diff --git a/clang/docs/OpenMPSupport.rst b/clang/docs/OpenMPSupport.rst index cdb3b331a25e7..78f14ac849edf 100644 --- a/clang/docs/OpenMPSupport.rst +++ b/clang/docs/OpenMPSupport.rst @@ -559,7 +559,7 @@ implementation. +-------------------------------------------------------------+---------------------------+---------------------------+--------------------------------------------------------------------------+ | Clarifications to Fortran map semantics | :none:`unclaimed` | :none:`unclaimed` | | +-------------------------------------------------------------+---------------------------+---------------------------+--------------------------------------------------------------------------+ -| default clause at target construct | :part:`In Progress` | :none:`unclaimed` | | +| default clause at target construct | :good:`done` | :none:`unclaimed` | https://github.com/llvm/llvm-project/pull/162910 | +-------------------------------------------------------------+---------------------------+---------------------------+--------------------------------------------------------------------------+ | ref count update use_device_{ptr, addr} | :none:`unclaimed` | :none:`unclaimed` | | +-------------------------------------------------------------+---------------------------+---------------------------+--------------------------------------------------------------------------+ diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst index 65b086caf3652..15e50d39a706b 100644 --- a/clang/docs/ReleaseNotes.rst +++ b/clang/docs/ReleaseNotes.rst @@ -635,6 +635,7 @@ OpenMP Support - Added support for ``defaultmap`` directive implicit-behavior ``private``. - Added parsing and semantic analysis support for ``groupprivate`` directive. - Added support for 'omp fuse' directive. +- Added support for ``default`` clause on ``target`` directive. Improvements ^^^^^^^^^^^^ diff --git a/clang/lib/Sema/SemaOpenMP.cpp b/clang/lib/Sema/SemaOpenMP.cpp index 0fa21e89b1236..3416ff2632db6 100644 --- a/clang/lib/Sema/SemaOpenMP.cpp +++ b/clang/lib/Sema/SemaOpenMP.cpp @@ -17297,6 +17297,43 @@ OMPClause *SemaOpenMP::ActOnOpenMPSimpleClause( return Res; } +static std::pair +getDefaultmapModifierAndKind(llvm::omp::DefaultKind M, + OpenMPDefaultClauseVariableCategory VCKind) { + OpenMPDefaultmapClauseModifier DefMapMod; + OpenMPDefaultmapClauseKind DefMapKind; + switch (M) { + case OMP_DEFAULT_none: + DefMapMod = OMPC_DEFAULTMAP_MODIFIER_none; + break; + case OMP_DEFAULT_firstprivate: + DefMapMod = OMPC_DEFAULTMAP_MODIFIER_firstprivate; + break; + case OMP_DEFAULT_private: + DefMapMod = OMPC_DEFAULTMAP_MODIFIER_private; + break; + default: + llvm_unreachable("unexpected DSA in OpenMP default clause"); + } + switch (VCKind) { + case OMPC_DEFAULT_VC_aggregate: + DefMapKind = OMPC_DEFAULTMAP_aggregate; + break; + case OMPC_DEFAULT_VC_pointer: + DefMapKind = OMPC_DEFAULTMAP_pointer; + break; + case OMPC_DEFAULT_VC_scalar: + DefMapKind = OMPC_DEFAULTMAP_scalar; + break; + case OMPC_DEFAULT_VC_all: + DefMapKind = OMPC_DEFAULTMAP_all; + break; + default: + llvm_unreachable("unexpected variable category in OpenMP default clause"); + } + return std::make_pair(DefMapMod, DefMapKind); +} + OMPClause *SemaOpenMP::ActOnOpenMPDefaultClause( llvm::omp::DefaultKind M, SourceLocation MLoc, OpenMPDefaultClauseVariableCategory VCKind, SourceLocation VCKindLoc, @@ -17309,21 +17346,40 @@ OMPClause *SemaOpenMP::ActOnOpenMPDefaultClause( return nullptr; } - switch (M) { - case OMP_DEFAULT_none: - DSAStack->setDefaultDSANone(MLoc); - break; - case OMP_DEFAULT_shared: - DSAStack->setDefaultDSAShared(MLoc); - break; - case OMP_DEFAULT_firstprivate: - DSAStack->setDefaultDSAFirstPrivate(MLoc); - break; - case OMP_DEFAULT_private: - DSAStack->setDefaultDSAPrivate(MLoc); - break; - default: - llvm_unreachable("DSA unexpected in OpenMP default clause"); + if (getLangOpts().OpenMP >= 60 && + DSAStack->getCurrentDirective() == OMPD_target) { + // OpenMP 6.0 (see page 224, lines 3-5) default Clause, Semantics + // If data-sharing-attribute is shared then the clause has no effect + // on a target construct; otherwise, its effect on a target construct is + // equivalent to specifying the defaultmap clause with the same + // data-sharing-attribute and variable-category. + if (M != OMP_DEFAULT_shared) { + auto [DefMapMod, DefMapKind] = getDefaultmapModifierAndKind(M, VCKind); + if (DefMapKind == OMPC_DEFAULTMAP_all) { + DSAStack->setDefaultDMAAttr(DefMapMod, OMPC_DEFAULTMAP_aggregate, MLoc); + DSAStack->setDefaultDMAAttr(DefMapMod, OMPC_DEFAULTMAP_scalar, MLoc); + DSAStack->setDefaultDMAAttr(DefMapMod, OMPC_DEFAULTMAP_pointer, MLoc); + } else { + DSAStack->setDefaultDMAAttr(DefMapMod, DefMapKind, MLoc); + } + } + } else { + switch (M) { + case OMP_DEFAULT_none: + DSAStack->setDefaultDSANone(MLoc); + break; + case OMP_DEFAULT_shared: + DSAStack->setDefaultDSAShared(MLoc); + break; + case OMP_DEFAULT_firstprivate: + DSAStack->setDefaultDSAFirstPrivate(MLoc); + break; + case OMP_DEFAULT_private: + DSAStack->setDefaultDSAPrivate(MLoc); + break; + default: + llvm_unreachable("DSA unexpected in OpenMP default clause"); + } } switch (VCKind) { diff --git a/clang/test/OpenMP/target_default_codegen.cpp b/clang/test/OpenMP/target_default_codegen.cpp new file mode 100644 index 0000000000000..a3ee569e343ef --- /dev/null +++ b/clang/test/OpenMP/target_default_codegen.cpp @@ -0,0 +1,1531 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --version 5 +// expected-no-diagnostics + +// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -fopenmp -fopenmp-version=60 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK-64 +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp -fopenmp-version=60 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp -fopenmp-version=60 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK-64 +// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -fopenmp -fopenmp-version=60 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK-32 +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp -fopenmp-version=60 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp -fopenmp-version=60 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CK-32 + +// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -fopenmp-simd -fopenmp-version=60 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY-64 %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp-simd -fopenmp-version=60 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp-simd -fopenmp-version=60 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY-64 %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -fopenmp-simd -fopenmp-version=60 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY-32 %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp-simd -fopenmp-version=60 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s +// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp-simd -fopenmp-version=60 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY-32 %s + +#ifndef HEADER +#define HEADER +void foo1(int a) { + double d = (double)a; + + #pragma omp target default(private: scalar) + { + d += 1.0; + } +} + +void foo2() { + int pvtArr[10]; + + #pragma omp target default(private: aggregate) + { + pvtArr[5]++; + } +} + +void foo3() { + int *pa; + + #pragma omp target default(private: pointer) + { + pa[50]++; + } +} + +// Specified variable-category doesn't apply to referenced variable, so +// normal implicitly determined data-sharing applies. +void foo4() { + int p; + + #pragma omp target default(private: pointer) + { + p++; + } +} + +// Verify default clause with variable-category 'all' is equivalent to no +// variable-category. IR checks generated with 'all' but test runs without +// variable-category. +void foo5(int a) { + double d = (double)a; + int pvtArr[10]; + int *pa; + + #pragma omp target default(private) + { + d += 1.0; + pvtArr[5]++; + pa[50]++; + } +} + +// Verify default clause with 'shared' DSA is ignored. This makes it +// equivalent to target with no default clause. IR checks generated with +// no default clause but test runs with default 'shared'. +void foo6(int a) { + double d = (double)a; + int pvtArr[10]; + int *pa; + + #pragma omp target default(shared) + { + d += 1.0; + pvtArr[5]++; + pa[50]++; + } +} + +// Verify default clause with 'firstprivate' DSA is equivalent to specifying +// defaultmap with 'firstprivate'. IR checks generated with +// defaultmap(firstprivate) but test runs with default(firstprivate). +void foo7(int a) { + double d = (double)a; + int pvtArr[10]; + int *pa; + + #pragma omp target default(firstprivate) + { + d += 1.0; + pvtArr[5]++; + pa[50]++; + } +} +#endif // HEADER +// CK-64-LABEL: define dso_local void @_Z4foo1i( +// CK-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0:[0-9]+]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-64-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-64-NEXT: [[D_CASTED:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// CK-64-NEXT: store double [[TMP1]], ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = load i64, ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP3]], align 8 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP4]], align 8 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP5]], align 8 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP8]], align 4 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 1, ptr [[TMP9]], align 4 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes, ptr [[TMP13]], align 8 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP14]], align 8 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP15]], align 8 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP16]], align 8 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP17]], align 8 +// CK-64-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP18]], align 4 +// CK-64-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP19]], align 4 +// CK-64-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP20]], align 4 +// CK-64-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo1i_l23.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 +// CK-64-NEXT: br i1 [[TMP22]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo1i_l23(i64 [[TMP2]]) #[[ATTR2:[0-9]+]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo1i_l23( +// CK-64-SAME: i64 [[D:%.*]]) #[[ATTR1:[0-9]+]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[D_ADDR:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[D1:%.*]] = alloca double, align 8 +// CK-64-NEXT: store i64 [[D]], ptr [[D_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load double, ptr [[D1]], align 8 +// CK-64-NEXT: [[ADD:%.*]] = fadd double [[TMP0]], 1.000000e+00 +// CK-64-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define dso_local void @_Z4foo2v( +// CK-64-SAME: ) #[[ATTR0]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP0]], align 8 +// CK-64-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP1]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP2]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP5]], align 4 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 1, ptr [[TMP6]], align 4 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes.1, ptr [[TMP9]], align 8 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP13]], align 8 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP14]], align 8 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP15]], align 4 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP17]], align 4 +// CK-64-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l32.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 +// CK-64-NEXT: br i1 [[TMP19]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l32(ptr [[PVTARR]]) #[[ATTR2]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l32( +// CK-64-SAME: ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]]) #[[ATTR1]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[PVTARR1:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 8, !nonnull [[META17:![0-9]+]], !align [[META18:![0-9]+]] +// CK-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR1]], i64 0, i64 5 +// CK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 +// CK-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define dso_local void @_Z4foo3v( +// CK-64-SAME: ) #[[ATTR0]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PA]], align 8 +// CK-64-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store ptr [[TMP0]], ptr [[TMP2]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP3]], align 8 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP6]], align 4 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 1, ptr [[TMP7]], align 4 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes.3, ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP13]], align 8 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP14]], align 8 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP15]], align 8 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP16]], align 4 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4 +// CK-64-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP18]], align 4 +// CK-64-NEXT: [[TMP19:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo3v_l41.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 +// CK-64-NEXT: br i1 [[TMP20]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo3v_l41(ptr [[TMP0]]) #[[ATTR2]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo3v_l41( +// CK-64-SAME: ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[PA1:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PA1]], align 8 +// CK-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 50 +// CK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 +// CK-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define dso_local void @_Z4foo4v( +// CK-64-SAME: ) #[[ATTR0]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[P:%.*]] = alloca i32, align 4 +// CK-64-NEXT: [[P_CASTED:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[P]], align 4 +// CK-64-NEXT: store i32 [[TMP0]], ptr [[P_CASTED]], align 4 +// CK-64-NEXT: [[TMP1:%.*]] = load i64, ptr [[P_CASTED]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP1]], ptr [[TMP2]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP1]], ptr [[TMP3]], align 8 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP4]], align 8 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP7]], align 4 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 1, ptr [[TMP8]], align 4 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes.5, ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP13]], align 8 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP14]], align 8 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP15]], align 8 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP16]], align 8 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP17]], align 4 +// CK-64-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4 +// CK-64-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP19]], align 4 +// CK-64-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo4v_l52.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0 +// CK-64-NEXT: br i1 [[TMP21]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo4v_l52(i64 [[TMP1]]) #[[ATTR2]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo4v_l52( +// CK-64-SAME: i64 [[P:%.*]]) #[[ATTR1]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[P_ADDR:%.*]] = alloca i64, align 8 +// CK-64-NEXT: store i64 [[P]], ptr [[P_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[P_ADDR]], align 4 +// CK-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP0]], 1 +// CK-64-NEXT: store i32 [[INC]], ptr [[P_ADDR]], align 4 +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define dso_local void @_Z4foo5i( +// CK-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-64-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[D_CASTED:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// CK-64-NEXT: store double [[TMP1]], ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = load i64, ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 8 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP4]], align 8 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP5]], align 8 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP6]], align 8 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP7]], align 8 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP8]], align 8 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 +// CK-64-NEXT: store ptr null, ptr [[TMP9]], align 8 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 +// CK-64-NEXT: store ptr null, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP15]], align 4 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 3, ptr [[TMP16]], align 4 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP13]], ptr [[TMP17]], align 8 +// CK-64-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP14]], ptr [[TMP18]], align 8 +// CK-64-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes.7, ptr [[TMP19]], align 8 +// CK-64-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP20]], align 8 +// CK-64-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP21]], align 8 +// CK-64-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP22]], align 8 +// CK-64-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP23]], align 8 +// CK-64-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP24]], align 8 +// CK-64-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP25]], align 4 +// CK-64-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP26]], align 4 +// CK-64-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP27]], align 4 +// CK-64-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo5i_l66.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 +// CK-64-NEXT: br i1 [[TMP29]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo5i_l66(i64 [[TMP2]], ptr [[PVTARR]], ptr [[TMP3]]) #[[ATTR2]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo5i_l66( +// CK-64-SAME: i64 [[D:%.*]], ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]], ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[D_ADDR:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[D1:%.*]] = alloca double, align 8 +// CK-64-NEXT: [[PVTARR2:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: [[PA3:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: store i64 [[D]], ptr [[D_ADDR]], align 8 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 8 +// CK-64-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 8, !nonnull [[META17]], !align [[META18]] +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D1]], align 8 +// CK-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// CK-64-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// CK-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR2]], i64 0, i64 5 +// CK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// CK-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA3]], align 8 +// CK-64-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 50 +// CK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 +// CK-64-NEXT: [[INC5:%.*]] = add nsw i32 [[TMP4]], 1 +// CK-64-NEXT: store i32 [[INC5]], ptr [[ARRAYIDX4]], align 4 +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define dso_local void @_Z4foo6i( +// CK-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-64-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[D_CASTED:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// CK-64-NEXT: store double [[TMP1]], ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = load i64, ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 8 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP4]], align 8 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP5]], align 8 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP6]], align 8 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP7]], align 8 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP8]], align 8 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 +// CK-64-NEXT: store ptr null, ptr [[TMP9]], align 8 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 +// CK-64-NEXT: store ptr null, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP15]], align 4 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 3, ptr [[TMP16]], align 4 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP13]], ptr [[TMP17]], align 8 +// CK-64-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP14]], ptr [[TMP18]], align 8 +// CK-64-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes.9, ptr [[TMP19]], align 8 +// CK-64-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP20]], align 8 +// CK-64-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP21]], align 8 +// CK-64-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP22]], align 8 +// CK-64-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP23]], align 8 +// CK-64-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP24]], align 8 +// CK-64-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP25]], align 4 +// CK-64-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP26]], align 4 +// CK-64-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP27]], align 4 +// CK-64-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo6i_l82.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 +// CK-64-NEXT: br i1 [[TMP29]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo6i_l82(i64 [[TMP2]], ptr [[PVTARR]], ptr [[TMP3]]) #[[ATTR2]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo6i_l82( +// CK-64-SAME: i64 [[D:%.*]], ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]], ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[D_ADDR:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: store i64 [[D]], ptr [[D_ADDR]], align 8 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 8 +// CK-64-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 8, !nonnull [[META17]], !align [[META18]] +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D_ADDR]], align 8 +// CK-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// CK-64-NEXT: store double [[ADD]], ptr [[D_ADDR]], align 8 +// CK-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 5 +// CK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// CK-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA_ADDR]], align 8 +// CK-64-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 50 +// CK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4 +// CK-64-NEXT: [[INC2:%.*]] = add nsw i32 [[TMP4]], 1 +// CK-64-NEXT: store i32 [[INC2]], ptr [[ARRAYIDX1]], align 4 +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define dso_local void @_Z4foo7i( +// CK-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-64-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[D_CASTED:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 +// CK-64-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// CK-64-NEXT: store double [[TMP1]], ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP2:%.*]] = load i64, ptr [[D_CASTED]], align 8 +// CK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 8 +// CK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP4]], align 8 +// CK-64-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: store i64 [[TMP2]], ptr [[TMP5]], align 8 +// CK-64-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 +// CK-64-NEXT: store ptr null, ptr [[TMP6]], align 8 +// CK-64-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP7]], align 8 +// CK-64-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[TMP8]], align 8 +// CK-64-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 +// CK-64-NEXT: store ptr null, ptr [[TMP9]], align 8 +// CK-64-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP10]], align 8 +// CK-64-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP3]], ptr [[TMP11]], align 8 +// CK-64-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 +// CK-64-NEXT: store ptr null, ptr [[TMP12]], align 8 +// CK-64-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-64-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-64-NEXT: store i32 3, ptr [[TMP15]], align 4 +// CK-64-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-64-NEXT: store i32 3, ptr [[TMP16]], align 4 +// CK-64-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-64-NEXT: store ptr [[TMP13]], ptr [[TMP17]], align 8 +// CK-64-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-64-NEXT: store ptr [[TMP14]], ptr [[TMP18]], align 8 +// CK-64-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-64-NEXT: store ptr @.offload_sizes.11, ptr [[TMP19]], align 8 +// CK-64-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-64-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP20]], align 8 +// CK-64-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-64-NEXT: store ptr null, ptr [[TMP21]], align 8 +// CK-64-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-64-NEXT: store ptr null, ptr [[TMP22]], align 8 +// CK-64-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-64-NEXT: store i64 0, ptr [[TMP23]], align 8 +// CK-64-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-64-NEXT: store i64 0, ptr [[TMP24]], align 8 +// CK-64-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-64-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP25]], align 4 +// CK-64-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-64-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP26]], align 4 +// CK-64-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-64-NEXT: store i32 0, ptr [[TMP27]], align 4 +// CK-64-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo7i_l98.region_id, ptr [[KERNEL_ARGS]]) +// CK-64-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 +// CK-64-NEXT: br i1 [[TMP29]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-64: [[OMP_OFFLOAD_FAILED]]: +// CK-64-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo7i_l98(i64 [[TMP2]], ptr [[PVTARR]], ptr [[TMP3]]) #[[ATTR2]] +// CK-64-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-64: [[OMP_OFFLOAD_CONT]]: +// CK-64-NEXT: ret void +// +// +// CK-64-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo7i_l98( +// CK-64-SAME: i64 [[D:%.*]], ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]], ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-64-NEXT: [[ENTRY:.*:]] +// CK-64-NEXT: [[D_ADDR:%.*]] = alloca i64, align 8 +// CK-64-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 8 +// CK-64-NEXT: [[PVTARR1:%.*]] = alloca [10 x i32], align 4 +// CK-64-NEXT: store i64 [[D]], ptr [[D_ADDR]], align 8 +// CK-64-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 8 +// CK-64-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 8 +// CK-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 8, !nonnull [[META17]], !align [[META18]] +// CK-64-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[PVTARR1]], ptr align 4 [[TMP0]], i64 40, i1 false) +// CK-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D_ADDR]], align 8 +// CK-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// CK-64-NEXT: store double [[ADD]], ptr [[D_ADDR]], align 8 +// CK-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR1]], i64 0, i64 5 +// CK-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// CK-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA_ADDR]], align 8 +// CK-64-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 50 +// CK-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4 +// CK-64-NEXT: [[INC3:%.*]] = add nsw i32 [[TMP4]], 1 +// CK-64-NEXT: store i32 [[INC3]], ptr [[ARRAYIDX2]], align 4 +// CK-64-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo1i( +// CK-32-SAME: i32 [[A:%.*]]) #[[ATTR0:[0-9]+]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-32-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-32-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP1]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP3]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP6]], align 4 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 1, ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes, ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes, ptr [[TMP11]], align 4 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP12]], align 4 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP13]], align 4 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP14]], align 8 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP15]], align 8 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP16]], align 4 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP18]], align 4 +// CK-32-NEXT: [[TMP19:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo1i_l23.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 +// CK-32-NEXT: br i1 [[TMP20]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo1i_l23(ptr [[D]]) #[[ATTR2:[0-9]+]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo1i_l23( +// CK-32-SAME: ptr nonnull align 4 dereferenceable(8) [[D:%.*]]) #[[ATTR1:[0-9]+]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[D1:%.*]] = alloca double, align 8 +// CK-32-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4, !nonnull [[META18:![0-9]+]], !align [[META19:![0-9]+]] +// CK-32-NEXT: [[TMP1:%.*]] = load double, ptr [[D1]], align 8 +// CK-32-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// CK-32-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo2v( +// CK-32-SAME: ) #[[ATTR0]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP0]], align 4 +// CK-32-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP1]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP5]], align 4 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 1, ptr [[TMP6]], align 4 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes.1, ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP11]], align 4 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP12]], align 4 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP13]], align 8 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP14]], align 8 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP15]], align 4 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l32.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 +// CK-32-NEXT: br i1 [[TMP19]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l32(ptr [[PVTARR]]) #[[ATTR2]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo2v_l32( +// CK-32-SAME: ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]]) #[[ATTR1]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PVTARR1:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR1]], i32 0, i32 5 +// CK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 +// CK-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo3v( +// CK-32-SAME: ) #[[ATTR0]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PA]], align 4 +// CK-32-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[TMP0]], ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP3]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP6]], align 4 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 1, ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes.3, ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP11]], align 4 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP12]], align 4 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP13]], align 4 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP14]], align 8 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP15]], align 8 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP16]], align 4 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP18]], align 4 +// CK-32-NEXT: [[TMP19:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo3v_l41.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 +// CK-32-NEXT: br i1 [[TMP20]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo3v_l41(ptr [[TMP0]]) #[[ATTR2]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo3v_l41( +// CK-32-SAME: ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PA1:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PA1]], align 4 +// CK-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 50 +// CK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 +// CK-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo4v( +// CK-32-SAME: ) #[[ATTR0]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[P:%.*]] = alloca i32, align 4 +// CK-32-NEXT: [[P_CASTED:%.*]] = alloca i32, align 4 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[P]], align 4 +// CK-32-NEXT: store i32 [[TMP0]], ptr [[P_CASTED]], align 4 +// CK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[P_CASTED]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store i32 [[TMP1]], ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store i32 [[TMP1]], ptr [[TMP3]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP4]], align 4 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 1, ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes.5, ptr [[TMP11]], align 4 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP12]], align 4 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP13]], align 4 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP14]], align 4 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP15]], align 8 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP16]], align 8 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP18]], align 4 +// CK-32-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP19]], align 4 +// CK-32-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo4v_l52.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0 +// CK-32-NEXT: br i1 [[TMP21]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo4v_l52(i32 [[TMP1]]) #[[ATTR2]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo4v_l52( +// CK-32-SAME: i32 [[P:%.*]]) #[[ATTR1]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[P_ADDR:%.*]] = alloca i32, align 4 +// CK-32-NEXT: store i32 [[P]], ptr [[P_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[P_ADDR]], align 4 +// CK-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP0]], 1 +// CK-32-NEXT: store i32 [[INC]], ptr [[P_ADDR]], align 4 +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo5i( +// CK-32-SAME: i32 [[A:%.*]]) #[[ATTR0]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-32-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PA]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP3]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP4]], align 4 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP5]], align 4 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP6]], align 4 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 +// CK-32-NEXT: store ptr null, ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP1]], ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP1]], ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 +// CK-32-NEXT: store ptr null, ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP13]], align 4 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 3, ptr [[TMP14]], align 4 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 4 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP12]], ptr [[TMP16]], align 4 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes.7, ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP18]], align 4 +// CK-32-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP19]], align 4 +// CK-32-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP20]], align 4 +// CK-32-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP21]], align 8 +// CK-32-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP22]], align 8 +// CK-32-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP23]], align 4 +// CK-32-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP24]], align 4 +// CK-32-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP25]], align 4 +// CK-32-NEXT: [[TMP26:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo5i_l66.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 +// CK-32-NEXT: br i1 [[TMP27]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo5i_l66(ptr [[D]], ptr [[PVTARR]], ptr [[TMP1]]) #[[ATTR2]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo5i_l66( +// CK-32-SAME: ptr nonnull align 4 dereferenceable(8) [[D:%.*]], ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]], ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[D1:%.*]] = alloca double, align 8 +// CK-32-NEXT: [[PVTARR2:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: [[PA3:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 4 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 4 +// CK-32-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[TMP2:%.*]] = load double, ptr [[D1]], align 8 +// CK-32-NEXT: [[ADD:%.*]] = fadd double [[TMP2]], 1.000000e+00 +// CK-32-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// CK-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR2]], i32 0, i32 5 +// CK-32-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP3]], 1 +// CK-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = load ptr, ptr [[PA3]], align 4 +// CK-32-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 50 +// CK-32-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 +// CK-32-NEXT: [[INC5:%.*]] = add nsw i32 [[TMP5]], 1 +// CK-32-NEXT: store i32 [[INC5]], ptr [[ARRAYIDX4]], align 4 +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo6i( +// CK-32-SAME: i32 [[A:%.*]]) #[[ATTR0]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-32-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PA]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP3]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP4]], align 4 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP5]], align 4 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP6]], align 4 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 +// CK-32-NEXT: store ptr null, ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP1]], ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP1]], ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 +// CK-32-NEXT: store ptr null, ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP13]], align 4 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 3, ptr [[TMP14]], align 4 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 4 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP12]], ptr [[TMP16]], align 4 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes.9, ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP18]], align 4 +// CK-32-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP19]], align 4 +// CK-32-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP20]], align 4 +// CK-32-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP21]], align 8 +// CK-32-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP22]], align 8 +// CK-32-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP23]], align 4 +// CK-32-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP24]], align 4 +// CK-32-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP25]], align 4 +// CK-32-NEXT: [[TMP26:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo6i_l82.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 +// CK-32-NEXT: br i1 [[TMP27]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo6i_l82(ptr [[D]], ptr [[PVTARR]], ptr [[TMP1]]) #[[ATTR2]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo6i_l82( +// CK-32-SAME: ptr nonnull align 4 dereferenceable(8) [[D:%.*]], ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]], ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[D1:%.*]] = alloca double, align 8 +// CK-32-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 4 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 4 +// CK-32-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[TMP2:%.*]] = load double, ptr [[TMP0]], align 8 +// CK-32-NEXT: store double [[TMP2]], ptr [[D1]], align 8 +// CK-32-NEXT: [[TMP3:%.*]] = load double, ptr [[D1]], align 8 +// CK-32-NEXT: [[ADD:%.*]] = fadd double [[TMP3]], 1.000000e+00 +// CK-32-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// CK-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP1]], i32 0, i32 5 +// CK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1 +// CK-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[TMP5:%.*]] = load ptr, ptr [[PA_ADDR]], align 4 +// CK-32-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 50 +// CK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4 +// CK-32-NEXT: [[INC3:%.*]] = add nsw i32 [[TMP6]], 1 +// CK-32-NEXT: store i32 [[INC3]], ptr [[ARRAYIDX2]], align 4 +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define dso_local void @_Z4foo7i( +// CK-32-SAME: i32 [[A:%.*]]) #[[ATTR0]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// CK-32-NEXT: [[D:%.*]] = alloca double, align 8 +// CK-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 +// CK-32-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 +// CK-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// CK-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// CK-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// CK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PA]], align 4 +// CK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP2]], align 4 +// CK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: store ptr [[D]], ptr [[TMP3]], align 4 +// CK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 +// CK-32-NEXT: store ptr null, ptr [[TMP4]], align 4 +// CK-32-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP5]], align 4 +// CK-32-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[TMP6]], align 4 +// CK-32-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 +// CK-32-NEXT: store ptr null, ptr [[TMP7]], align 4 +// CK-32-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP1]], ptr [[TMP8]], align 4 +// CK-32-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP1]], ptr [[TMP9]], align 4 +// CK-32-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 +// CK-32-NEXT: store ptr null, ptr [[TMP10]], align 4 +// CK-32-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 +// CK-32-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 +// CK-32-NEXT: store i32 3, ptr [[TMP13]], align 4 +// CK-32-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 +// CK-32-NEXT: store i32 3, ptr [[TMP14]], align 4 +// CK-32-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 +// CK-32-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 4 +// CK-32-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 +// CK-32-NEXT: store ptr [[TMP12]], ptr [[TMP16]], align 4 +// CK-32-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 +// CK-32-NEXT: store ptr @.offload_sizes.11, ptr [[TMP17]], align 4 +// CK-32-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 +// CK-32-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP18]], align 4 +// CK-32-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 +// CK-32-NEXT: store ptr null, ptr [[TMP19]], align 4 +// CK-32-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 +// CK-32-NEXT: store ptr null, ptr [[TMP20]], align 4 +// CK-32-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 +// CK-32-NEXT: store i64 0, ptr [[TMP21]], align 8 +// CK-32-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 +// CK-32-NEXT: store i64 0, ptr [[TMP22]], align 8 +// CK-32-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 +// CK-32-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP23]], align 4 +// CK-32-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 +// CK-32-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP24]], align 4 +// CK-32-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 +// CK-32-NEXT: store i32 0, ptr [[TMP25]], align 4 +// CK-32-NEXT: [[TMP26:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo7i_l98.region_id, ptr [[KERNEL_ARGS]]) +// CK-32-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 +// CK-32-NEXT: br i1 [[TMP27]], label %[[OMP_OFFLOAD_FAILED:.*]], label %[[OMP_OFFLOAD_CONT:.*]] +// CK-32: [[OMP_OFFLOAD_FAILED]]: +// CK-32-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo7i_l98(ptr [[D]], ptr [[PVTARR]], ptr [[TMP1]]) #[[ATTR2]] +// CK-32-NEXT: br label %[[OMP_OFFLOAD_CONT]] +// CK-32: [[OMP_OFFLOAD_CONT]]: +// CK-32-NEXT: ret void +// +// +// CK-32-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z4foo7i_l98( +// CK-32-SAME: ptr nonnull align 4 dereferenceable(8) [[D:%.*]], ptr nonnull align 4 dereferenceable(40) [[PVTARR:%.*]], ptr [[PA:%.*]]) #[[ATTR1]] { +// CK-32-NEXT: [[ENTRY:.*:]] +// CK-32-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PVTARR_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[PA_ADDR:%.*]] = alloca ptr, align 4 +// CK-32-NEXT: [[D1:%.*]] = alloca double, align 8 +// CK-32-NEXT: [[PVTARR2:%.*]] = alloca [10 x i32], align 4 +// CK-32-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 4 +// CK-32-NEXT: store ptr [[PVTARR]], ptr [[PVTARR_ADDR]], align 4 +// CK-32-NEXT: store ptr [[PA]], ptr [[PA_ADDR]], align 4 +// CK-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PVTARR_ADDR]], align 4, !nonnull [[META18]], !align [[META19]] +// CK-32-NEXT: [[TMP2:%.*]] = load double, ptr [[TMP0]], align 8 +// CK-32-NEXT: store double [[TMP2]], ptr [[D1]], align 8 +// CK-32-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[PVTARR2]], ptr align 4 [[TMP1]], i32 40, i1 false) +// CK-32-NEXT: [[TMP3:%.*]] = load double, ptr [[D1]], align 8 +// CK-32-NEXT: [[ADD:%.*]] = fadd double [[TMP3]], 1.000000e+00 +// CK-32-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// CK-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR2]], i32 0, i32 5 +// CK-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1 +// CK-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// CK-32-NEXT: [[TMP5:%.*]] = load ptr, ptr [[PA_ADDR]], align 4 +// CK-32-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 50 +// CK-32-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX3]], align 4 +// CK-32-NEXT: [[INC4:%.*]] = add nsw i32 [[TMP6]], 1 +// CK-32-NEXT: store i32 [[INC4]], ptr [[ARRAYIDX3]], align 4 +// CK-32-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo1i( +// SIMD-ONLY-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0:[0-9]+]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-64-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-64-NEXT: [[D1:%.*]] = alloca double, align 8 +// SIMD-ONLY-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D1]], align 8 +// SIMD-ONLY-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-64-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo2v( +// SIMD-ONLY-64-SAME: ) #[[ATTR0]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-64-NEXT: [[PVTARR1:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR1]], i64 0, i64 5 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP0]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo3v( +// SIMD-ONLY-64-SAME: ) #[[ATTR0]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// SIMD-ONLY-64-NEXT: [[PA1:%.*]] = alloca ptr, align 8 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PA1]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 50 +// SIMD-ONLY-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo4v( +// SIMD-ONLY-64-SAME: ) #[[ATTR0]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[P:%.*]] = alloca i32, align 4 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[P]], align 4 +// SIMD-ONLY-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP0]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC]], ptr [[P]], align 4 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo5i( +// SIMD-ONLY-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-64-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// SIMD-ONLY-64-NEXT: [[D1:%.*]] = alloca double, align 8 +// SIMD-ONLY-64-NEXT: [[PVTARR2:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-64-NEXT: [[PA3:%.*]] = alloca ptr, align 8 +// SIMD-ONLY-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D1]], align 8 +// SIMD-ONLY-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-64-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR2]], i64 0, i64 5 +// SIMD-ONLY-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA3]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 50 +// SIMD-ONLY-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 +// SIMD-ONLY-64-NEXT: [[INC5:%.*]] = add nsw i32 [[TMP4]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC5]], ptr [[ARRAYIDX4]], align 4 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo6i( +// SIMD-ONLY-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-64-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// SIMD-ONLY-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-64-NEXT: store double [[ADD]], ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR]], i64 0, i64 5 +// SIMD-ONLY-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 50 +// SIMD-ONLY-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-64-NEXT: [[INC2:%.*]] = add nsw i32 [[TMP4]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC2]], ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-64-LABEL: define dso_local void @_Z4foo7i( +// SIMD-ONLY-64-SAME: i32 signext [[A:%.*]]) #[[ATTR0]] { +// SIMD-ONLY-64-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-64-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-64-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-64-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-64-NEXT: [[PA:%.*]] = alloca ptr, align 8 +// SIMD-ONLY-64-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-64-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-64-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-64-NEXT: store double [[ADD]], ptr [[D]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR]], i64 0, i64 5 +// SIMD-ONLY-64-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-64-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 8 +// SIMD-ONLY-64-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 50 +// SIMD-ONLY-64-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-64-NEXT: [[INC2:%.*]] = add nsw i32 [[TMP4]], 1 +// SIMD-ONLY-64-NEXT: store i32 [[INC2]], ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-64-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo1i( +// SIMD-ONLY-32-SAME: i32 [[A:%.*]]) #[[ATTR0:[0-9]+]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-32-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-32-NEXT: [[D1:%.*]] = alloca double, align 8 +// SIMD-ONLY-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[TMP1:%.*]] = load double, ptr [[D1]], align 8 +// SIMD-ONLY-32-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-32-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// SIMD-ONLY-32-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo2v( +// SIMD-ONLY-32-SAME: ) #[[ATTR0]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-32-NEXT: [[PVTARR1:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR1]], i32 0, i32 5 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP0]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo3v( +// SIMD-ONLY-32-SAME: ) #[[ATTR0]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// SIMD-ONLY-32-NEXT: [[PA1:%.*]] = alloca ptr, align 4 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PA1]], align 4 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 50 +// SIMD-ONLY-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo4v( +// SIMD-ONLY-32-SAME: ) #[[ATTR0]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[P:%.*]] = alloca i32, align 4 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[P]], align 4 +// SIMD-ONLY-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP0]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC]], ptr [[P]], align 4 +// SIMD-ONLY-32-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo5i( +// SIMD-ONLY-32-SAME: i32 [[A:%.*]]) #[[ATTR0]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-32-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// SIMD-ONLY-32-NEXT: [[D1:%.*]] = alloca double, align 8 +// SIMD-ONLY-32-NEXT: [[PVTARR2:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-32-NEXT: [[PA3:%.*]] = alloca ptr, align 4 +// SIMD-ONLY-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[TMP1:%.*]] = load double, ptr [[D1]], align 8 +// SIMD-ONLY-32-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-32-NEXT: store double [[ADD]], ptr [[D1]], align 8 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR2]], i32 0, i32 5 +// SIMD-ONLY-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA3]], align 4 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i32 50 +// SIMD-ONLY-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 +// SIMD-ONLY-32-NEXT: [[INC5:%.*]] = add nsw i32 [[TMP4]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC5]], ptr [[ARRAYIDX4]], align 4 +// SIMD-ONLY-32-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo6i( +// SIMD-ONLY-32-SAME: i32 [[A:%.*]]) #[[ATTR0]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-32-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// SIMD-ONLY-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-32-NEXT: store double [[ADD]], ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR]], i32 0, i32 5 +// SIMD-ONLY-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 4 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i32 50 +// SIMD-ONLY-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-32-NEXT: [[INC2:%.*]] = add nsw i32 [[TMP4]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC2]], ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-32-NEXT: ret void +// +// +// SIMD-ONLY-32-LABEL: define dso_local void @_Z4foo7i( +// SIMD-ONLY-32-SAME: i32 [[A:%.*]]) #[[ATTR0]] { +// SIMD-ONLY-32-NEXT: [[ENTRY:.*:]] +// SIMD-ONLY-32-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 +// SIMD-ONLY-32-NEXT: [[D:%.*]] = alloca double, align 8 +// SIMD-ONLY-32-NEXT: [[PVTARR:%.*]] = alloca [10 x i32], align 4 +// SIMD-ONLY-32-NEXT: [[PA:%.*]] = alloca ptr, align 4 +// SIMD-ONLY-32-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4 +// SIMD-ONLY-32-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP0]] to double +// SIMD-ONLY-32-NEXT: store double [[CONV]], ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[TMP1:%.*]] = load double, ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[ADD:%.*]] = fadd double [[TMP1]], 1.000000e+00 +// SIMD-ONLY-32-NEXT: store double [[ADD]], ptr [[D]], align 8 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[PVTARR]], i32 0, i32 5 +// SIMD-ONLY-32-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[INC:%.*]] = add nsw i32 [[TMP2]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC]], ptr [[ARRAYIDX]], align 4 +// SIMD-ONLY-32-NEXT: [[TMP3:%.*]] = load ptr, ptr [[PA]], align 4 +// SIMD-ONLY-32-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i32 50 +// SIMD-ONLY-32-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-32-NEXT: [[INC2:%.*]] = add nsw i32 [[TMP4]], 1 +// SIMD-ONLY-32-NEXT: store i32 [[INC2]], ptr [[ARRAYIDX1]], align 4 +// SIMD-ONLY-32-NEXT: ret void +// +//. +// CK-64: [[META17]] = !{} +// CK-64: [[META18]] = !{i64 4} +//. +// CK-32: [[META18]] = !{} +// CK-32: [[META19]] = !{i64 4} +//. diff --git a/clang/test/OpenMP/target_default_messages.cpp b/clang/test/OpenMP/target_default_messages.cpp index be677dffa21ca..6a1a1f99360b5 100644 --- a/clang/test/OpenMP/target_default_messages.cpp +++ b/clang/test/OpenMP/target_default_messages.cpp @@ -24,6 +24,8 @@ int main(int argc, char **argv) { for (int i=0; i<200; i++) foo(); #pragma omp target default(x) // expected-error {{expected 'none', 'shared', 'private' or 'firstprivate' in OpenMP clause 'default'}} for (int i=0; i<200; i++) foo(); +#pragma omp target default(none) // expected-note {{explicit data sharing attribute, data mapping attribute, or is_device_ptr clause requested here}} + x++; // expected-error {{variable 'x' must have explicitly specified data sharing attributes, data mapping attributes, or in an is_device_ptr clause}} #endif #ifdef OMP52