HLSL: Include shape-changing conversions in overloaded signature selection.

This also enables vecN -> vec1 shape conversions for all places doing shape
conversions.

For signature selection, makes shape changes worse than any other comparison
when deciding what conversions are better than others.

Test/baseResults/hlsl.intrinsics.negative.frag.out

@@ -1,12 +1,6 @@
 hlsl.intrinsics.negative.frag
-ERROR: 0:8: 'cross' : no matching overloaded function found 
-ERROR: 0:9: 'D3DCOLORtoUBYTE4' : no matching overloaded function found 
 ERROR: 0:10: 'determinant' : no matching overloaded function found 
 ERROR: 0:12: 'f32tof16' : unimplemented intrinsic: handle natively 
-ERROR: 0:23: 'length' : no matching overloaded function found 
-ERROR: 0:25: 'normalize' : no matching overloaded function found 
-ERROR: 0:26: 'reflect' : no matching overloaded function found 
-ERROR: 0:27: 'refract' : no matching overloaded function found 
 ERROR: 0:28: 'refract' : no matching overloaded function found 
 ERROR: 0:30: 'transpose' : no matching overloaded function found 
 ERROR: 0:39: 'GetRenderTargetSamplePosition' : no matching overloaded function found 
@@ -23,7 +17,6 @@ ERROR: 0:67: 'determinant' : no matching overloaded function found
 ERROR: 0:68: 'f32tof16' : unimplemented intrinsic: handle natively 
 ERROR: 0:73: 'transpose' : no matching overloaded function found 
 ERROR: 0:81: 'CheckAccessFullyMapped' : no matching overloaded function found 
-ERROR: 0:83: 'cross' : no matching overloaded function found 
 ERROR: 0:84: 'determinant' : no matching overloaded function found 
 ERROR: 0:85: 'f32tof16' : unimplemented intrinsic: handle natively 
 ERROR: 0:90: 'transpose' : no matching overloaded function found 
@@ -66,7 +59,7 @@ ERROR: 0:133: 'normalize' : no matching overloaded function found
 ERROR: 0:133: 'reflect' : no matching overloaded function found 
 ERROR: 0:133: 'refract' : no matching overloaded function found 
 ERROR: 0:133: 'reversebits' : no matching overloaded function found 
-ERROR: 67 compilation errors.  No code generated.
+ERROR: 60 compilation errors.  No code generated.
 
 
 Shader version: 450
@@ -91,10 +84,14 @@ ERROR: node is still EOpNull!
 0:7      bitCount (global uint)
 0:7        Convert float to uint (temp uint)
 0:7          'inF0' (in float)
-0:8      Constant:
-0:8        0.000000
-0:9      Constant:
-0:9        0.000000
+0:8      cross-product (global 3-component vector of float)
+0:8        Construct vec3 (in 3-component vector of float)
+0:8          'inF0' (in float)
+0:8        Construct vec3 (in 3-component vector of float)
+0:8          'inF1' (in float)
+0:9      Function Call: D3DCOLORtoUBYTE4(vf4; (global 4-component vector of int)
+0:9        Construct vec4 (in 4-component vector of float)
+0:9          'inF0' (in float)
 0:10      Constant:
 0:10        0.000000
 0:12      ERROR: Bad unary op
@@ -107,8 +104,9 @@ ERROR: node is still EOpNull!
 0:14      findLSB (global uint)
 0:14        Convert float to uint (temp uint)
 0:14          'inF0' (in float)
-0:23      Constant:
-0:23        0.000000
+0:23      length (global float)
+0:23        Construct vec2 (in 2-component vector of float)
+0:23          'inF0' (in float)
 0:24      Function Call: msad4(u1;vu2;vu4; (global 4-component vector of uint)
 0:24        Convert float to uint (temp uint)
 0:24          'inF0' (in float)
@@ -120,12 +118,20 @@ ERROR: node is still EOpNull!
 0:24          0 (const uint)
 0:24          0 (const uint)
 0:24          0 (const uint)
-0:25      Constant:
-0:25        0.000000
-0:26      Constant:
-0:26        0.000000
-0:27      Constant:
-0:27        0.000000
+0:25      normalize (global 2-component vector of float)
+0:25        Construct vec2 (in 2-component vector of float)
+0:25          'inF0' (in float)
+0:26      reflect (global 2-component vector of float)
+0:26        Construct vec2 (in 2-component vector of float)
+0:26          'inF0' (in float)
+0:26        Construct vec2 (in 2-component vector of float)
+0:26          'inF1' (in float)
+0:27      refract (global 2-component vector of float)
+0:27        Construct vec2 (in 2-component vector of float)
+0:27          'inF0' (in float)
+0:27        Construct vec2 (in 2-component vector of float)
+0:27          'inF1' (in float)
+0:27        'inF2' (in float)
 0:28      Constant:
 0:28        0.000000
 0:29      bitFieldReverse (global uint)
@@ -239,8 +245,11 @@ ERROR: node is still EOpNull!
 0:82      bitCount (global 4-component vector of uint)
 0:82        Convert float to uint (temp 4-component vector of uint)
 0:82          'inF0' (in 4-component vector of float)
-0:83      Constant:
-0:83        0.000000
+0:83      cross-product (global 3-component vector of float)
+0:83        Construct vec3 (in 3-component vector of float)
+0:83          'inF0' (in 4-component vector of float)
+0:83        Construct vec3 (in 3-component vector of float)
+0:83          'inF1' (in 4-component vector of float)
 0:84      Constant:
 0:84        0.000000
 0:85      ERROR: Bad unary op
@@ -423,10 +432,14 @@ ERROR: node is still EOpNull!
 0:7      bitCount (global uint)
 0:7        Convert float to uint (temp uint)
 0:7          'inF0' (in float)
-0:8      Constant:
-0:8        0.000000
-0:9      Constant:
-0:9        0.000000
+0:8      cross-product (global 3-component vector of float)
+0:8        Construct vec3 (in 3-component vector of float)
+0:8          'inF0' (in float)
+0:8        Construct vec3 (in 3-component vector of float)
+0:8          'inF1' (in float)
+0:9      Function Call: D3DCOLORtoUBYTE4(vf4; (global 4-component vector of int)
+0:9        Construct vec4 (in 4-component vector of float)
+0:9          'inF0' (in float)
 0:10      Constant:
 0:10        0.000000
 0:12      ERROR: Bad unary op
@@ -439,8 +452,9 @@ ERROR: node is still EOpNull!
 0:14      findLSB (global uint)
 0:14        Convert float to uint (temp uint)
 0:14          'inF0' (in float)
-0:23      Constant:
-0:23        0.000000
+0:23      length (global float)
+0:23        Construct vec2 (in 2-component vector of float)
+0:23          'inF0' (in float)
 0:24      Function Call: msad4(u1;vu2;vu4; (global 4-component vector of uint)
 0:24        Convert float to uint (temp uint)
 0:24          'inF0' (in float)
@@ -452,12 +466,20 @@ ERROR: node is still EOpNull!
 0:24          0 (const uint)
 0:24          0 (const uint)
 0:24          0 (const uint)
-0:25      Constant:
-0:25        0.000000
-0:26      Constant:
-0:26        0.000000
-0:27      Constant:
-0:27        0.000000
+0:25      normalize (global 2-component vector of float)
+0:25        Construct vec2 (in 2-component vector of float)
+0:25          'inF0' (in float)
+0:26      reflect (global 2-component vector of float)
+0:26        Construct vec2 (in 2-component vector of float)
+0:26          'inF0' (in float)
+0:26        Construct vec2 (in 2-component vector of float)
+0:26          'inF1' (in float)
+0:27      refract (global 2-component vector of float)
+0:27        Construct vec2 (in 2-component vector of float)
+0:27          'inF0' (in float)
+0:27        Construct vec2 (in 2-component vector of float)
+0:27          'inF1' (in float)
+0:27        'inF2' (in float)
 0:28      Constant:
 0:28        0.000000
 0:29      bitFieldReverse (global uint)
@@ -571,8 +593,11 @@ ERROR: node is still EOpNull!
 0:82      bitCount (global 4-component vector of uint)
 0:82        Convert float to uint (temp 4-component vector of uint)
 0:82          'inF0' (in 4-component vector of float)
-0:83      Constant:
-0:83        0.000000
+0:83      cross-product (global 3-component vector of float)
+0:83        Construct vec3 (in 3-component vector of float)
+0:83          'inF0' (in 4-component vector of float)
+0:83        Construct vec3 (in 3-component vector of float)
+0:83          'inF1' (in 4-component vector of float)
 0:84      Constant:
 0:84        0.000000
 0:85      ERROR: Bad unary op

Test/hlsl.overload.frag

@@ -1,117 +1,142 @@
 // function selection under type conversion
-void foo1(double a, bool b)  {}
-void foo1(double a, uint b)  {}
-void foo1(double a, int b)   {}
-void foo1(double a, float b) {}
+void foo1(double a, bool b)  {}
+void foo1(double a, uint b)  {}
+void foo1(double a, int b)   {}
+void foo1(double a, float b) {}
 void foo1(double a, double b){}
 
 // uint -> int
-void foo2(int a, bool b)  {}
-void foo2(int a, uint b)  {}
-void foo2(int a, int b)   {}
-void foo2(int a, float b) {}
+void foo2(int a, bool b)  {}
+void foo2(int a, uint b)  {}
+void foo2(int a, int b)   {}
+void foo2(int a, float b) {}
 void foo2(int a, double b){}
 
 // everything can promote
-void foo3(bool b)  {}
-void foo4(uint b)  {}
-void foo5(int b)   {}
-void foo6(float b) {}
+void foo3(bool b)  {}
+void foo4(uint b)  {}
+void foo5(int b)   {}
+void foo6(float b) {}
 void foo7(double b){}
 
 // shorter forward chain better than longer or backward chain
-void foo8(float);
-void foo8(double);
-void foo9(int);
-void foo9(uint);
-void foo10(bool);
-void foo10(int);
+void foo8(float)  {}
+void foo8(double) {}
+void foo9(int)    {}
+void foo9(uint)   {}
+void foo10(bool)  {}
+void foo10(int)   {}
+
+// shape change is worse
+void foo11(float3)  {}
+void foo11(double)  {}
+void foo11(int3)    {}
+void foo11(uint)    {}
+void foo12(float1)  {}
+void foo12(double3) {}
+void foo16(uint)    {}
+void foo16(uint2)   {}
+
+// shape change
+void foo13(float3)  {}
+void foo14(int1)     {}
+void foo15(bool1)   {}
 
 float4 PixelShaderFunction(float4 input) : COLOR0
 {
-    bool b;
-    double d;
-    uint u;
-    int i;
-    float f;
+    bool b;
+    double d;
+    uint u;
+    int i;
+    float f;
+
+    foo1(d, b);
+    foo1(d, d);
+    foo1(d, u);
+    foo1(d, i);
+    foo1(d, f);
+
+    foo1(f, b);
+    foo1(f, d);
+    foo1(f, u);
+    foo1(f, i);
+    foo1(f, f);
+
+    foo1(u, b);
+    foo1(u, d);
+    foo1(u, u);
+    foo1(u, i);
+    foo1(u, f);
+
+    foo1(i, b);
+    foo1(i, d);
+    foo1(i, u);
+    foo1(i, i);
+    foo1(i, f);
+
+    foo2(u, b);
+    foo2(u, d);
+    foo2(u, u);
+    foo2(u, i);
+    foo2(u, f);
+
+    foo2(i, b);
+    foo2(i, d);
+    foo2(i, u);
+    foo2(i, i);
+    foo2(i, f);
+
+    foo3(b);
+    foo3(d);
+    foo3(u);
+    foo3(i);
+    foo3(f);
+
+    foo4(b);
+    foo4(d);
+    foo4(u);
+    foo4(i);
+    foo4(f);
+
+    foo5(b);
+    foo5(d);
+    foo5(u);
+    foo5(i);
+    foo5(f);
+
+    foo6(b);
+    foo6(d);
+    foo6(u);
+    foo6(i);
+    foo6(f);
+
+    foo7(b);
+    foo7(d);
+    foo7(u);
+    foo7(i);
+    foo7(f);
+
+    foo8(b);
+    foo8(u);
+    foo8(i);
+
+    foo9(b);
+    foo9(f);
+    foo9(d);
+
+    foo10(u);
+    foo10(f);
+    foo10(d);
+
+    foo11(b);
+    foo11(f);
+    foo12(float3(f));
+    foo16(int2(i,i));
+
+    foo13(f);
+    foo14(int4(i));
+    foo15(b);
+    foo15(bool3(b));
 
-    foo1(d, b);
-    foo1(d, d);
-    foo1(d, u);
-    foo1(d, i);
-    foo1(d, f);
-
-    foo1(f, b);
-    foo1(f, d);
-    foo1(f, u);
-    foo1(f, i);
-    foo1(f, f);
-
-    foo1(u, b);
-    foo1(u, d);
-    foo1(u, u);
-    foo1(u, i);
-    foo1(u, f);
-
-    foo1(i, b);
-    foo1(i, d);
-    foo1(i, u);
-    foo1(i, i);
-    foo1(i, f);
-
-    foo2(u, b);
-    foo2(u, d);
-    foo2(u, u);
-    foo2(u, i);
-    foo2(u, f);
-
-    foo2(i, b);
-    foo2(i, d);
-    foo2(i, u);
-    foo2(i, i);
-    foo2(i, f);
-
-    foo3(b);
-    foo3(d);
-    foo3(u);
-    foo3(i);
-    foo3(f);
-
-    foo4(b);
-    foo4(d);
-    foo4(u);
-    foo4(i);
-    foo4(f);
-
-    foo5(b);
-    foo5(d);
-    foo5(u);
-    foo5(i);
-    foo5(f);
-
-    foo6(b);
-    foo6(d);
-    foo6(u);
-    foo6(i);
-    foo6(f);
-
-    foo7(b);
-    foo7(d);
-    foo7(u);
-    foo7(i);
-    foo7(f);
-
-    foo8(b);
-    foo8(u);
-    foo8(i);
-
-    foo9(b);
-    foo9(f);
-    foo9(d);
-
-    foo10(u);
-    foo10(f);
-    foo10(d);
-
     return input;
 }