Home | History | Annotate | Download | only in CodeGen
      1 // RUN: %clang_cc1 -emit-llvm %s -o - -triple=i386-pc-win32 | FileCheck %s
      2 // RUN: %clang_cc1 -emit-llvm %s -o - -triple=x86_64-pc-win32 | FileCheck %s --check-prefix=X64
      3 
      4 void __vectorcall v1(int a, int b) {}
      5 // CHECK: define x86_vectorcallcc void @"\01v1@@8"(i32 inreg %a, i32 inreg %b)
      6 // X64: define x86_vectorcallcc void @"\01v1@@16"(i32 %a, i32 %b)
      7 
      8 void __vectorcall v2(char a, char b) {}
      9 // CHECK: define x86_vectorcallcc void @"\01v2@@8"(i8 inreg signext %a, i8 inreg signext %b)
     10 // X64: define x86_vectorcallcc void @"\01v2@@16"(i8 %a, i8 %b)
     11 
     12 struct Small { int x; };
     13 void __vectorcall v3(int a, struct Small b, int c) {}
     14 // CHECK: define x86_vectorcallcc void @"\01v3@@12"(i32 inreg %a, i32 %b.0, i32 inreg %c)
     15 // X64: define x86_vectorcallcc void @"\01v3@@24"(i32 %a, i32 %b.coerce, i32 %c)
     16 
     17 struct Large { int a[5]; };
     18 void __vectorcall v4(int a, struct Large b, int c) {}
     19 // CHECK: define x86_vectorcallcc void @"\01v4@@28"(i32 inreg %a, %struct.Large* byval align 4 %b, i32 inreg %c)
     20 // X64: define x86_vectorcallcc void @"\01v4@@40"(i32 %a, %struct.Large* %b, i32 %c)
     21 
     22 struct HFA2 { double x, y; };
     23 struct HFA4 { double w, x, y, z; };
     24 struct HFA5 { double v, w, x, y, z; };
     25 
     26 void __vectorcall hfa1(int a, struct HFA4 b, int c) {}
     27 // CHECK: define x86_vectorcallcc void @"\01hfa1@@40"(i32 inreg %a, double %b.0, double %b.1, double %b.2, double %b.3, i32 inreg %c)
     28 // X64: define x86_vectorcallcc void @"\01hfa1@@48"(i32 %a, double %b.0, double %b.1, double %b.2, double %b.3, i32 %c)
     29 
     30 // HFAs that would require more than six total SSE registers are passed
     31 // indirectly. Additional vector arguments can consume the rest of the SSE
     32 // registers.
     33 void __vectorcall hfa2(struct HFA4 a, struct HFA4 b, double c) {}
     34 // CHECK: define x86_vectorcallcc void @"\01hfa2@@72"(double %a.0, double %a.1, double %a.2, double %a.3, %struct.HFA4* inreg %b, double %c)
     35 // X64: define x86_vectorcallcc void @"\01hfa2@@72"(double %a.0, double %a.1, double %a.2, double %a.3, %struct.HFA4* %b, double %c)
     36 
     37 // Ensure that we pass builtin types directly while counting them against the
     38 // SSE register usage.
     39 void __vectorcall hfa3(double a, double b, double c, double d, double e, struct HFA2 f) {}
     40 // CHECK: define x86_vectorcallcc void @"\01hfa3@@56"(double %a, double %b, double %c, double %d, double %e, %struct.HFA2* inreg %f)
     41 // X64: define x86_vectorcallcc void @"\01hfa3@@56"(double %a, double %b, double %c, double %d, double %e, %struct.HFA2* %f)
     42 
     43 // Aggregates with more than four elements are not HFAs and are passed byval.
     44 // Because they are not classified as homogeneous, they don't get special
     45 // handling to ensure alignment.
     46 void __vectorcall hfa4(struct HFA5 a) {}
     47 // CHECK: define x86_vectorcallcc void @"\01hfa4@@40"(%struct.HFA5* byval align 4)
     48 // X64: define x86_vectorcallcc void @"\01hfa4@@40"(%struct.HFA5* %a)
     49 
     50 // Return HFAs of 4 or fewer elements in registers.
     51 static struct HFA2 g_hfa2;
     52 struct HFA2 __vectorcall hfa5(void) { return g_hfa2; }
     53 // CHECK: define x86_vectorcallcc %struct.HFA2 @"\01hfa5@@0"()
     54 // X64: define x86_vectorcallcc %struct.HFA2 @"\01hfa5@@0"()
     55 
     56 typedef float __attribute__((vector_size(16))) v4f32;
     57 struct HVA2 { v4f32 x, y; };
     58 struct HVA4 { v4f32 w, x, y, z; };
     59 
     60 void __vectorcall hva1(int a, struct HVA4 b, int c) {}
     61 // CHECK: define x86_vectorcallcc void @"\01hva1@@72"(i32 inreg %a, <4 x float> %b.0, <4 x float> %b.1, <4 x float> %b.2, <4 x float> %b.3, i32 inreg %c)
     62 // X64: define x86_vectorcallcc void @"\01hva1@@80"(i32 %a, <4 x float> %b.0, <4 x float> %b.1, <4 x float> %b.2, <4 x float> %b.3, i32 %c)
     63 
     64 void __vectorcall hva2(struct HVA4 a, struct HVA4 b, v4f32 c) {}
     65 // CHECK: define x86_vectorcallcc void @"\01hva2@@144"(<4 x float> %a.0, <4 x float> %a.1, <4 x float> %a.2, <4 x float> %a.3, %struct.HVA4* inreg %b, <4 x float> %c)
     66 // X64: define x86_vectorcallcc void @"\01hva2@@144"(<4 x float> %a.0, <4 x float> %a.1, <4 x float> %a.2, <4 x float> %a.3, %struct.HVA4* %b, <4 x float> %c)
     67 
     68 void __vectorcall hva3(v4f32 a, v4f32 b, v4f32 c, v4f32 d, v4f32 e, struct HVA2 f) {}
     69 // CHECK: define x86_vectorcallcc void @"\01hva3@@112"(<4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x float> %d, <4 x float> %e, %struct.HVA2* inreg %f)
     70 // X64: define x86_vectorcallcc void @"\01hva3@@112"(<4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x float> %d, <4 x float> %e, %struct.HVA2* %f)
     71 
     72 typedef float __attribute__((ext_vector_type(3))) v3f32;
     73 struct OddSizeHVA { v3f32 x, y; };
     74 
     75 void __vectorcall odd_size_hva(struct OddSizeHVA a) {}
     76 // CHECK: define x86_vectorcallcc void @"\01odd_size_hva@@32"(<3 x float> %a.0, <3 x float> %a.1)
     77 // X64: define x86_vectorcallcc void @"\01odd_size_hva@@32"(<3 x float> %a.0, <3 x float> %a.1)
     78