HLSL: add geometry stage support for clip/cull distance

Test/hlsl.clipdistance-1.geom

 struct S {
-    float4 pos   : SV_Position;
-    float  clip  : SV_ClipDistance0;
-    float  cull  : SV_CullDistance0;
+    float4 pos     : SV_Position;
+    float2 clip    : SV_ClipDistance0;
 };
 
 [maxvertexcount(3)]
 void main(triangle in float4 pos[3] : SV_Position, 
           triangle in uint VertexID[3] : VertexID,
           inout LineStream<S> OutputStream,
-          triangle in float clip[3] : SV_ClipDistance,  // scalar float
-          triangle in float cull[3] : SV_CullDistance)  // scalar float
+          triangle in float4 clip[3] : SV_ClipDistance)   // externally: an array 3 of array 4 (not vec4!) of float.
 {
     S s;
 
     s.pos = pos[0];
-    s.clip = clip[0];
-    s.cull = cull[0];
+    s.clip = clip[0].xy;
 
     OutputStream.Append(s);
 }

Test/hlsl.clipdistance-2.geom

@@ -7,7 +7,7 @@ struct S {
 void main(triangle in float4 pos[3] : SV_Position, 
           triangle in uint VertexID[3] : VertexID,
           inout LineStream<S> OutputStream,
-          triangle in float2 clip[3][2] : SV_ClipDistance)  // scalar float
+          triangle in float2 clip[3][2] : SV_ClipDistance) // externally: an array 3 of array 4 of float.
 {
     S s;
 

Test/hlsl.clipdistance-3.frag

@@ -4,5 +4,3 @@ float4 main(in float4 pos : SV_Position,
 {
     return pos + clip[0] + cull[0];
 }
-
-

Test/hlsl.clipdistance-3.geom

+struct S {
+    float4 pos     : SV_Position;
+    float2 clip0   : SV_ClipDistance0;  // clip0 and clip1 form an array of float[4] externally.
+    float2 clip1   : SV_ClipDistance1;
+};
+
+[maxvertexcount(3)]
+void main(triangle in float4 pos[3] : SV_Position, 
+          triangle in uint VertexID[3] : VertexID,
+          inout LineStream<S> OutputStream,
+          triangle in float4 clip[3] : SV_ClipDistance)   // externally: an array 3 of array 4 (not vec4!) of float.
+{
+    S s;
+
+    s.pos = pos[0];
+    s.clip0 = clip[0].xy;
+    s.clip1 = clip[0].zw;
+
+    OutputStream.Append(s);
+}

Test/hlsl.clipdistance-4.geom

+struct S {
+    float4 pos     : SV_Position;
+    float2 clip0   : SV_ClipDistance0;  // clip0 and clip1 form an array of float[4] externally.
+    float2 clip1   : SV_ClipDistance1;
+};
+
+[maxvertexcount(3)]
+void main(triangle in float4 pos[3] : SV_Position, 
+          triangle in uint VertexID[3] : VertexID,
+          inout LineStream<S> OutputStream,
+          triangle in float2 clip0[3] : SV_ClipDistance0,  // test input arrayed semantic vars
+          triangle in float2 clip1[3] : SV_ClipDistance1)
+{
+    S s;
+
+    s.pos = pos[0];
+    s.clip0 = clip0[0];
+    s.clip1 = clip1[0];
+
+    OutputStream.Append(s);
+}

gtests/Hlsl.FromFile.cpp

@@ -104,8 +104,10 @@ INSTANTIATE_TEST_CASE_P(
         {"hlsl.clipdistance-2.geom", "main"},
         {"hlsl.clipdistance-2.vert", "main"},
         {"hlsl.clipdistance-3.frag", "main"},
+        {"hlsl.clipdistance-3.geom", "main"},
         {"hlsl.clipdistance-3.vert", "main"},
         {"hlsl.clipdistance-4.frag", "main"},
+        {"hlsl.clipdistance-4.geom", "main"},
         {"hlsl.clipdistance-4.vert", "main"},
         {"hlsl.clipdistance-5.frag", "main"},
         {"hlsl.clipdistance-5.vert", "main"},

hlsl/hlslParseHelper.h

@@ -435,12 +435,16 @@ protected:
 
     TVariable* gsStreamOutput;               // geometry shader stream outputs, for emit (Append method)
 
-    TVariable* clipDistanceVariable;         // synthesized clip distance variable (shader might have >1)
-    TVariable* cullDistanceVariable;         // synthesized cull distance variable (shader might have >1)
+    TVariable* clipDistanceOutput;           // synthesized clip distance out variable (shader might have >1)
+    TVariable* cullDistanceOutput;           // synthesized cull distance out variable (shader might have >1)
+    TVariable* clipDistanceInput;            // synthesized clip distance in variable (shader might have >1)
+    TVariable* cullDistanceInput;            // synthesized cull distance in variable (shader might have >1)
 
     static const int maxClipCullRegs = 2;
-    std::array<int, maxClipCullRegs> clipSemanticNSize; // vector, indexed by clip semantic ID
-    std::array<int, maxClipCullRegs> cullSemanticNSize; // vector, indexed by cull semantic ID
+    std::array<int, maxClipCullRegs> clipSemanticNSizeIn;  // vector, indexed by clip semantic ID
+    std::array<int, maxClipCullRegs> cullSemanticNSizeIn;  // vector, indexed by cull semantic ID
+    std::array<int, maxClipCullRegs> clipSemanticNSizeOut; // vector, indexed by clip semantic ID
+    std::array<int, maxClipCullRegs> cullSemanticNSizeOut; // vector, indexed by cull semantic ID
 
     // This tracks the first (mip level) argument to the .mips[][] operator.  Since this can be nested as
     // in tx.mips[tx.mips[0][1].x][2], we need a stack.  We also track the TSourceLoc for error reporting