parallel_codegen.cpp 9.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136
  1. // RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -triple %itanium_abi_triple -fexceptions -fcxx-exceptions -o - | FileCheck %s
  2. // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
  3. // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-DEBUG %s
  4. // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -emit-llvm %s -triple %itanium_abi_triple -fexceptions -fcxx-exceptions -o - | FileCheck --check-prefix SIMD-ONLY0 %s
  5. // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
  6. // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
  7. // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
  8. // expected-no-diagnostics
  9. #ifndef HEADER
  10. #define HEADER
  11. // CHECK-DAG: %struct.ident_t = type { i32, i32, i32, i32, i8* }
  12. // CHECK-DAG: [[STR:@.+]] = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00"
  13. // CHECK-DAG: [[DEF_LOC_2:@.+]] = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* [[STR]], i32 0, i32 0) }
  14. // CHECK-DEBUG-DAG: %struct.ident_t = type { i32, i32, i32, i32, i8* }
  15. // CHECK-DEBUG-DAG: [[STR:@.+]] = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00"
  16. // CHECK-DEBUG-DAG: [[DEF_LOC_2:@.+]] = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* [[STR]], i32 0, i32 0) }
  17. // CHECK-DEBUG-DAG: [[LOC1:@.+]] = private unnamed_addr constant [{{.+}} x i8] c";{{.*}}parallel_codegen.cpp;main;[[@LINE+15]];1;;\00"
  18. // CHECK-DEBUG-DAG: [[LOC2:@.+]] = private unnamed_addr constant [{{.+}} x i8] c";{{.*}}parallel_codegen.cpp;tmain;[[@LINE+7]];1;;\00"
  19. template <class T>
  20. void foo(T argc) {}
  21. template <typename T>
  22. int tmain(T argc) {
  23. #pragma omp parallel
  24. foo(argc);
  25. return 0;
  26. }
  27. int main (int argc, char **argv) {
  28. int a[argc];
  29. #pragma omp parallel
  30. foo(a[1]);
  31. return tmain(argv);
  32. }
  33. // CHECK-LABEL: define {{[a-z\_\b]*[ ]?i32}} @main({{i32[ ]?[a-z]*}} %argc, i8** %argv)
  34. // CHECK: store i32 %argc, i32* [[ARGC_ADDR:%.+]],
  35. // CHECK: [[VLA:%.+]] = alloca i32, i{{[0-9]+}} [[VLA_SIZE:%[^,]+]],
  36. // CHECK: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[DEF_LOC_2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i{{[0-9]+}}, i32*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i{{[0-9]+}} [[VLA_SIZE]], i32* [[VLA]])
  37. // CHECK-NEXT: [[ARGV:%.+]] = load i8**, i8*** {{%[a-z0-9.]+}}
  38. // CHECK-NEXT: [[RET:%.+]] = call {{[a-z\_\b]*[ ]?i32}} [[TMAIN:@.+tmain.+]](i8** [[ARGV]])
  39. // CHECK: ret i32
  40. // CHECK-NEXT: }
  41. // CHECK-DEBUG-LABEL: define i32 @main(i32 %argc, i8** %argv)
  42. // CHECK-DEBUG: [[LOC_2_ADDR:%.+]] = alloca %struct.ident_t
  43. // CHECK-DEBUG: [[KMPC_LOC_VOIDPTR:%.+]] = bitcast %struct.ident_t* [[LOC_2_ADDR]] to i8*
  44. // CHECK-DEBUG-NEXT: [[KMPC_DEFAULT_LOC_VOIDPTR:%.+]] = bitcast %struct.ident_t* [[DEF_LOC_2]] to i8*
  45. // CHECK-DEBUG-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[KMPC_LOC_VOIDPTR]], i8* align 8 [[KMPC_DEFAULT_LOC_VOIDPTR]], i64 24, i1 false)
  46. // CHECK-DEBUG: store i32 %argc, i32* [[ARGC_ADDR:%.+]],
  47. // CHECK-DEBUG: [[VLA:%.+]] = alloca i32, i64 [[VLA_SIZE:%[^,]+]],
  48. // CHECK-DEBUG: [[KMPC_LOC_PSOURCE_REF:%.+]] = getelementptr inbounds %struct.ident_t, %struct.ident_t* [[LOC_2_ADDR]], i32 0, i32 4
  49. // CHECK-DEBUG-NEXT: store i8* getelementptr inbounds ([{{.+}} x i8], [{{.+}} x i8]* [[LOC1]], i32 0, i32 0), i8** [[KMPC_LOC_PSOURCE_REF]]
  50. // CHECK-DEBUG: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[LOC_2_ADDR]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i32*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i64 [[VLA_SIZE]], i32* [[VLA]])
  51. // CHECK-DEBUG-NEXT: [[ARGV:%.+]] = load i8**, i8*** {{%[a-z0-9.]+}}
  52. // CHECK-DEBUG-NEXT: [[RET:%.+]] = call i32 [[TMAIN:@.+tmain.+]](i8** [[ARGV]])
  53. // CHECK-DEBUG: ret i32
  54. // CHECK-DEBUG-NEXT: }
  55. // CHECK: define internal {{.*}}void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i{{[0-9]+}}{{.*}} [[VLA_SIZE:%.+]], i32* {{.+}} [[VLA_ADDR:%[^)]+]])
  56. // CHECK-SAME: #[[FN_ATTRS:[0-9]+]]
  57. // CHECK: store i32* [[VLA_ADDR]], i32** [[VLA_PTR_ADDR:%.+]],
  58. // CHECK: [[VLA_REF:%.+]] = load i32*, i32** [[VLA_PTR_ADDR]]
  59. // CHECK: [[VLA_ELEM_REF:%.+]] = getelementptr inbounds i32, i32* [[VLA_REF]], i{{[0-9]+}} 1
  60. // CHECK-NEXT: [[VLA_ELEM:%.+]] = load i32, i32* [[VLA_ELEM_REF]]
  61. // CHECK-NEXT: invoke {{.*}}void [[FOO:@.+foo.+]](i32{{[ ]?[a-z]*}} [[VLA_ELEM]])
  62. // CHECK: ret void
  63. // CHECK: call {{.*}}void @{{.+terminate.*|abort}}(
  64. // CHECK-NEXT: unreachable
  65. // CHECK-NEXT: }
  66. // CHECK-DEBUG: define internal void [[OMP_OUTLINED_DEBUG:@.+]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i64 [[VLA_SIZE:%.+]], i32* {{.+}} [[VLA_ADDR:%[^)]+]])
  67. // CHECK-DEBUG-SAME: #[[FN_ATTRS:[0-9]+]]
  68. // CHECK-DEBUG: store i32* [[VLA_ADDR]], i32** [[VLA_PTR_ADDR:%.+]],
  69. // CHECK-DEBUG: [[VLA_REF:%.+]] = load i32*, i32** [[VLA_PTR_ADDR]]
  70. // CHECK-DEBUG: [[VLA_ELEM_REF:%.+]] = getelementptr inbounds i32, i32* [[VLA_REF]], i64 1
  71. // CHECK-DEBUG-NEXT: [[VLA_ELEM:%.+]] = load i32, i32* [[VLA_ELEM_REF]]
  72. // CHECK-DEBUG-NEXT: invoke void [[FOO:@.+foo.+]](i32 [[VLA_ELEM]])
  73. // CHECK-DEBUG: ret void
  74. // CHECK-DEBUG: call void @{{.+terminate.*|abort}}(
  75. // CHECK-DEBUG-NEXT: unreachable
  76. // CHECK-DEBUG-NEXT: }
  77. // CHECK-DAG: define linkonce_odr {{.*}}void [[FOO]]({{i32[ ]?[a-z]*}} %argc)
  78. // CHECK-DAG: declare !callback ![[cbid:[0-9]+]] {{.*}}void @__kmpc_fork_call(%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...)
  79. // CHECK-DEBUG-DAG: define linkonce_odr void [[FOO]](i32 %argc)
  80. // CHECK-DEBUG-DAG: declare !callback ![[cbid:[0-9]+]] void @__kmpc_fork_call(%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...)
  81. // CHECK-DEBUG-DAG: define internal void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i64 [[VLA_SIZE:%.+]], i32* {{.+}} [[VLA_ADDR:%[^)]+]])
  82. // CHECK-DEBUG-DAG: call void [[OMP_OUTLINED_DEBUG]]
  83. // CHECK: define linkonce_odr {{[a-z\_\b]*[ ]?i32}} [[TMAIN]](i8** %argc)
  84. // CHECK: store i8** %argc, i8*** [[ARGC_ADDR:%.+]],
  85. // CHECK-NEXT: call {{.*}}void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[DEF_LOC_2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8***)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8*** [[ARGC_ADDR]])
  86. // CHECK-NEXT: ret i32 0
  87. // CHECK-NEXT: }
  88. // CHECK-DEBUG: define linkonce_odr i32 [[TMAIN]](i8** %argc)
  89. // CHECK-DEBUG-DAG: [[LOC_2_ADDR:%.+]] = alloca %struct.ident_t
  90. // CHECK-DEBUG: [[KMPC_LOC_VOIDPTR:%.+]] = bitcast %struct.ident_t* [[LOC_2_ADDR]] to i8*
  91. // CHECK-DEBUG-NEXT: [[KMPC_DEFAULT_LOC_VOIDPTR:%.+]] = bitcast %struct.ident_t* [[DEF_LOC_2]] to i8*
  92. // CHECK-DEBUG-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[KMPC_LOC_VOIDPTR]], i8* align 8 [[KMPC_DEFAULT_LOC_VOIDPTR]], i64 24, i1 false)
  93. // CHECK-DEBUG-NEXT: store i8** %argc, i8*** [[ARGC_ADDR:%.+]],
  94. // CHECK-DEBUG: [[KMPC_LOC_PSOURCE_REF:%.+]] = getelementptr inbounds %struct.ident_t, %struct.ident_t* [[LOC_2_ADDR]], i32 0, i32 4
  95. // CHECK-DEBUG-NEXT: store i8* getelementptr inbounds ([{{.+}} x i8], [{{.+}} x i8]* [[LOC2]], i32 0, i32 0), i8** [[KMPC_LOC_PSOURCE_REF]]
  96. // CHECK-DEBUG-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[LOC_2_ADDR]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8***)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8*** [[ARGC_ADDR]])
  97. // CHECK-DEBUG-NEXT: ret i32 0
  98. // CHECK-DEBUG-NEXT: }
  99. // CHECK: define internal {{.*}}void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i8*** dereferenceable({{4|8}}) %argc)
  100. // CHECK: store i8*** %argc, i8**** [[ARGC_PTR_ADDR:%.+]],
  101. // CHECK: [[ARGC_REF:%.+]] = load i8***, i8**** [[ARGC_PTR_ADDR]]
  102. // CHECK-NEXT: [[ARGC:%.+]] = load i8**, i8*** [[ARGC_REF]]
  103. // CHECK-NEXT: invoke {{.*}}void [[FOO1:@.+foo.+]](i8** [[ARGC]])
  104. // CHECK: ret void
  105. // CHECK: call {{.*}}void @{{.+terminate.*|abort}}(
  106. // CHECK-NEXT: unreachable
  107. // CHECK-NEXT: }
  108. // CHECK-DEBUG: define internal void [[OMP_OUTLINED_DEBUG:@.+]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i8*** dereferenceable({{4|8}}) %argc)
  109. // CHECK-DEBUG: store i8*** %argc, i8**** [[ARGC_PTR_ADDR:%.+]],
  110. // CHECK-DEBUG: [[ARGC_REF:%.+]] = load i8***, i8**** [[ARGC_PTR_ADDR]]
  111. // CHECK-DEBUG-NEXT: [[ARGC:%.+]] = load i8**, i8*** [[ARGC_REF]]
  112. // CHECK-DEBUG-NEXT: invoke void [[FOO1:@.+foo.+]](i8** [[ARGC]])
  113. // CHECK-DEBUG: ret void
  114. // CHECK-DEBUG: call void @{{.+terminate.*|abort}}(
  115. // CHECK-DEBUG-NEXT: unreachable
  116. // CHECK-DEBUG-NEXT: }
  117. // CHECK: define linkonce_odr {{.*}}void [[FOO1]](i8** %argc)
  118. // CHECK-DEBUG-DAG: define linkonce_odr void [[FOO1]](i8** %argc)
  119. // CHECK-DEBUG-DAG: define internal void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i8*** dereferenceable({{4|8}}) %argc)
  120. // CHECK-DEBUG-DAG: call void [[OMP_OUTLINED_DEBUG]]({{[^)]+}}){{[^,]*}}, !dbg
  121. // CHECK: attributes #[[FN_ATTRS]] = {{.+}} nounwind
  122. // CHECK-DEBUG: attributes #[[FN_ATTRS]] = {{.+}} nounwind
  123. // CHECK: ![[cbid]] = !{![[cbidb:[0-9]+]]}
  124. // CHECK: ![[cbidb]] = !{i64 2, i64 -1, i64 -1, i1 true}
  125. #endif