in[25].vInputs_0, patchIn.gl_in[25].vInputs_1, patchIn.gl_in[25].vInputs_2, patchIn.gl_in[25].vInputs_3), float4x4(patchIn.gl_in[26].vInputs_0, patchIn.gl_in[26].vInputs_1, patchIn.gl_in[26].vInputs_2, patchIn.gl_in[26].vInputs_3), float4x4(patchIn.gl_in[27].vInputs_0, patchIn.gl_in[27].vInputs_1, patchIn.gl_in[27].vInputs_2, patchIn.gl_in[27].vInputs_3), float4x4(patchIn.gl_in[28].vInputs_0, patchIn.gl_in[28].vInputs_1, patchIn.gl_in[28].vInputs_2, patchIn.gl_in[28].vInputs_3), float4x4(patchIn.gl_in[29].vInputs_0, patchIn.gl_in[29].vInputs_1, patchIn.gl_in[29].vInputs_2, patchIn.gl_in[29].vInputs_3), float4x4(patchIn.gl_in[30].vInputs_0, patchIn.gl_in[30].vInputs_1, patchIn.gl_in[30].vInputs_2, patchIn.gl_in[30].vInputs_3), float4x4(patchIn.gl_in[31].vInputs_0, patchIn.gl_in[31].vInputs_1, patchIn.gl_in[31].vInputs_2, patchIn.gl_in[31].vInputs_3) });
spvUnsafeArray<float4x4, 32> tmp;
tmp = _16;
spvUnsafeArray<float4x4, 32> tmp = _16;
out.gl_Position = (tmp[0][patchIn.vIndex] + tmp[1][patchIn.vIndex]) + vBoo[patchIn.vIndex];
return out;
}

3
reference/shaders-msl/tese/load-control-point-array.tese
View File

@ -73,8 +73,7 @@ struct main0_patchIn
vBoo[2] = patchIn.vBoo_2;
vBoo[3] = patchIn.vBoo_3;
spvUnsafeArray<float4, 32> _15 = spvUnsafeArray<float4, 32>({ patchIn.gl_in[0].vInputs, patchIn.gl_in[1].vInputs, patchIn.gl_in[2].vInputs, patchIn.gl_in[3].vInputs, patchIn.gl_in[4].vInputs, patchIn.gl_in[5].vInputs, patchIn.gl_in[6].vInputs, patchIn.gl_in[7].vInputs, patchIn.gl_in[8].vInputs, patchIn.gl_in[9].vInputs, patchIn.gl_in[10].vInputs, patchIn.gl_in[11].vInputs, patchIn.gl_in[12].vInputs, patchIn.gl_in[13].vInputs, patchIn.gl_in[14].vInputs, patchIn.gl_in[15].vInputs, patchIn.gl_in[16].vInputs, patchIn.gl_in[17].vInputs, patchIn.gl_in[18].vInputs, patchIn.gl_in[19].vInputs, patchIn.gl_in[20].vInputs, patchIn.gl_in[21].vInputs, patchIn.gl_in[22].vInputs, patchIn.gl_in[23].vInputs, patchIn.gl_in[24].vInputs, patchIn.gl_in[25].vInputs, patchIn.gl_in[26].vInputs, patchIn.gl_in[27].vInputs, patchIn.gl_in[28].vInputs, patchIn.gl_in[29].vInputs, patchIn.gl_in[30].vInputs, patchIn.gl_in[31].vInputs });
spvUnsafeArray<float4, 32> tmp;
tmp = _15;
spvUnsafeArray<float4, 32> tmp = _15;
out.gl_Position = (tmp[0] + tmp[1]) + vBoo[patchIn.vIndex];
return out;
}

12
reference/shaders-ue4-no-opt/asm/tese/ds-texcoord-array.asm.tese
View File

@ -277,10 +277,8 @@ struct main0_patchIn
float4 _113 = float4(gl_TessCoord.y);
float4 _116 = float4(gl_TessCoord.z);
float4 _118 = ((_97[0] * _111) + (_97[1] * _113)) + (_97[2] * _116);
spvUnsafeArray<float4, 1> _72;
_72 = _79[0];
spvUnsafeArray<float4, 1> _71;
_71 = _79[1];
spvUnsafeArray<float4, 1> _72 = _79[0];
spvUnsafeArray<float4, 1> _71 = _79[1];
float3 _120 = float3(gl_TessCoord.x);
float3 _123 = float3(gl_TessCoord.y);
spvUnsafeArray<float4, 1> _73;
@ -290,10 +288,8 @@ struct main0_patchIn
_133++;
continue;
}
spvUnsafeArray<float4, 1> _75;
_75 = _73;
spvUnsafeArray<float4, 1> _74;
_74 = _79[2];
spvUnsafeArray<float4, 1> _75 = _73;
spvUnsafeArray<float4, 1> _74 = _79[2];
float3 _155 = float3(gl_TessCoord.z);
float3 _157 = ((_77[0].xyz * _120) + (_77[1].xyz * _123)).xyz + (_77[2].xyz * _155);
spvUnsafeArray<float4, 1> _76;

3
reference/shaders-ue4/asm/tesc/hs-incorrect-base-type.invalid.asm.tesc
View File

@ -317,8 +317,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_View& View [[buffer(0)
spvUnsafeArray<uint, 12> _150 = spvUnsafeArray<uint, 12>({ gl_in[0].in_var_LIGHTMAP_ID, gl_in[1].in_var_LIGHTMAP_ID, gl_in[2].in_var_LIGHTMAP_ID, gl_in[3].in_var_LIGHTMAP_ID, gl_in[4].in_var_LIGHTMAP_ID, gl_in[5].in_var_LIGHTMAP_ID, gl_in[6].in_var_LIGHTMAP_ID, gl_in[7].in_var_LIGHTMAP_ID, gl_in[8].in_var_LIGHTMAP_ID, gl_in[9].in_var_LIGHTMAP_ID, gl_in[10].in_var_LIGHTMAP_ID, gl_in[11].in_var_LIGHTMAP_ID });
spvUnsafeArray<float4, 12> _259 = spvUnsafeArray<float4, 12>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position, gl_in[3].in_var_VS_To_DS_Position, gl_in[4].in_var_VS_To_DS_Position, gl_in[5].in_var_VS_To_DS_Position, gl_in[6].in_var_VS_To_DS_Position, gl_in[7].in_var_VS_To_DS_Position, gl_in[8].in_var_VS_To_DS_Position, gl_in[9].in_var_VS_To_DS_Position, gl_in[10].in_var_VS_To_DS_Position, gl_in[11].in_var_VS_To_DS_Position });
spvUnsafeArray<FBasePassVSToDS, 12> _284 = spvUnsafeArray<FBasePassVSToDS, 12>({ FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[0], _145[0], _146[0], _147[0], _148[0], _149[0], _150[0] } }, FBasePassInterpolantsVSToDS{ { } }, _259[0] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[1], _145[1], _146[1], _147[1], _148[1], _149[1], _150[1] } }, FBasePassInterpolantsVSToDS{ { } }, _259[1] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[2], _145[2], _146[2], _147[2], _148[2], _149[2], _150[2] } }, FBasePassInterpolantsVSToDS{ { } }, _259[2] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[3], _145[3], _146[3], _147[3], _148[3], _149[3], _150[3] } }, FBasePassInterpolantsVSToDS{ { } }, _259[3] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[4], _145[4], _146[4], _147[4], _148[4], _149[4], _150[4] } }, FBasePassInterpolantsVSToDS{ { } }, _259[4] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[5], _145[5], _146[5], _147[5], _148[5], _149[5], _150[5] } }, FBasePassInterpolantsVSToDS{ { } }, _259[5] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[6], _145[6], _146[6], _147[6], _148[6], _149[6], _150[6] } }, FBasePassInterpolantsVSToDS{ { } }, _259[6] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[7], _145[7], _146[7], _147[7], _148[7], _149[7], _150[7] } }, FBasePassInterpolantsVSToDS{ { } }, _259[7] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[8], _145[8], _146[8], _147[8], _148[8], _149[8], _150[8] } }, FBasePassInterpolantsVSToDS{ { } }, _259[8] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[9], _145[9], _146[9], _147[9], _148[9], _149[9], _150[9] } }, FBasePassInterpolantsVSToDS{ { } }, _259[9] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[10], _145[10], _146[10], _147[10], _148[10], _149[10], _150[10] } }, FBasePassInterpolantsVSToDS{ { } }, _259[10] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _144[11], _145[11], _146[11], _147[11], _148[11], _149[11], _150[11] } }, FBasePassInterpolantsVSToDS{ { } }, _259[11] } });
spvUnsafeArray<FBasePassVSToDS, 12> param_var_I;
param_var_I = _284;
spvUnsafeArray<FBasePassVSToDS, 12> param_var_I = _284;
float4 _301 = float4(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, 0.0);
float3 _310 = View_PrimitiveSceneData._m0[(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.PrimitiveId * 26u) + 22u].xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
uint _313 = (gl_InvocationID < 2u) ? (gl_InvocationID + 1u) : 0u;

3
reference/shaders-ue4/asm/tesc/hs-input-array-access.invalid.asm.tesc
View File

@ -355,8 +355,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_View& View [[buffer(0)
spvUnsafeArray<float4, 12> _192 = spvUnsafeArray<float4, 12>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position, gl_in[3].in_var_VS_To_DS_Position, gl_in[4].in_var_VS_To_DS_Position, gl_in[5].in_var_VS_To_DS_Position, gl_in[6].in_var_VS_To_DS_Position, gl_in[7].in_var_VS_To_DS_Position, gl_in[8].in_var_VS_To_DS_Position, gl_in[9].in_var_VS_To_DS_Position, gl_in[10].in_var_VS_To_DS_Position, gl_in[11].in_var_VS_To_DS_Position });
spvUnsafeArray<uint, 12> _193 = spvUnsafeArray<uint, 12>({ gl_in[0].in_var_VS_To_DS_VertexID, gl_in[1].in_var_VS_To_DS_VertexID, gl_in[2].in_var_VS_To_DS_VertexID, gl_in[3].in_var_VS_To_DS_VertexID, gl_in[4].in_var_VS_To_DS_VertexID, gl_in[5].in_var_VS_To_DS_VertexID, gl_in[6].in_var_VS_To_DS_VertexID, gl_in[7].in_var_VS_To_DS_VertexID, gl_in[8].in_var_VS_To_DS_VertexID, gl_in[9].in_var_VS_To_DS_VertexID, gl_in[10].in_var_VS_To_DS_VertexID, gl_in[11].in_var_VS_To_DS_VertexID });
spvUnsafeArray<FHitProxyVSToDS, 12> _230 = spvUnsafeArray<FHitProxyVSToDS, 12>({ FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[0], _143[0] } }, _192[0], _193[0] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[1], _143[1] } }, _192[1], _193[1] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[2], _143[2] } }, _192[2], _193[2] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[3], _143[3] } }, _192[3], _193[3] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[4], _143[4] } }, _192[4], _193[4] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[5], _143[5] } }, _192[5], _193[5] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[6], _143[6] } }, _192[6], _193[6] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[7], _143[7] } }, _192[7], _193[7] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[8], _143[8] } }, _192[8], _193[8] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[9], _143[9] } }, _192[9], _193[9] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[10], _143[10] } }, _192[10], _193[10] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _142[11], _143[11] } }, _192[11], _193[11] } });
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I;
param_var_I = _230;
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I = _230;
float4 _247 = float4(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, 0.0);
float3 _251 = Primitive.Primitive_NonUniformScale.xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
uint _254 = (gl_InvocationID < 2u) ? (gl_InvocationID + 1u) : 0u;

3
reference/shaders-ue4/asm/tesc/hs-texcoord-array.invalid.asm.tesc
View File

@ -332,8 +332,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_View& View [[buffer(0)
spvUnsafeArray<spvUnsafeArray<float2, 2>, 12> _132 = spvUnsafeArray<spvUnsafeArray<float2, 2>, 12>({ spvUnsafeArray<float2, 2>({ gl_in[0].in_var_TEXCOORD0_0, gl_in[0].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[1].in_var_TEXCOORD0_0, gl_in[1].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[2].in_var_TEXCOORD0_0, gl_in[2].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[3].in_var_TEXCOORD0_0, gl_in[3].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[4].in_var_TEXCOORD0_0, gl_in[4].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[5].in_var_TEXCOORD0_0, gl_in[5].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[6].in_var_TEXCOORD0_0, gl_in[6].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[7].in_var_TEXCOORD0_0, gl_in[7].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[8].in_var_TEXCOORD0_0, gl_in[8].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[9].in_var_TEXCOORD0_0, gl_in[9].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[10].in_var_TEXCOORD0_0, gl_in[10].in_var_TEXCOORD0_1 }), spvUnsafeArray<float2, 2>({ gl_in[11].in_var_TEXCOORD0_0, gl_in[11].in_var_TEXCOORD0_1 }) });
spvUnsafeArray<float4, 12> _205 = spvUnsafeArray<float4, 12>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position, gl_in[3].in_var_VS_To_DS_Position, gl_in[4].in_var_VS_To_DS_Position, gl_in[5].in_var_VS_To_DS_Position, gl_in[6].in_var_VS_To_DS_Position, gl_in[7].in_var_VS_To_DS_Position, gl_in[8].in_var_VS_To_DS_Position, gl_in[9].in_var_VS_To_DS_Position, gl_in[10].in_var_VS_To_DS_Position, gl_in[11].in_var_VS_To_DS_Position });
spvUnsafeArray<FHitProxyVSToDS, 12> _230 = spvUnsafeArray<FHitProxyVSToDS, 12>({ FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[0], _130[0], _131[0], _132[0] } }, _205[0] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[1], _130[1], _131[1], _132[1] } }, _205[1] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[2], _130[2], _131[2], _132[2] } }, _205[2] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[3], _130[3], _131[3], _132[3] } }, _205[3] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[4], _130[4], _131[4], _132[4] } }, _205[4] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[5], _130[5], _131[5], _132[5] } }, _205[5] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[6], _130[6], _131[6], _132[6] } }, _205[6] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[7], _130[7], _131[7], _132[7] } }, _205[7] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[8], _130[8], _131[8], _132[8] } }, _205[8] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[9], _130[9], _131[9], _132[9] } }, _205[9] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[10], _130[10], _131[10], _132[10] } }, _205[10] }, FHitProxyVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _129[11], _130[11], _131[11], _132[11] } }, _205[11] } });
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I;
param_var_I = _230;
spvUnsafeArray<FHitProxyVSToDS, 12> param_var_I = _230;
float4 _247 = float4(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, 0.0);
float3 _251 = Primitive.Primitive_NonUniformScale.xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
uint _254 = (gl_InvocationID < 2u) ? (gl_InvocationID + 1u) : 0u;

3
reference/shaders-ue4/asm/tesc/tess-factor-must-be-threadgroup.invalid.asm.tesc
View File

@ -147,8 +147,7 @@ kernel void main0(main0_in in [[stage_in]], constant type_Primitive& Primitive [
spvUnsafeArray<float4, 3> _91 = spvUnsafeArray<float4, 3>({ gl_in[0].in_var_TEXCOORD11_centroid, gl_in[1].in_var_TEXCOORD11_centroid, gl_in[2].in_var_TEXCOORD11_centroid });
spvUnsafeArray<float4, 3> _104 = spvUnsafeArray<float4, 3>({ gl_in[0].in_var_VS_To_DS_Position, gl_in[1].in_var_VS_To_DS_Position, gl_in[2].in_var_VS_To_DS_Position });
spvUnsafeArray<FBasePassVSToDS, 3> _111 = spvUnsafeArray<FBasePassVSToDS, 3>({ FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _90[0], _91[0] } }, FBasePassInterpolantsVSToDS{ { } }, _104[0] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _90[1], _91[1] } }, FBasePassInterpolantsVSToDS{ { } }, _104[1] }, FBasePassVSToDS{ FVertexFactoryInterpolantsVSToDS{ FVertexFactoryInterpolantsVSToPS{ _90[2], _91[2] } }, FBasePassInterpolantsVSToDS{ { } }, _104[2] } });
spvUnsafeArray<FBasePassVSToDS, 3> param_var_I;
param_var_I = _111;
spvUnsafeArray<FBasePassVSToDS, 3> param_var_I = _111;
float3 _128 = Primitive.Primitive_NonUniformScale.xyz * float3x3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz, cross(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz, param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0.xyz) * float3(param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.w), param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2.xyz);
gl_out[gl_InvocationID].out_var_TEXCOORD10_centroid = param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld0;
gl_out[gl_InvocationID].out_var_TEXCOORD11_centroid = param_var_I[gl_InvocationID].FactoryInterpolants.InterpolantsVSToPS.TangentToWorld2;

58
shaders-msl-no-opt/frag/struct-bool-edge-cases.frag Normal file
View File

@ -0,0 +1,58 @@
#version 450
layout(location = 0) out float FragColor;
struct Test
{
bool a;
bvec2 b;
bvec3 c;
bvec4 d;
};
struct Test2
{
bool a[2];
bvec2 b[2];
bvec3 c[2];
bvec4 d[2];
Test e;
};
void main()
{
Test t;
Test2 t2;
t.a = true;
t.b = bvec2(true, false);
t.c = bvec3(true, false, true);
t.d = bvec4(true, false, true, false);
t2.a = bool[](true, false);
t2.b = bvec2[](bvec2(true, false), bvec2(false, true));
t2.c = bvec3[](bvec3(true, false, true), bvec3(false, true, false));
t2.d = bvec4[](bvec4(true, false, true, false), bvec4(false, true, false, true));
bool a = t.a;
bvec2 b = t.b;
bvec3 c = t.c;
bvec4 d = t.d;
bool a2[2] = t2.a;
bvec2 b2[2] = t2.b;
bvec3 c2[2] = t2.c;
bvec4 d2[2] = t2.d;
t = Test(true, bvec2(true, false), bvec3(true), bvec4(false));
t2 = Test2(
bool[](true, true),
bvec2[](bvec2(true), bvec2(false)),
bvec3[](bvec3(true), bvec3(false)),
bvec4[](bvec4(true), bvec4(false)),
Test(true, bvec2(true, false), bvec3(true), bvec4(false)));
Test t3[2] = Test[](
Test(true, bvec2(true, false), bvec3(true), bvec4(false)),
Test(false, bvec2(false, true), bvec3(false), bvec4(true)));
}

140
spirv_glsl.cpp
View File

@ -5102,8 +5102,27 @@ string CompilerGLSL::to_extract_constant_composite_expression(uint32_t result_ty
return constant_expression(tmp);
}
string CompilerGLSL::to_rerolled_array_expression(const string &base_expr, const SPIRType &type)
string CompilerGLSL::to_rerolled_array_expression(const SPIRType &parent_type,
const string &base_expr, const SPIRType &type)
{
bool remapped_boolean = parent_type.basetype == SPIRType::Struct &&
type.basetype == SPIRType::Boolean &&
backend.boolean_in_struct_remapped_type != SPIRType::Boolean;
SPIRType tmp_type;
if (remapped_boolean)
{
tmp_type = get<SPIRType>(type.parent_type);
tmp_type.basetype = backend.boolean_in_struct_remapped_type;
}
else if (type.basetype == SPIRType::Boolean && backend.boolean_in_struct_remapped_type != SPIRType::Boolean)
{
// It's possible that we have an r-value expression that was OpLoaded from a struct.
// We have to reroll this and explicitly cast the input to bool, because the r-value is short.
tmp_type = get<SPIRType>(type.parent_type);
remapped_boolean = true;
}
uint32_t size = to_array_size_literal(type);
auto &parent = get<SPIRType>(type.parent_type);
string expr = "{ ";
@ -5111,10 +5130,14 @@ string CompilerGLSL::to_rerolled_array_expression(const string &base_expr, const
for (uint32_t i = 0; i < size; i++)
{
auto subexpr = join(base_expr, "[", convert_to_string(i), "]");
if (parent.array.empty())
if (!type_is_top_level_array(parent))
{
if (remapped_boolean)
subexpr = join(type_to_glsl(tmp_type), "(", subexpr, ")");
expr += subexpr;
}
else
expr += to_rerolled_array_expression(subexpr, parent);
expr += to_rerolled_array_expression(parent_type, subexpr, parent);
if (i + 1 < size)
expr += ", ";
@ -5124,13 +5147,26 @@ string CompilerGLSL::to_rerolled_array_expression(const string &base_expr, const
return expr;
}
string CompilerGLSL::to_composite_constructor_expression(uint32_t id, bool block_like_type)
string CompilerGLSL::to_composite_constructor_expression(const SPIRType &parent_type, uint32_t id, bool block_like_type)
{
auto &type = expression_type(id);
bool reroll_array = !type.array.empty() &&
(!backend.array_is_value_type ||
(block_like_type && !backend.array_is_value_type_in_buffer_blocks));
bool reroll_array = false;
bool remapped_boolean = parent_type.basetype == SPIRType::Struct &&
type.basetype == SPIRType::Boolean &&
backend.boolean_in_struct_remapped_type != SPIRType::Boolean;
if (type_is_top_level_array(type))
{
reroll_array = !backend.array_is_value_type ||
(block_like_type && !backend.array_is_value_type_in_buffer_blocks);
if (remapped_boolean)
{
// Forced to reroll if we have to change bool[] to short[].
reroll_array = true;
}
}
if (reroll_array)
{
@ -5144,10 +5180,20 @@ string CompilerGLSL::to_composite_constructor_expression(uint32_t id, bool block
// We're only triggering one read of the array expression, but this is fine since arrays have to be declared
// as temporaries anyways.
return to_rerolled_array_expression(to_enclosed_expression(id), type);
return to_rerolled_array_expression(parent_type, to_enclosed_expression(id), type);
}
else
return to_unpacked_expression(id);
{
auto expr = to_unpacked_expression(id);
if (remapped_boolean)
{
auto tmp_type = type;
tmp_type.basetype = backend.boolean_in_struct_remapped_type;
expr = join(type_to_glsl(tmp_type), "(", expr, ")");
}
return expr;
}
}
string CompilerGLSL::to_non_uniform_aware_expression(uint32_t id)
@ -5657,11 +5703,13 @@ string CompilerGLSL::constant_op_expression(const SPIRConstantOp &cop)
}
}
string CompilerGLSL::constant_expression(const SPIRConstant &c, bool inside_block_like_struct_scope)
string CompilerGLSL::constant_expression(const SPIRConstant &c,
bool inside_block_like_struct_scope,
bool inside_struct_scope)
{
auto &type = get<SPIRType>(c.constant_type);
if (type.pointer)
if (type_is_top_level_pointer(type))
{
return backend.null_pointer_literal;
}
@ -5676,19 +5724,32 @@ string CompilerGLSL::constant_expression(const SPIRConstant &c, bool inside_bloc
// with Offset = 0, using no ArrayStride on the enclosed array type.
// A particular CTS test hits this scenario.
bool array_type_decays = inside_block_like_struct_scope &&
!type.array.empty() && !backend.array_is_value_type_in_buffer_blocks;
type_is_top_level_array(type) &&
!backend.array_is_value_type_in_buffer_blocks;
// Allow Metal to use the array<T> template to make arrays a value type
bool needs_trailing_tracket = false;
if (backend.use_initializer_list && backend.use_typed_initializer_list && type.basetype == SPIRType::Struct &&
type.array.empty())
!type_is_top_level_array(type))
{
res = type_to_glsl_constructor(type) + "{ ";
}
else if (backend.use_initializer_list && backend.use_typed_initializer_list && backend.array_is_value_type &&
!type.array.empty() && !array_type_decays)
type_is_top_level_array(type) && !array_type_decays)
{
res = type_to_glsl_constructor(type) + "({ ";
const auto *p_type = &type;
SPIRType tmp_type;
if (inside_struct_scope &&
backend.boolean_in_struct_remapped_type != SPIRType::Boolean &&
type.basetype == SPIRType::Boolean)
{
tmp_type = type;
tmp_type.basetype = backend.boolean_in_struct_remapped_type;
p_type = &tmp_type;
}
res = type_to_glsl_constructor(*p_type) + "({ ";
needs_trailing_tracket = true;
}
else if (backend.use_initializer_list)
@ -5718,7 +5779,7 @@ string CompilerGLSL::constant_expression(const SPIRConstant &c, bool inside_bloc
res += to_name(elem);
else
{
if (type.array.empty() && type.basetype == SPIRType::Struct)
if (!type_is_top_level_array(type) && type.basetype == SPIRType::Struct)
{
// When we get down to emitting struct members, override the block-like information.
// For constants, we can freely mix and match block-like state.
@ -5726,7 +5787,10 @@ string CompilerGLSL::constant_expression(const SPIRConstant &c, bool inside_bloc
has_member_decoration(type.self, subconstant_index, DecorationOffset);
}
res += constant_expression(subc, inside_block_like_struct_scope);
if (type.basetype == SPIRType::Struct)
inside_struct_scope = true;
res += constant_expression(subc, inside_block_like_struct_scope, inside_struct_scope);
}
}
@ -5748,19 +5812,30 @@ string CompilerGLSL::constant_expression(const SPIRConstant &c, bool inside_bloc
if (backend.supports_empty_struct)
return "{ }";
else if (backend.use_typed_initializer_list)
return join(type_to_glsl(get<SPIRType>(c.constant_type)), "{ 0 }");
return join(type_to_glsl(type), "{ 0 }");
else if (backend.use_initializer_list)
return "{ 0 }";
else
return join(type_to_glsl(get<SPIRType>(c.constant_type)), "(0)");
return join(type_to_glsl(type), "(0)");
}
else if (c.columns() == 1)
{
return constant_expression_vector(c, 0);
auto res = constant_expression_vector(c, 0);
if (inside_struct_scope &&
backend.boolean_in_struct_remapped_type != SPIRType::Boolean &&
type.basetype == SPIRType::Boolean)
{
SPIRType tmp_type = type;
tmp_type.basetype = backend.boolean_in_struct_remapped_type;
res = join(type_to_glsl(tmp_type), "(", res, ")");
}
return res;
}
else
{
string res = type_to_glsl(get<SPIRType>(c.constant_type)) + "(";
string res = type_to_glsl(type) + "(";
for (uint32_t col = 0; col < c.columns(); col++)
{
if (c.specialization_constant_id(col) != 0)
@ -5772,6 +5847,16 @@ string CompilerGLSL::constant_expression(const SPIRConstant &c, bool inside_bloc
res += ", ";
}
res += ")";
if (inside_struct_scope &&
backend.boolean_in_struct_remapped_type != SPIRType::Boolean &&
type.basetype == SPIRType::Boolean)
{
SPIRType tmp_type = type;
tmp_type.basetype = backend.boolean_in_struct_remapped_type;
res = join(type_to_glsl(tmp_type), "(", res, ")");
}
return res;
}
}
@ -11107,7 +11192,7 @@ string CompilerGLSL::build_composite_combiner(uint32_t return_type, const uint32
bool uses_buffer_offset =
type.basetype == SPIRType::Struct && has_member_decoration(type.self, i, DecorationOffset);
subop = to_composite_constructor_expression(elems[i], uses_buffer_offset);
subop = to_composite_constructor_expression(type, elems[i], uses_buffer_offset);
}
base = e ? e->base_expression : ID(0);
@ -11737,7 +11822,7 @@ void CompilerGLSL::emit_instruction(const Instruction &instruction)
// it is an array, and our backend does not support arrays as value types.
// Emit the temporary, and copy it explicitly.
e = &emit_uninitialized_temporary_expression(result_type, id);
emit_array_copy(to_expression(id), id, ptr, StorageClassFunction, get_expression_effective_storage_class(ptr));
emit_array_copy(nullptr, id, ptr, StorageClassFunction, get_expression_effective_storage_class(ptr));
}
else
e = &emit_op(result_type, id, expr, forward, !usage_tracking);
@ -17343,9 +17428,16 @@ uint32_t CompilerGLSL::mask_relevant_memory_semantics(uint32_t semantics)
MemorySemanticsCrossWorkgroupMemoryMask | MemorySemanticsSubgroupMemoryMask);
}
void CompilerGLSL::emit_array_copy(const string &lhs, uint32_t, uint32_t rhs_id, StorageClass, StorageClass)
bool CompilerGLSL::emit_array_copy(const char *expr, uint32_t lhs_id, uint32_t rhs_id, StorageClass, StorageClass)
{
string lhs;
if (expr)
lhs = expr;
else
lhs = to_expression(lhs_id);
statement(lhs, " = ", to_expression(rhs_id), ";");
return true;
}
bool CompilerGLSL::unroll_array_to_complex_store(uint32_t target_id, uint32_t source_id)

11
spirv_glsl.hpp
View File

@ -430,7 +430,9 @@ protected:
const std::string &qualifier = "", uint32_t base_offset = 0);
virtual void emit_struct_padding_target(const SPIRType &type);
virtual std::string image_type_glsl(const SPIRType &type, uint32_t id = 0);
std::string constant_expression(const SPIRConstant &c, bool inside_block_like_struct_scope = false);
std::string constant_expression(const SPIRConstant &c,
bool inside_block_like_struct_scope = false,
bool inside_struct_scope = false);
virtual std::string constant_op_expression(const SPIRConstantOp &cop);
virtual std::string constant_expression_vector(const SPIRConstant &c, uint32_t vector);
virtual void emit_fixup();
@ -608,6 +610,7 @@ protected:
const char *uint16_t_literal_suffix = "us";
const char *nonuniform_qualifier = "nonuniformEXT";
const char *boolean_mix_function = "mix";
SPIRType::BaseType boolean_in_struct_remapped_type = SPIRType::Boolean;
bool swizzle_is_function = false;
bool shared_is_implied = false;
bool unsized_array_supported = true;
@ -776,8 +779,8 @@ protected:
void append_global_func_args(const SPIRFunction &func, uint32_t index, SmallVector<std::string> &arglist);
std::string to_non_uniform_aware_expression(uint32_t id);
std::string to_expression(uint32_t id, bool register_expression_read = true);
std::string to_composite_constructor_expression(uint32_t id, bool block_like_type);
std::string to_rerolled_array_expression(const std::string &expr, const SPIRType &type);
std::string to_composite_constructor_expression(const SPIRType &parent_type, uint32_t id, bool block_like_type);
std::string to_rerolled_array_expression(const SPIRType &parent_type, const std::string &expr, const SPIRType &type);
std::string to_enclosed_expression(uint32_t id, bool register_expression_read = true);
std::string to_unpacked_expression(uint32_t id, bool register_expression_read = true);
std::string to_unpacked_row_major_matrix_expression(uint32_t id);
@ -810,7 +813,7 @@ protected:
std::string layout_for_variable(const SPIRVariable &variable);
std::string to_combined_image_sampler(VariableID image_id, VariableID samp_id);
virtual bool skip_argument(uint32_t id) const;
virtual void emit_array_copy(const std::string &lhs, uint32_t lhs_id, uint32_t rhs_id,
virtual bool emit_array_copy(const char *expr, uint32_t lhs_id, uint32_t rhs_id,
spv::StorageClass lhs_storage, spv::StorageClass rhs_storage);
virtual void emit_block_hints(const SPIRBlock &block);
virtual std::string to_initializer_expression(const SPIRVariable &var);

73
spirv_msl.cpp
View File

@ -1467,6 +1467,7 @@ string CompilerMSL::compile()
backend.support_small_type_sampling_result = true;
backend.supports_empty_struct = true;
backend.support_64bit_switch = true;
backend.boolean_in_struct_remapped_type = SPIRType::Short;
// Allow Metal to use the array<T> template unless we force it off.
backend.can_return_array = !msl_options.force_native_arrays;
@ -7234,7 +7235,7 @@ void CompilerMSL::declare_constant_arrays()
// FIXME: However, hoisting constants to main() means we need to pass down constant arrays to leaf functions if they are used there.
// If there are multiple functions in the module, drop this case to avoid breaking use cases which do not need to
// link into Metal libraries. This is hacky.
if (!type.array.empty() && (!fully_inlined || is_scalar(type) || is_vector(type)))
if (type_is_top_level_array(type) && (!fully_inlined || is_scalar(type) || is_vector(type)))
{
add_resource_name(c.self);
auto name = to_name(c.self);
@ -7266,7 +7267,7 @@ void CompilerMSL::declare_complex_constant_arrays()
return;
auto &type = this->get<SPIRType>(c.constant_type);
if (!type.array.empty() && !(is_scalar(type) || is_vector(type)))
if (type_is_top_level_array(type) && !(is_scalar(type) || is_vector(type)))
{
add_resource_name(c.self);
auto name = to_name(c.self);
@ -9469,7 +9470,7 @@ static bool storage_class_array_is_thread(StorageClass storage)
}
}
void CompilerMSL::emit_array_copy(const string &lhs, uint32_t lhs_id, uint32_t rhs_id,
bool CompilerMSL::emit_array_copy(const char *expr, uint32_t lhs_id, uint32_t rhs_id,
StorageClass lhs_storage, StorageClass rhs_storage)
{
// Allow Metal to use the array<T> template to make arrays a value type.
@ -9508,10 +9509,21 @@ void CompilerMSL::emit_array_copy(const string &lhs, uint32_t lhs_id, uint32_t r
// Avoid spvCopy* wrapper functions; Otherwise, spvUnsafeArray<> template cannot be used with that storage qualifier.
if (lhs_is_array_template && rhs_is_array_template && !using_builtin_array())
{
statement(lhs, " = ", to_expression(rhs_id), ";");
// Fall back to normal copy path.
return false;
}
else
{
// Ensure the LHS variable has been declared
if (lhs_var)
flush_variable_declaration(lhs_var->self);
string lhs;
if (expr)
lhs = expr;
else
lhs = to_expression(lhs_id);
// Assignment from an array initializer is fine.
auto &type = expression_type(rhs_id);
auto *var = maybe_get_backing_variable(rhs_id);
@ -9585,6 +9597,8 @@ void CompilerMSL::emit_array_copy(const string &lhs, uint32_t lhs_id, uint32_t r
else
statement("spvArrayCopy", tag, type.array.size(), "(", lhs, ", ", to_expression(rhs_id), ");");
}
return true;
}
uint32_t CompilerMSL::get_physical_tess_level_array_size(spv::BuiltIn builtin) const
@ -9641,14 +9655,11 @@ bool CompilerMSL::maybe_emit_array_assignment(uint32_t id_lhs, uint32_t id_rhs)
}
}
// Ensure the LHS variable has been declared
auto *p_v_lhs = maybe_get_backing_variable(id_lhs);
if (p_v_lhs)
flush_variable_declaration(p_v_lhs->self);
auto lhs_storage = get_expression_effective_storage_class(id_lhs);
auto rhs_storage = get_expression_effective_storage_class(id_rhs);
emit_array_copy(to_expression(id_lhs), id_lhs, id_rhs, lhs_storage, rhs_storage);
if (!emit_array_copy(nullptr, id_lhs, id_rhs, lhs_storage, rhs_storage))
return false;
register_write(id_lhs);
return true;
@ -16784,20 +16795,31 @@ void CompilerMSL::cast_from_variable_load(uint32_t source_id, std::string &expr,
auto *source_expr = maybe_get<SPIRExpression>(source_id);
auto *var = maybe_get_backing_variable(source_id);
const SPIRType *var_type = nullptr, *phys_type = nullptr;
if (uint32_t phys_id = get_extended_decoration(source_id, SPIRVCrossDecorationPhysicalTypeID))
phys_type = &get<SPIRType>(phys_id);
else
phys_type = &expr_type;
if (var)
{
source_id = var->self;
var_type = &get_variable_data_type(*var);
}
bool rewrite_boolean_load =
expr_type.basetype == SPIRType::Boolean &&
(var && (var->storage == StorageClassWorkgroup || var_type->basetype == SPIRType::Struct));
// Type fixups for workgroup variables if they are booleans.
if (var && (var->storage == StorageClassWorkgroup || var_type->basetype == SPIRType::Struct) &&
expr_type.basetype == SPIRType::Boolean)
expr = join(type_to_glsl(expr_type), "(", expr, ")");
if (rewrite_boolean_load)
{
if (type_is_top_level_array(expr_type))
expr = to_rerolled_array_expression(expr_type, expr, expr_type);
else
expr = join(type_to_glsl(expr_type), "(", expr, ")");
}
// Type fixups for workgroup variables if they are matrices.
// Don't do fixup for packed types; those are handled specially.
// FIXME: Maybe use a type like spvStorageMatrix for packed matrices?
@ -16910,24 +16932,37 @@ void CompilerMSL::cast_to_variable_store(uint32_t target_id, std::string &expr,
auto *target_expr = maybe_get<SPIRExpression>(target_id);
auto *var = maybe_get_backing_variable(target_id);
const SPIRType *var_type = nullptr, *phys_type = nullptr;
if (uint32_t phys_id = get_extended_decoration(target_id, SPIRVCrossDecorationPhysicalTypeID))
phys_type = &get<SPIRType>(phys_id);
else
phys_type = &expr_type;
if (var)
{
target_id = var->self;
var_type = &get_variable_data_type(*var);
}
// Type fixups for workgroup variables if they are booleans.
if (var && (var->storage == StorageClassWorkgroup || var_type->basetype == SPIRType::Struct) &&
expr_type.basetype == SPIRType::Boolean)
bool rewrite_boolean_store =
expr_type.basetype == SPIRType::Boolean &&
(var && (var->storage == StorageClassWorkgroup || var_type->basetype == SPIRType::Struct));
// Type fixups for workgroup variables or struct members if they are booleans.
if (rewrite_boolean_store)
{
auto short_type = expr_type;
short_type.basetype = SPIRType::Short;
expr = join(type_to_glsl(short_type), "(", expr, ")");
if (type_is_top_level_array(expr_type))
{
expr = to_rerolled_array_expression(*var_type, expr, expr_type);
}
else
{
auto short_type = expr_type;
short_type.basetype = SPIRType::Short;
expr = join(type_to_glsl(short_type), "(", expr, ")");
}
}
// Type fixups for workgroup variables if they are matrices.
// Don't do fixup for packed types; those are handled specially.
// FIXME: Maybe use a type like spvStorageMatrix for packed matrices?

2
spirv_msl.hpp
View File

@ -1032,7 +1032,7 @@ protected:
void add_pragma_line(const std::string &line);
void add_typedef_line(const std::string &line);
void emit_barrier(uint32_t id_exe_scope, uint32_t id_mem_scope, uint32_t id_mem_sem);
void emit_array_copy(const std::string &lhs, uint32_t lhs_id, uint32_t rhs_id,
bool emit_array_copy(const char *expr, uint32_t lhs_id, uint32_t rhs_id,
spv::StorageClass lhs_storage, spv::StorageClass rhs_storage) override;
void build_implicit_builtins();
uint32_t build_constant_uint_array_pointer();