clang/test/CodeGen/variadic-nvptx.c

   1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
   2 // RUN: %clang_cc1 -triple nvptx64-nvidia-cuda -emit-llvm -o - %s | FileCheck %s
   3
   4 extern void varargs_simple(int, ...);
   5
   6 // CHECK-LABEL: define dso_local void @foo(
   7 // CHECK-SAME: ) #[[ATTR0:[0-9]+]] {
   8 // CHECK-NEXT:  [[ENTRY:.*:]]
   9 // CHECK-NEXT:    [[C:%.*]] = alloca i8, align 1
  10 // CHECK-NEXT:    [[S:%.*]] = alloca i16, align 2
  11 // CHECK-NEXT:    [[I:%.*]] = alloca i32, align 4
  12 // CHECK-NEXT:    [[L:%.*]] = alloca i64, align 8
  13 // CHECK-NEXT:    [[F:%.*]] = alloca float, align 4
  14 // CHECK-NEXT:    [[D:%.*]] = alloca double, align 8
  15 // CHECK-NEXT:    [[A:%.*]] = alloca [[STRUCT_ANON:%.*]], align 4
  16 // CHECK-NEXT:    [[V:%.*]] = alloca <4 x i32>, align 16
  17 // CHECK-NEXT:    [[T:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 1
  18 // CHECK-NEXT:    store i8 1, ptr [[C]], align 1
  19 // CHECK-NEXT:    store i16 1, ptr [[S]], align 2
  20 // CHECK-NEXT:    store i32 1, ptr [[I]], align 4
  21 // CHECK-NEXT:    store i64 1, ptr [[L]], align 8
  22 // CHECK-NEXT:    store float 1.000000e+00, ptr [[F]], align 4
  23 // CHECK-NEXT:    store double 1.000000e+00, ptr [[D]], align 8
  24 // CHECK-NEXT:    [[TMP0:%.*]] = load i8, ptr [[C]], align 1
  25 // CHECK-NEXT:    [[CONV:%.*]] = sext i8 [[TMP0]] to i32
  26 // CHECK-NEXT:    [[TMP1:%.*]] = load i16, ptr [[S]], align 2
  27 // CHECK-NEXT:    [[CONV1:%.*]] = sext i16 [[TMP1]] to i32
  28 // CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr [[I]], align 4
  29 // CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr [[L]], align 8
  30 // CHECK-NEXT:    [[TMP4:%.*]] = load float, ptr [[F]], align 4
  31 // CHECK-NEXT:    [[CONV2:%.*]] = fpext float [[TMP4]] to double
  32 // CHECK-NEXT:    [[TMP5:%.*]] = load double, ptr [[D]], align 8
  33 // CHECK-NEXT:    call void (i32, ...) @varargs_simple(i32 noundef 0, i32 noundef [[CONV]], i32 noundef [[CONV1]], i32 noundef [[TMP2]], i64 noundef [[TMP3]], double noundef [[CONV2]], double noundef [[TMP5]]) #[[ATTR3:[0-9]+]]
  34 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[A]], ptr align 4 @__const.foo.a, i64 12, i1 false)
  35 // CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[A]], i32 0, i32 0
  36 // CHECK-NEXT:    [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
  37 // CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[A]], i32 0, i32 1
  38 // CHECK-NEXT:    [[TMP9:%.*]] = load i8, ptr [[TMP8]], align 4
  39 // CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[A]], i32 0, i32 2
  40 // CHECK-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
  41 // CHECK-NEXT:    call void (i32, ...) @varargs_simple(i32 noundef 0, i32 [[TMP7]], i8 [[TMP9]], i32 [[TMP11]]) #[[ATTR3]]
  42 // CHECK-NEXT:    store <4 x i32> splat (i32 1), ptr [[V]], align 16
  43 // CHECK-NEXT:    [[TMP12:%.*]] = load <4 x i32>, ptr [[V]], align 16
  44 // CHECK-NEXT:    call void (i32, ...) @varargs_simple(i32 noundef 0, <4 x i32> noundef [[TMP12]]) #[[ATTR3]]
  45 // CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[T]], i32 0, i32 0
  46 // CHECK-NEXT:    [[TMP14:%.*]] = load i8, ptr [[TMP13]], align 1
  47 // CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[T]], i32 0, i32 1
  48 // CHECK-NEXT:    [[TMP16:%.*]] = load i8, ptr [[TMP15]], align 1
  49 // CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[T]], i32 0, i32 0
  50 // CHECK-NEXT:    [[TMP18:%.*]] = load i8, ptr [[TMP17]], align 1
  51 // CHECK-NEXT:    [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[T]], i32 0, i32 1
  52 // CHECK-NEXT:    [[TMP20:%.*]] = load i8, ptr [[TMP19]], align 1
  53 // CHECK-NEXT:    [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[T]], i32 0, i32 0
  54 // CHECK-NEXT:    [[TMP22:%.*]] = load i8, ptr [[TMP21]], align 1
  55 // CHECK-NEXT:    [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[T]], i32 0, i32 1
  56 // CHECK-NEXT:    [[TMP24:%.*]] = load i8, ptr [[TMP23]], align 1
  57 // CHECK-NEXT:    call void (i32, ...) @varargs_simple(i32 noundef 0, i8 [[TMP14]], i8 [[TMP16]], i8 [[TMP18]], i8 [[TMP20]], i32 noundef 0, i8 [[TMP22]], i8 [[TMP24]]) #[[ATTR3]]
  58 // CHECK-NEXT:    ret void
  59 //
  60 void foo() {
  61   char c = '\x1';
  62   short s = 1;
  63   int i = 1;
  64   long l = 1;
  65   float f = 1.f;
  66   double d = 1.;
  67   varargs_simple(0, c, s, i, l, f, d);
  68
  69   struct {int x; char c; int y;} a = {1, '\x1', 1};
  70   varargs_simple(0, a);
  71
  72   typedef int __attribute__((ext_vector_type(4))) int4;
  73   int4 v = {1, 1, 1, 1};
  74   varargs_simple(0, v);
  75
  76   struct {char c, d;} t;
  77   varargs_simple(0, t, t, 0, t);
  78 }
  79
  80 typedef struct {long x; long y;} S;
  81 extern void varargs_complex(S, S, ...);
  82
  83 // CHECK-LABEL: define dso_local void @bar(
  84 // CHECK-SAME: ) #[[ATTR0]] {
  85 // CHECK-NEXT:  [[ENTRY:.*:]]
  86 // CHECK-NEXT:    [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
  87 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[S]], ptr align 8 @__const.bar.s, i64 16, i1 false)
  88 // CHECK-NEXT:    call void (ptr, ptr, ...) @varargs_complex(ptr noundef byval([[STRUCT_S]]) align 8 [[S]], ptr noundef byval([[STRUCT_S]]) align 8 [[S]], i32 noundef 1, i64 noundef 1, double noundef 1.000000e+00) #[[ATTR3]]
  89 // CHECK-NEXT:    ret void
  90 //
  91 void bar() {
  92   S s = {1l, 1l};
  93   varargs_complex(s, s, 1, 1l, 1.0);
  94 }